data analysis and resultsshodhganga.inflibnet.ac.in/bitstream/10603/68283/11... · list– one...
TRANSCRIPT
79
CHAPTER 5
DATA ANALYSIS AND RESULTS
5.1 Results from Study 1
5.1.1 Sample Demographics of Study 1
In the first study, a 2x2x2 full factorial between-group design is implemented to test
the validity of hypothetical models 1, 2, 3 and 4 with job-seekers who are in their final
semester of coursework and are about to enter the job-market. Experimental designs
are preferred as research technique when the objective of a study is to test the impact
of new scientific or social developments on human perceptions. Experimental studies,
also known as causal research, have therefore found extensive application in clinical
trials of new drugs on humans (Schulz et al. 1996; Al-Metwalli et al. 2008), impact of
new technologies on user experience (Leong et al. 2006) and in understanding how
multiple factors simultaneously affect perceptions of end users (Lutz and Reilly 1974)
to name a few. Since the objective of this study is to understand the impact of
technical modifications in existing job advertisement designs through the internet on
job-seekers‘ inclination towards applying for the same online, the choice of
experimental design for conducting this study is justified.
Experimental studies are conducted on a sample of subjects who should be selected
from a representative portion of the target population (Trochim 2004; Wharrad and
Silcocks 2009). In non-experimental studies, it is essential that the sample be selected
in a random order in order to ensure unbiased inclusion of participants (Altman and
Bland 1999). However, in experimental studies, there is the issue of voluntary consent
from the participants to take part in the experiments (Wharrad and Silcocks 2009). At
the same time, the fact that the researchers have to control for extraneous variables in
80
order to enhance internal validity requires that some eligibility criteria have to be set
before allowing any individual to take part in the study. Therefore, the final set of
respondents can never be a complete representation of the target population- it is that
section of the population which has agreed to take part in the study and at the same
time meet the eligibility criteria to participate in the study. As a consequence, we can
never generalize the experimental study findings. In other words, experimental
designs lack external validity (Trochim 2004). However, the central objective of
experimental designs is to ensure internal validity, i.e., whether from the study one
can conclude conclusively that the change in the dependent or observed variable is
caused mainly by the predictor variables (Altman and Bland 1999). One way of
bypassing this problem of lack of random selection in experiments is to adopt the
strategy of random allocation (Featherstone and Donovan 1998) where the final pool
of study participants are randomly assigned to the different treatment conditions using
random number tables or by using software for random number generators. As long as
the participants are allocated in this manner to the treatment groups, the internal
validity of the experiment remains intact and the researchers may continue with
stochastic measurement techniques for further evaluation of participant responses
(Trochim 2004) even if the participants only represent a fraction of the actual target
population and may not cover the entire heterogeneity of the overall population.
Previous experimental studies have been conducted with students enrolled in a single
university (Dineen, Ash and Noe 2002), or members of a particular community (Lutz
and reilly 1974) which have not threatened the internal validity of the findings. In the
context of E-recruitment itself, students studying nursing at a single large hospital in
Belgium formed the final sample for one recent study (Van Hoye and Lievens
2007)Hence, to conduct any experiment, the focus of the researcher should be to
81
randomly allocate participants to treatment conditions rather than concentrate on
random sampling (Trochim 2004).
From the India Online Landscape Survey, it can be seen that students with mostly
engineering (43 %) and MBA backgrounds (37 %) occupy a significant portion of the
active job-seeker population registered in various E-recruitment websites. Hence
students who are enrolled in Bachelors of Engineering (B.E.) or Bachelors of
Technology (B.Tech.) courses and those enrolled in post graduate courses in business
administration (MBA) have been considered eligible for the final sampling frame for
study 1. It is further taken into account that the Metro cities of India account for over
70 % of active urban job-seekers (India Online Landscape Survey 2012). Hence the
sampling frame has been further narrowed down to students studying at institutes
located in Metro cities only. In India there are 8 metro cities in all which include
include Mumbai, Ahmedabad and Pune from Western India; Delhi National Capital
Region (NCR) comprising of New Delhi, Gurgaon, Noida, Faridabad and Ghaziabad
from Northern India; Chennai, Bangalore and Hyderabad from South India; and
Kolkata, the only metro city in Eastern India, as per India Population Statistics 2011
developed by Census Commission (www.censusindia.gov.in). A city has been
considered as a metro city as per definition given by Census Commission (2011)
which mandates that any city with population greater than 4 million should be treated
as a metro city.
Respondents/ final year students residing and/or studying at any of the Indian Metro
cities have been further identified as active job-seekers from urban India who are
about to enter the job market in less than a year. Invitation to take part in the study
was sent to universities/ institutes imparting post-graduate courses on business
administration and engineering which were located in the major metro cities identified
82
as above and which have featured in the top 50 of the ‗Best B-School Survey‘ of 2012
(Pandey and Bhattacharya 2012) conducted by Business Today and Nielsen
Company and top 50 engineering colleges featured in India Today‘s ‗India‘s Best
Colleges Survey 2012‘ (www.indiatoday.intoday.in) conducted by India Today and
Nielsen Company. These two surveys are considered respectable benchmarks for a
good institute in India (Tyagi 2012). Three institutes featured in the Business Today
list– one private university from Hyderabad, one private B-School from Delhi NCR
and another privately owned B-School from Kolkata gave their consent to have their
students take part in the experiments. No engineering college/ institute agreed to the
study invitation from the list provided by India Today. Hence the final study sample
was further restricted down to only final year Business Administration students
studying at three institutes located in three different Indian Metros. Researchers who
have used experimental designs as research methods have often used students as
surrogate for the actual target population (Greenberg 1987; Peterson 2001). Such
attempts have been criticized by many (Shuptrine 1975; Lynch 1982), citing such
practices may reduce the external validity of the study even further. Calder and
colleagues (1982) have retaliated by pointing out that since the objective is to verify
internal validity, the lack of external validity should not be a concern. Host and his
associates (2000) have empirically proven that there is no notable difference between
responses given by students and those given by mature adults given that the students
have proper idea about the research context, even better if they have themselves
experienced the conditions as have mature people. In this study context, since the
active job-seekers who are about to enter the job market are by definition the principal
users of E- recruitment, hence the question of surrogacy should not arise here. Hence
83
the choice of final year students as the first study‘s target population should help in
enhancing the external validity of this study.
Table 3: Demographic Characteristics for Sample of Study 1
Demographic Variable Category Frequency Percentage
Gender Male
Female
255
202
55.8
44.2
Age < 20 yrs
20-25 yrs
26-30 yrs
28
403
26
6.1
88.2
5.7
Work Experience Nil
< 1 yr
1-2 yrs
2-3 yrs
>3 yrs
183
196
46
22
10
40.04
42.9
10.1
4.8
2.2
Involvement in online
job search
< 1 yr
1-2 yrs
2-3 yrs
3-4 yrs
> 4 yrs
295
99
29
10
24
64.55
21.66
6.34
2.18
5.25
Registration in online
job portals
Nil
Naukri
Monster
Timesjobs
Shine
Others
70
302
188
129
52
40
15.31
66.08
41.11
28.22
11.37
8.75
84
During the data collection process, proper permission has been sought from respective
institute authority to permit the students to take part in the data collection process.
Random Allocation technique has been used to ensure parametric estimation of results
(Forder et al. 2005). The students were contacted from the list of Final year students
enrolled in the respective institutes which was available at the admission office of the
university. It was ensured from the placement departments that none of the students
who were contacted have received any job-offer till the time of interview. The
respondents were then contacted through a notice in the student bulletin board where
it was clearly stated that the purpose of the study and voluntary participation was
invited from the students.
From the institute in Hyderabad, out of total batch strength of 800 students, 256
students agreed to take part in the study (32 %). From the institute in Gurgaon, out of
657 students in final year, 165 students agreed to volunteer for the experiment (25.12
%), while in the institute in Kolkata, out of 135 students in the passing out batch, 45
(33.3 %) gave their consent to take part in the study. In total, 466 students responded
to the invitation and agreed to take part in the study. Students in their respective
institutes were assembled in the institute auditorium on a given date and then using
the online random number generator software – research randomizer (Urbaniak and
Plous 2013), they were assigned to eight different groups. Research randomizer gives
the provision to assign treatment conditions and number of participants for a
particular experiment and computes which participant should be exposed to which
treatment. It has been successfully used in experimental research in recent times and
has become popular for its precise and easy assignment of participants randomly to
different treatments (Leong et al. 2006; Al-Metwalli et al. 2008).
85
With the help of student and faculty volunteers, the experimental survey was
conducted in eight different classrooms in each institute, where members of the eight
groups were randomly assigned by drawing cards where the names of specific
treatments were written. For example, treatment 1 was written on one card, treatment
2 on another and so on. To maintain allocation concealment (Schulz and Grimes
2002; Forder et al. 2005), the cards were put in opaque envelopes and then randomly
distributed to the students to achieve completely randomized experiment. Experiment
volunteers present at each classroom gave an initial brief about the objectives of the
study and facilitated in the survey fill-up process. On completion of the survey, the
respondents submitted their responses to the moderator and left. On final inspection, it
was found that despite repeated announcement by the moderators, nine questionnaires
had incomplete responses. These responses were dropped from the final analysis.
Thus our final sample size came down from 466 to 457 respondents. This sample size
is adequate for running experimental designs with a statistical power of 80 % and
effect size of 0.5 (Cohen 1988; Pierce et al. 2004). As a token of gratitude each
participant received a chocolate bar. Approximately 56 % respondents of study 1 are
male and 88 % are in the age group of 20-25 years. Nearly 83 % of these individuals
have less than one year of experience. Activity in online job search has been dating
back mostly to the past 1-2 years. Most of the respondents have registered in the job
portal Naukri.com (66.08 %), followed by Monster.com (41.11 %).
5.1.2 Data Analysis for Study 1
Data analysis for Study 1 was conducted in five phases. At Phase I, a multivariate
Analysis of Variance (MANOVA) was conducted to test the theoretical model 1,
wherein the effect of preview-mode, testimonial-type and testimonial-source on
quality, credibility, attractiveness and intention to apply taken together as the
86
dependent variable set was examined. This was followed by Phase II, where perceived
credibility and perceived quality were treated as covariates in the model and the
relationships stated in theoretical model 2 were evaluated using the multivariate
analysis of covariance (MANCOVA) technique. At Phase III, theoretical model 3 was
tested, where attractiveness was also included as a covariate along with perceived
credibility and quality in the model and analysis of covariance (ANCOVA) was
conducted to test the postulated hypotheses.
Finally, at Phase IV, a moderated mediation analysis was conducted to understand
whether the treatments have any moderating effect on the relationship between
perceived credibility, perceived quality, organizational attractiveness and intention to
apply and whether organizational attractiveness acted as a mediator for the inter-
relationship between perceived credibility, perceived quality and intention to apply
using structural equation modeling (SEM). In an attempt to identify which set of
treatments is best conducive of generating the most positive perceptions of quality,
credibility, organizational attractiveness and intention to apply, a post-hoc test using
Scheffe‘s procedure (Klockars and Hancock 2000) has been conducted at Phase V.
Similar exercises have been followed for the Study 2 data analysis.
5.1.2.1 Descriptive Statistics and Scale Reliability Tests
Overall, the means gave indications that the respondents in our sample had fairly high
perceptions of quality of the website (M = 3.76, SD = .7492) and credibility (M =
3.65, SD = .8342). Perceived organizational attractiveness (M = 3.5, SD = .8246) and
intention to apply (M = 3.56, SD = .9278) were also moderately high in the
individuals on an average. The correlation matrix (Table 4) gave indications that
perceived credibility is positively correlated with perceived quality (r = .383),
organizational attractiveness (r = .272) and intention to apply (r = .278). Perceived
87
quality is also positively correlated with attractiveness (r = .213) and intention to
apply (r = .203).
Table 4: Descriptive Statistics and Correlation Diagnostics for Study 1
Name of construct M SD 1. 2. 3. 4.
1.Perceived Credibility 3.7561 .74918 .72** .383* .272* .278*
2. Perceived Quality 3.6543 .83426 .383* .85** .213* .203*
3. Organizational Attractiveness 3.5098 .82465 .272* .213* .82** .665*
4. Intention to Apply 3.5602 .92788 .278* .203* .665* .81**
* Correlation is significant at the 0.01 level (2-tailed).
**Cronbach α reliability coefficients reported in the diagonal, N =457
Organizational attractiveness is found to have a high correlation with intention to
apply (r=.665). The reliability coefficients have been reported in the diagonals for
each construct. Cronbach Alpha values are all above 0.7 (Nunally 1978) indicating
good reliability of the scales.
5.1.2.2 Manipulation Checks (Study 1)
Manipulation checks help in determining whether the participants have clearly
perceived the manipulations in the manner intended by the researchers (Perdue and
Summers 1986). Three separate manipulation checks were conducted with the overall
sample to verify the following three premises- to check the perceived difference in
intention to apply for the audio-visual-preview and text preview; to check the
perceived difference in intention to apply for positive and realistic testimonial and to
check the perceived difference in intention to apply for company-dependent and
company neutral testimonial sources
88
Table 5: Results from Manipulation Check 1 [Study 1]
Descriptive Statistics t-test for equality of means
Manipulation
(Preview-mode)
N
(Sample
Size)
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Text 238 3.193 1.14 .07397 Equal variance
Assumed
Equal variance not
assumed
-1.609
-1.633
456
448
.008
.003
Audio-visual 219 3.984 1.02 .07464
Table 6: Results from Manipulation Check 2 [Study 1]
Descriptive Statistics t-test for equality of means
Manipulation
(Testimonial-
type)
N
(Sample
Size)
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Positive 234 3.143 .8693 -.07102 Equal variance
Assumed
Equal variance not
assumed
-.883
-.895
456
448
.038
.031
Realistic 223 3.813 .8607 -.07102
The mean intention to apply for text preview is found to be 3.193 and that for audio-
visual previews is 3.984, and the mean difference is significant at 5 % level of
significance (refer to table 5). This indicates that the manipulation of preview-mode
89
was comprehended properly by the respondents. The mean intention to apply for
positive testimonial is found to be 3.143 and that for audio-visual previews is 3.813,
and the mean difference is significant at 5 % level of significance (refer to table 6).
This indicates that the manipulation of testimonial-type was comprehended properly
by the respondents. The mean intention to apply for company dependent testimonial-
source is found to be 3.161 and that for audio-visual previews is 3.867, and the mean
difference is significant at 5 % level of significance (refer to table 7).
Table 7: Results from manipulation check 3 [Study 1]
Descriptive Statistics t-test for equality of means
Manipulation
(Testimonial-
source)
N
(Sample
Size)
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Company-
dependent
230 3.161 .9587 .0598 Equal variance
Assumed
Equal variance not
assumed
-.958
-1.122
456
455
.016
.002 Company-
neutral
227 3.867 .9211 .0625
This indicates that the manipulation of testimonial-source was comprehended
properly by the respondents. Therefore the results from the manipulation checks
indicated that the respondents have clearly understood the design manipulations
before responding to the survey questionnaire.
90
5.1.3 Phase I Results (Study 1)
The experimental design for phase 1 (study 1) can be represented as under:
Perceived Quality + Perceived Credibility + Organizational Attractiveness + Intention to apply= β0 +
β1 (Preview-mode) + β2 (Testimonial-type) + β3 (Testimonial-source) + β4 (Preview-mode x
Testimonial-type) + β5 (Testimonial-type x Testimonial-source) + β6 (Preview-mode x Testimonial-
source) + β7 (Preview-mode x Testimonial-type x Testimonial-source) + Error
A multiple analysis of variance (MANOVA) was conducted using SPSS 16 with the
dependent variable block including the constructs of perceived credibility, perceived
quality, organizational attractiveness and intention to apply and preview-mode (two
factors-text/ audio-visual), testimonial-type (two factors-positive/ realistic) and
testimonial-source (two factors- company dependent/ company neutral) as fixed
factors. The Levene‘s test for homogeneity of variance was found to be insignificant
(p > .05) in case of each dependent variable which helped in meeting the necessary
assumptions (Hair, Anderson, Tatham and Black 1998) to conduct a MANOVA (refer
Table 8).
Table 8: Levene’s Test for Significance for MANOVA [Study1]
Name of variable F df1 df2 Significance
Perceived Quality 1.714 7 449 .104
Perceived Credibility .767 7 449 .615
Organizational Attractiveness .985 7 449 .441
Intention to apply and recommend 4.832 7 449 .089
The results from the MANOVA analysis indicate that the multivariate test of
significance (Wilk‘s Lambda) is significant for preview-mode (p = .001), testimonial-
91
type (p = .001) and testimonial-source (p = .001), and the two-way and three-way
interaction terms all have significant F values. The multivariate tests indicate presence
of significant main and interaction effect of the treatment variables (preview-mode,
testimonial-type and testimonial-source) on the dependent variables. The observed
power for the above variables (greater than 0.8) also indicates strong possibility of
main and interaction effect (Hair et al. 1998). Table 9 shows the effect of treatment
variables on each dependent variable separately.
TABLE 9: Multivariate Test of Significance for Perceived Quality,
Perceived Credibility, Attractiveness and Intention to Apply [Study 1]
Name of Variable Test of
Significance
Value F
Value
Partial Eta
Squared
Observed
Power
Previewmode Wilk‘s Lambda 0.958 4.867* 0.042 0.957
Testimonial type Wilk‘s Lambda 0.911 10.851* 0.089 1.000
Testimonial source Wilk‘s Lambda 0.955 5.255* 0.045 0.970
Preview-mode X Testimonial-type Wilk‘s Lambda 0.959 4.822** 0.041 0.956
Preview-mode X Testimonial-source Wilk‘s Lambda 0.941 6.990* 0.059 0.995
Testimonial-type X Testimonial-source Wilk‘s Lambda 0.982 2.042*** 0.018 0.850
Preview-mode X Testimonial-type X
Testimonial-source
Wilk‘s Lambda 0.922 9.476* 0.078 1.000
*p<0.001;**p<0.01; ***p<0.05
The univariate results further indicate that preview-mode does not have any
significant main effect on perceived quality [F (4, 446) = .167, p = .683] and
perceived credibility [F (4, 446) = .082, p = .774]. However, preview-mode is found
to have a significant main effect on organizational attractiveness [F (4, 446) = 7.752,
p = .006, effect size = .02] and intention to apply [F (4, 446) = 18.768, p = .001, effect
size = .043]. Testimonial-type has a main effect only on intention to apply [F (4, 446)
92
= 27.662, p = .001]. Testimonial-source is noticed to have significant main effect on
organizational attractiveness [F (4, 446) = 8.224, p = .004, effect size = .02] and
intention to apply [F (4, 446) = 16.281, p = .001, effect size = .058]. The two-way
interaction effect of preview-mode and testimonial-type is significant only for
organizational attractiveness [F (4, 446) =13.922, p=.001, effect size= .065] and
intention to apply [F (4, 446) = 13.424, p = .001, effect size = .079]. Two-way
interaction between preview-mode and testimonial-source is significant only on
organizational attractiveness [F (4, 446) = 14.028, p = .001, effect size = .06]. As for
the two-way interaction between testimonial-type and testimonial-source, significant
effect is observed for organizational attractiveness [F (4, 446) = 5.620, p = .018, effect
size = .032] and intention to apply [F (4, 446) = 3.196, p = .044, effect size = .037].
The three-way interaction between preview-mode, testimonial-type and testimonial-
source is significant at 99 % confidence interval for perceived quality [F (4, 446) =
8.449, p = .004, effect size = .097], perceived credibility [F (4, 446) = 8.401, p =.004,
effect size = .088] and intention to apply [F (4, 446) = 14.487, p =.001, effect size =
.038] respectively, and at 95 % confidence level for organizational attractiveness [F
(4, 446) = 3.704, p = .045, effect size = .121].
The effect sizes for all variables fall under the range of .03 - .12 which indicates
moderate effect of the treatments on the dependent variables (Kotrlik and Williams
2003). The findings suggest that all four dependent variables are perceived
differently in case of the three-way interaction indicating that websites with different
combinations of preview-mode, testimonial-type and testimonial-source have a
possibility of being differently perceived by job-seekers.
93
Table 10: Univariate Test of Significance for Perceived Quality, Perceived
Credibility, Organizational Attractiveness and Intention to Apply
[Study 1]
Main & Interaction Effects F Value Partial
Eta
Squared
Observed
power
Preview-mode Quality
Preview-mode Credibility
Preview-mode Attractiveness
Preview-modeIntention to apply
0.167
0.082
7.752*
18.768*
0.000
0.000
0.021
0.043
0.069
0.059
0.803
0.991
Testimonial-type Quality
Testimonial-type Credibility Testimonial-type Attractiveness
Testimonial-type Intention to apply
0.248
1.037 0.474
27.662*
0.001
0.002 0.001
0.058
0.079
0.174 0.105
0.999
Testimonial-source Quality
Testimonial-source Credibility Testimonial-source Attractiveness
Testimonial-source Intention to apply
0.034
0.856 8.244*
16.281*
0.000
0.002 0.058
0.035
0.054
0.152 0.817
0.981
Preview-mode X Testimonial-typeQuality
Preview-mode X Testimonial-typeCredibility
Preview-mode X Testimonial-typeAttractiveness
Preview-mode X Testimonial-typeIntention to apply
0.041
0.047
13.922*
13.424*
0.000
0.000
0.065
0.079
0.055
0.055
0.961
0.955
Preview-mode X Testimonial-source Quality
Preview-mode X Testimonial-source Credibility
Preview-mode X Testimonial-source Attractiveness
Preview-mode X Testimonial-source Intention to apply
0.596
2.152
14.028*
0.332
0.001
0.005
0.060
0.001
0.120
0.310
0.962
0.089
Testimonial-type X Testimonial-source Quality
Testimonial-type X Testimonial-source Credibility
Testimonial-type X Testimonial-source Attractiveness
Testimonial-type X Testimonial-source Intention to apply
0.529
0.044
5.620**
3.196**
0.001
0.000
0.032
0.037
0.112
0.055
0.757
0.730
Preview-mode X Testimonial-type X Testimonial-source Quality
Preview-mode X Testimonial-type X Testimonial-source Credibility
Preview-mode X Testimonial-type X Testimonial-source Attractiveness
Preview-mode X Testimonial-type X Testimonial-source Intention to
apply
8.449*
8.401*
3.704**
14.487*
0.097
0.088
0.038
0.121
0.827
0.824
0.884
0.967
*p<0.01, **p<0.05
94
5.1.3.1 Profile Plot Analysis
Profile Plot 1. Dependent Variable-Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-type
Figure 11 demonstrates the marginal means of perceived quality, given preview-mode
and testimonial-type as treatment variables.
Figure 11: Profile Plot 1: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type
The profile plot shows that for video-based preview and positive employee
testimonial, perceived quality of website is highest. Perceived quality was lower than
the above case in both the following situations:
a. Text based preview and positive testimonial
b. Video based preview and realistic testimonial.
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text-based and testimonial-type is realistic.
3.772
3.783
3.733
3.767
Testimonial-type
Estimated Marginal Means of Perceived Quality
95
Profile Plot 2. Dependent Variable-Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-source
Figure 12 demonstrates the marginal means of perceived quality, given preview-mode
and testimonial-source as treatment variables.
Figure 12: Profile Plot 2: DV- Perceived Quality; IV- Preview-mode,
Testimonial-source
The profile plot shows that for video-based preview and company-dependent source
of employee testimonial, estimated marginal means for perceived quality of website is
highest. Perceived quality is lower than the above case in both the following
situations:
a. Text based preview and company neutral testimonial-source
b. Video based preview and company neutral testimonial-source.
3.738
3.75
3.77
3.81
Estimated Marginal Means of Perceived Quality
Testimonial-source
96
The lowest marginal mean was recorded for the scenario where the preview-mode is
text-based and testimonial-type is company-dependent.
Profile Plot 3. Dependent Variable-Perceived Quality, Treatment Variables-
Testimonial-type, testimonial-Source
Figure 13 demonstrates the marginal means of perceived quality, given testimonial-
type and testimonial-source as treatment variables. The profile plot shows that for
positive employee testimonial and company-dependent source of employee
testimonial, estimated marginal means for perceived quality of website is highest.
Figure 13: Profile Plot 3: DV- Perceived Quality; IV- Testimonial-type,
Testimonial-source
Perceived quality is lower than the above case in both the following situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company neutral testimonial-source.
3.75
3.762
3.737
3.80
Testimonial-source
Testimonial-type
Estimated Marginal Means of Perceived Quality
97
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-dependent.
Profile Plot 4. Dependent Variable-Perceived Credibility, Treatment Variables-
Testimonial-type, Preview-mode
Figure 14 demonstrates the marginal means of perceived credibility, given
testimonial-type and preview-mode as treatment variables. The profile plot shows that
for video-based preview and positive employee testimonial, perceived credibility of
website is highest.
Figure 14: Profile Plot 4: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type
Perceived credibility for text based preview and positive testimonial is lower than the
above case but greater than in both the following situations:
a. Text based preview and realistic testimonial.
b. Video based preview and realistic testimonial
Estimated Marginal Means of Perceived Credibility
Testimonial-type
3.718
3.682
3.634 3.625
98
The difference in marginal mean for the scenario where the preview-mode is audio-
visual and testimonial-type is realistic and the scenario for which preview-mode is
text-based and testimonial-type is realistic is not very large, indicating that both these
combinations have more or less similar influence on perceived credibility.
Profile Plot 5. Dependent Variable-Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-source
Figure 15 demonstrates the marginal means of perceived credibility, given
testimonial-source and preview-mode as treatment variables.
Figure 15: Profile Plot 5: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-source
The profile plot shows that for text-based preview and company-dependent source of
employee testimonial, estimated marginal means for perceived credibility of website
3.653
3.73
3.69
3.575
Estimated Marginal Means of Perceived Credibility
Testimonial-source
99
is highest. Perceived credibility is lower than the above case in both the following
situations:
a. Video based preview and company dependent testimonial-source
b. Video based preview and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text-based and testimonial-source is company-neutral.
Profile Plot 6. Dependent Variable-Perceived Credibility, Treatment Variables-
Testimonial-type, Testimonial-source
Figure 16 demonstrates the marginal means of perceived credibility, given
testimonial-type and testimonial-source as treatment variables.
Figure 16: Profile Plot 6: DV- Perceived Credibility; IV-Testimonial-type,
Testimonial-source
The profile plot shows that for positive employee testimonial and company-dependent
source of employee testimonial, estimated marginal means for perceived credibility of
3.67
3.72
3.59
3.68
Testimonial-source
Testimonial-type
Estimated Marginal Means of Perceived Credibility
100
website is highest. Perceived credibility is lower than the above case in both the
following situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-neutral.
Profile Plot 7. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Preview-mode
Figure 17 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-type as treatment variables.
Figure 17: Profile Plot 7: DV- Organizational Attractiveness; IV-Testimonial-
type, Preview-mode
3.48
3.69
3.223
3.55
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
101
The profile plot shows that for video-based preview and positive employee
testimonial, organizational attractiveness is highest. Organizational attractiveness is
lower than the above case in both the following situations:
a. Text based preview and realistic testimonial
b. Video based preview and realistic testimonial.
However, the lowest marginal mean is recorded for the scenario where the preview-
mode is text-based and testimonial-type is positive.
Profile Plot 8. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-source
Figure 18 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-source as treatment variables.
Figure 18: Profile Plot 8: DV- Organizational Attractiveness; IV-Preview-mode,
Testimonial-source
The profile plot shows that the combinations video-based preview and company-
dependent source of employee testimonial, and text-based preview and company
neutral source of testimonial, have same level of estimated marginal means for
3.61
3.15
3.61
3.58
Testimonial-source Estimated Marginal Means of Organizational Attractiveness
102
organizational attractiveness, and it is also highest for the above two combinations.
Attractiveness is lower than the above cases in the situation where preview is video
based and testimonial-source is company neutral. However, the lowest marginal mean
was recorded for the scenario where the preview-mode is text-based but testimonial-
source is company-dependent.
Profile Plot 9. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Testimonial-source
Figure 19 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables.
Figure 19: Profile Plot 9: DV- Organizational Attractiveness; IV-Testimonial-
type, Testimonial-source
The profile plot shows that for positive employee testimonial and company-neutral
source of employee testimonial, estimated marginal means for organizational
attractiveness of website is highest. Attractiveness is lower than the above case in
both the following situations:
3.48
3.52
3.65
3.28
Testimonial-type
Testimonial-source
Estimated Marginal Means of Organizational Attractiveness
103
a. Realistic employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-dependent.
Profile Plot 10. Dependent Variable-Intention to Apply, Treatment Variables-
Testimonial-type, Preview-mode
Figure 20 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables. The profile plot shows that for realistic
employee testimonial and audio-visual preview, estimated marginal means for
intention to apply is highest.
Figure 20: Profile Plot 10: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
Intention to apply is lower than the above case in both the following situations:
a. Realistic employee testimonial and text based preview
b. Positive employee testimonial and audio-visual preview.
3.81
3.86 3.83
3.23
Testimonial-type
Estimated Marginal Means of Intention to apply
104
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive and corporate preview is text based.
Profile Plot 11. Dependent Variable- Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 21 demonstrates the marginal means of intention to apply, given testimonial-
source and preview-mode as treatment variables. The profile plot shows that for
audio-visual previews and company-neutral source of employee testimonial,
estimated marginal means for intention to apply is highest.
Figure 21: Profile Plot 11: DV- Intention to Apply; IV-Preview-mode,
Testimonial-source
Intention to apply is lower than the above case and roughly same in both the
following situations:
a. Text preview and company neutral testimonial-source
b. Video preview and company dependent testimonial-source.
3.77
4.00
3.76
3.40
Testimonial-source
Estimated Marginal Means of Intention to apply
105
However, the lowest marginal mean was recorded for the scenario where the preview
is text based but testimonial-source is company-dependent.
Profile Plot 12. Dependent Variable- Intention to Apply, Treatment Variables-
Testimonial-Type, Testimonial-Source
Figure 22 demonstrates the marginal means of intention to apply, given testimonial-
type and testimonial-source as treatment variables. The profile plot shows that for
realistic employee testimonial and company-neutral source of employee testimonial,
estimated marginal means for intention to apply is highest.
Figure 22: Profile Plot 12: DV- Intention to Apply; IV-Testimonial-type,
Testimonial-source
Intention to apply is lower than the above case in both the following situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company dependent testimonial-source.
3.845
3.35
4.05
3.772
Testimonial-type
Testimonial-source
Estimated Marginal Means of Intention to apply
106
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but testimonial-source is company-dependent.
Profile Plot 13. Dependent Variable- Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 23 demonstrates the marginal means of perceived quality, given testimonial-
type, preview-mode as treatment variables when testimonial-source is company
dependent.
Figure 23: Profile Plot 13: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
The profile plot shows that for positive employee testimonial and text preview, and
for realistic testimonial and video preview, estimated marginal means for perceived
quality of website is approximately same. Perceived quality is lower than the above
3.75
3.853 3.852
3.62
Testimonial-type
Estimated Marginal Means of Perceived Quality
At Testimonial-source =
107
case in the situation where the testimonial is positively framed and the corporate
preview is video based. However, the lowest marginal mean was recorded for the
scenario where the testimonial is realistic but firm‘s preview is given in plain text,
especially since the testimonial is posted through a company dependent channel.
Profile Plot 14. Dependent variable- Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 24: Profile Plot 14: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-neutral)
Figure 24 demonstrates the marginal means of perceived quality, given testimonial-
type, preview-mode as treatment variables when testimonial-source is company
neutral. The profile plot shows that for realistic employee testimonial and text
preview, estimated marginal means for perceived quality of website is highest.
Perceived quality is lower than the above case in the situation where the testimonial is
positively framed and the corporate preview is video based. However, the lowest
marginal means are recorded for the scenarios where the testimonial is positive but
3.803
3.85
3.67 3.66
Testimonial-type
Estimated Marginal Means of Perceived Quality
At Testimonial-source=
108
firm‘s preview is given in plain text, and for realistic testimonial and video preview,
the estimated marginal means for these two cases being more or less similar.
Profile Plot 15. Dependent Variable- Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 25: Profile Plot 15: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
Figure 25 demonstrates the marginal means of perceived credibility, given
testimonial-type, preview-mode as treatment variables when testimonial-source is
company dependent. The profile plot shows that for positive employee testimonial
and text preview, estimated marginal means for perceived credibility of website is
highest. Perceived credibility is lower than the above case in case of the situation
where there is realistic employee testimonial and video preview. For realistic
3.60
3.73
3.62
3.85 Testimonial-type
Estimated Marginal Means of Perceived credibility
At Testimonial-source
109
employee testimonial and text based preview, estimated marginal means is less than
the above two conditions, and it is less than the combination where the testimonial is
realistic and corporate preview is video-based. However, the lowest marginal mean
was recorded for the scenario where the testimonial is positive but preview is video
based.
Profile Plot 16. Dependent Variable- Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 26: Profile Plot 16: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-neutral)
Figure 26 demonstrates the marginal means of perceived credibility, given
testimonial-type, preview-mode as treatment variables when testimonial-source is
company neutral. The profile plot shows that for positive employee testimonial and
audio-visual preview, estimated marginal means for perceived quality of website is
highest. Perceived credibility is lower than the above case in the situation where the
testimonial is realistically framed but the corporate preview is text based when the
testimonial is displayed in a company neutral platform. However, the lowest marginal
3.85
3.55
3.65
3.53
Testimonial-type
Estimated Marginal Means of Perceived Credibility
At Testimonial-source = Company Neutral
110
means are recorded for the scenarios where the testimonial is positive but firm‘s
preview is given in plain text, and for realistic testimonial and video preview, the
estimated marginal means for these two cases being more or less similar.
Profile Plot 17. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-
dependent)
Figure 27 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables when testimonial-source is
company dependent.
Figure 27: Profile Plot 17: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-dependent)
The profile plot shows that for positive employee testimonial and audio-visual
corporate preview, estimated marginal means for organizational attractiveness is
highest. Organizational attractiveness is lower than the above case in both the
following situations:
3.50 3.45
3.70
2.85
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
At Testimonial-source
111
a. Realistic employee testimonial and text-based preview
b. Realistic employee testimonial and audiovisual preview
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive and corporate preview is plain text based.
Profile Plot 18. Dependent variable- Organizational Attractiveness, Treatment
variables- Preview-Mode, Testimonial-Type, Testimonial-Source (Company-
neutral)
Figure 28: Profile Plot 18: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-neutral)
Figure 28 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables when testimonial-source is
company neutral. The profile plot shows that for positive employee testimonial and
audio-visual corporate preview, estimated marginal means for organizational
attractiveness is highest. Organizational attractiveness is lower than the above case
and is marginally equal in both the following situations:
3.67
3.61
3.62
3.44
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
At Testimonial-source = Company Neutral
112
a. Positive employee testimonial and text based corporate preview
b. Realistic employee testimonial and text based corporate preview.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but corporate preview is video based.
Profile Plot 19. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 29: Profile Plot 19: Dependent Variable-Intention to Apply, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-
dependent)
Figure 29 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial-source is company
dependent. The profile plot shows that for realistic employee testimonial and textual
corporate preview, estimated marginal means for intention to apply is highest.
Intention to apply is lower than the above case and marginally equal in both the
following situations:
3.73
3.74 3.86
2.85
Testimonial-type
Estimated Marginal Means of intention to apply
At Testimonial-source
113
a. Realistic employee testimonial and audio-visual preview
b. Positive employee testimonial and audio-visual preview
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but preview is text based.
Profile Plot 20. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 30: Profile Plot 20: Dependent Variable-Intention to Apply, Treatment
Variables- Preview-mode and Testimonial-type at Testimonial-source
(Company-neutral)
Figure 30 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial-source is company
neutral. The profile plot shows that for realistic employee testimonial and video based
corporate preview, estimated marginal means for intention to apply is highest.
Intention to apply is lower than the above case and equal in both the following
situations:
a. Realistic employee testimonial and text based preview
b. Positive employee testimonial and audio-visual preview
4.15
3.86 3.86
3.63
Testimonial-type
Estimated Marginal Means of Intention to apply
At Testimonial-source
114
5.1.4 Phase II Results (Study 1)
The experimental design for phase II (study 1) can be represented as under:
Organizational Attractiveness + Intention to apply= β0 + β1 (Preview-mode) + β2 (Testimonial-type) +
β3 (Testimonial-source) + β4 (Preview-mode x Testimonial-type) + β5 (Testimonial-type x
Testimonial-source) + β6 (Preview-mode x Testimonial-source) + β7 (Preview-mode x Testimonial-
type x Testimonial-source) + β8 (Perceived Quality) + β9 (Perceived Quality) + Error
A multivariate analysis of covariance (MANCOVA) was conducted using SPSS 16
with the dependent variable block including the constructs of organizational
attractiveness and intention to apply; perceived credibility and perceived quality as
covariates and preview-mode (two factors-text/ audio-visual), testimonial-type (two
factors-positive/ realistic) and testimonial-source (two factors- company dependent/
company neutral) as fixed factors. The Levene‘s test for homogeneity of variance was
found to be insignificant (p > .05) in case of each dependent variable which helped in
meeting the necessary assumptions to conduct a MANCOVA (refer Table 11).
Table 11: Levene's Test of Equality of Error Variances MANCOVA
(Study 1)
Name of Variable F df1 df2 Significance
Organizational Attractiveness 1.715 7 449 .103
Intention to Apply 4.221 7 449 .074
The results from the MANCOVA analysis indicate that the multivariate test of
significance (Wilk‘s Lambda) is significant at 95 % confidence level and preview-
mode, testimonial-type and testimonial-source, as well as the two-way and three-way
interaction terms all have significant F values.
115
TABLE 12: Multivariate Test of Significance for Organizational
Attractiveness and Intention to Apply [Study 1]
Name of Variable Test of
Significance
Value F Value Sig. Partial
Eta
Squared
Observed
Power
Perceived Quality Wilk‘s
Lambda .948 12.165 .000 .052 .995
Perceived Credibility Wilk‘s
Lambda .945 13.091 .000 .055 .997
Preview-mode Wilk‘s
Lambda .959 9.637 .000 .041 .981
Testimonial-type Wilk‘s
Lambda .913 21.130 .000 .087 1.000
Testimonial-source Wilk‘s
Lambda .957 10.049 .000 .043 .985
Preview-mode X Testimonial-type Wilk‘s
Lambda .959 9.570 .000 .041 .981
Preview-mode X Testimonial-source Wilk‘s
Lambda .950 11.789 .000 .050 .994
Testimonial-type X Testimonial-source Wilk‘s
Lambda .984 3.680 .026 .026 .876
Preview-mode X Testimonial-type X
Testimonial-source
Wilk‘s
Lambda .947 12.527 .000 .053 .996
The multivariate tests indicate presence of significant main and interaction effect of
the treatment variables (preview-mode, testimonial-type and testimonial-source) on
the dependent variables. The observed power for the above variables (greater than
0.8) also indicates strong possibility of main and interaction effect (Hair et al. 1998).
Table 13 shows the effect of treatment variables on each dependent variable
separately. The univariate results indicate that perceived website quality has a
significant main effect on organizational attractiveness [F (2, 446) = 22.776, p = .001,
effect size = .098] and intention to apply [F (2, 446) = 15.426, p = .001, effect size =
.073] for the jobseekers.
116
TABLE 13: Univariate Test of Significance for Organizational Attractiveness
and Intention to Apply [Study 1]
Main & Interaction Effects F Value Partial
Eta
Squared
Observed
power
Perceived Quality Attractiveness Perceived Quality Intention to apply
22.776* 15.426*
.098
.073 .997 .975
Perceived Credibility Attractiveness
Perceived Credibility Intention to apply
26.179*
11.104*
.095
.044
.999
.914
Preview-mode Attractiveness
Preview-mode Intention to apply
8.049**
19.283*
.058
.081
.808
.992
Testimonial-type Attractiveness
Testimonial-type Intention to apply
1.276
33.137*
.003
.069
.204
1.000
Testimonial-source Attractiveness
Testimonial-sourceIntention to apply
11.483*
19.379*
.045
.112
.922
.993
Preview-mode X Testimonial-type Attractiveness
Preview-mode X Testimonial-type Intention to apply
16.225*
14.653*
.085
.101
.980
.969
Preview-mode X Testimonial-source Attractiveness
Preview-mode X Testimonial-source Intention to apply
17.862*
.480
.128
.001
.988
.106
Testimonial-type X Testimonial-source Attractiveness
Testimonial-type X Testimonial-source Intention to apply
7.202*
3.898**
.056
.033
.864
.804
Preview-mode X Testimonial-type X Testimonial-source Attractiveness
Preview-mode X Testimonial-type X Testimonial-source Intention to
apply
12.224*
24.794*
.057
.143
.937
.999
*p<0.01, **p<0.05
Perceived credibility also has a significant main effect on organizational attractiveness
[F (2, 446) = 26.179, p =.001, effect size = .095] and intention to apply [F (2, 446) =
11.104, p =.001, effect size = .044]. It is also observed that preview-mode exerts a
notable main effect on organizational attractiveness [F (2, 446) = 8.049, p = .005,
effect size = .058] and intention to apply [F (2, 446) = 19.283, p = .001, effect size =
.081). Testimonial-type has a main effect only on intention to apply [F (2, 446) =
33.137, p = .001, effect size = .069), but not on organizational attractiveness (p =
117
.259). Testimonial-source is noticed to have significant main effect on organizational
attractiveness [F (2, 446) = 11.483, p = .001, effect size = .045] and intention to apply
[F (2, 446) = 19.379, p = .001, effect size = .112].
The two-way interaction effect of preview-mode and testimonial-type is significant
for organizational attractiveness [F (2, 446) = 16.225, p = .001, effect size = .085] and
intention to apply [F (2, 446) = 14.653, p = .001, effect size= .101]. Two-way
interaction between preview-mode and testimonial-source is significant only on
organizational attractiveness [F (2, 446) = 17.862, p = .001, effect size = .128]. As for
the two-way interaction between testimonial-type and testimonial-source, significant
effect is observed for organizational attractiveness [F (2, 446) = 7.202, p = .008, effect
size = .056] and intention to apply [F (2, 446) = 3.898, p = .049, effect size = .033],
albeit at higher confidence level.
The three-way interaction between preview-mode, testimonial-type and testimonial-
source is also observed to be significant for organizational attractiveness [F (2, 446) =
12.224, p = .001, effect size= .057] and intention to apply [F (2, 446) = 24.794, p =
.001, effect size = .143] respectively. The effect sizes (partial eta squared) for all
significant effects range between .03 - .15 which indicates small to moderate effect of
the treatments on the dependent variables (Kotrlik and Williams 2003). The findings
suggest that perceptions regarding attractiveness of an organization as a potential
employer as well as job aspirant‘s intention to apply for any vacancy in the firm are
perceived differently depending on the main effects of perceived quality, perceived
credibility, preview-mode and testimonial-source. Testimonial-type accounts for only
the difference in perception towards intention to apply. Two-way interactions are
significant for combination of treatments preview-mode and testimonial-type, and for
combinations of testimonial-type and testimonial-source. The three-way interaction is
118
also significant on job-seekers‘ perception regarding organizational attraction and
intention to apply. However, a combination of preview-mode and testimonial-source
is not found to be significant for influencing intention to apply. This combination is
found to trigger only difference in perception towards attractiveness of a firm. This
indicates that websites with different combinations of preview-mode, testimonial-type
and testimonial-source have a possibility of being differently perceived by job-seekers
while seeking job-related information through internet compared to websites with
either a single treatment or a combination of any two treatments.
5.1.4.1 Profile Plot Analysis
Profile Plot 21. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Preview-mode
Figure 31 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-type as treatment variables.
Figure 31: Profile Plot 21: DV- Organizational Attractiveness; IV-Testimonial-
type, Preview-mode
3.23
3.47
3.55
3.67
Estimated Marginal Means for Organizational Attractiveness
Testimonial-type
119
The profile plot shows that for video-based preview and positive employee
testimonial, perceived quality of website is highest. Organizational attractiveness is
lower than the above case in both the following situations:
a. Text based preview and realistic testimonial
b. Video based preview and realistic testimonial.
However, the lowest marginal mean is recorded for the scenario where the preview-
mode is in plain text form and testimonial is positively positioned.
Profile Plot 22. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-source
Figure 32 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-source as treatment variables.
Figure 32: Profile Plot 22: DV- Organizational Attractiveness; IV-Preview-mode,
Testimonial-source
3.15
3.53
3.62 3.60
Estimated Marginal Means for Organizational Attractiveness
Testimonial-source
120
The profile plot shows that for text-based preview and company-independent source
of employee testimonial, estimated marginal means for perceived quality of website is
highest. Attractiveness is lower than the above case in both the following situations:
a. Video based preview and company neutral testimonial-source
b. Video based preview and company dependent testimonial-source.
Of all the treatment conditions, the lowest marginal mean was recorded for the
scenario where the preview-mode is text-based but testimonial-source is company-
dependent.
Profile Plot 23. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Testimonial-source
Figure 33 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and testimonial-source as treatment variables. The profile plot shows that
for positive employee testimonial and company-independent source of employee
testimonial, estimated marginal means for perceived attractiveness of the firm is highest.
Figure 33: Profile Plot 23: DV- Organizational Attractiveness; IV-Testimonial-
type, Testimonial-source
3.25
3.50
3.53 3.63
Estimated Marginal Means for Organizational Attractiveness
Testimonial-type
Testimonial-source
121
Organizational attractiveness is lower than the above case in both the following
situations:
a. Realistic employee testimonial and company neutral testimonial-source
b. Positive employee testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the testimonial
is positive but testimonial-source is company-dependent.
Profile Plot 24. Dependent Variable-Intention to Apply, Treatment Variables-
Testimonial-type, Preview-mode
Figure 34 demonstrates the marginal means of Intention to apply, given testimonial-type
and preview-mode as treatment variables.
Figure 34: Profile Plot 24: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
The profile plot shows that for positive employee testimonial and company-dependent
source of employee testimonial, estimated marginal means for intention to apply is
highest. Intention to apply is lower than the above case in both the following situations:
3.80
3.22
3.90
3.93
Estimated Marginal Means for Intention to apply
Testimonial-type
122
a. Realistic employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-dependent.
Profile Plot 25. Dependent Variable- Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 35 demonstrates the marginal means of intention to apply, given preview-mode
and testimonial-source as treatment variables.
Figure 35: Profile Plot 25: DV- Intention to Apply; IV-Preview-mode,
Testimonial-source
The profile plot shows that for video-based preview and company-dependent source
of employee testimonial, estimated marginal means for intention to apply is highest.
Intention to apply for job is lower than the above case in both the following situations:
a. Text based preview and company neutral testimonial-source
b. Video based preview and company neutral testimonial-source.
3.72
3.40
3.74
4.06
Estimated Marginal Means for Intention to apply
Testimonial-source
123
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text-based but testimonial-source is company-dependent.
Profile Plot 26. Dependent Variable- Intention to Apply, Treatment Variables-
Testimonial-type, Testimonial-source
Figure 36 demonstrates the marginal means of intention to apply, given testimonial-
type and testimonial-source as treatment variables.
Figure 36: Profile Plot 26: DV- Intention to Apply; IV-Testimonial-type,
Testimonial-source
The profile plot shows that for realistic employee testimonial and company-neutral
source of employee testimonial, estimated marginal means for intention to apply is
highest. Intention to apply is lower than the above case in both the following
situations:
a. Realistic employee testimonial and company dependent testimonial-source
b. Positive employee testimonial and company neutral testimonial-source.
3.84
3.31
3.76
4.03
Estimated Marginal Means for Intention to apply Testimonial-source
Testimonial-type
124
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but testimonial-source is company-dependent.
Profile Plot 27. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-
dependent)
Figure 37 demonstrates the marginal means of organizational attractiveness, given
testimonial-type, preview-mode and testimonial-source (company-dependent) as
treatment variables.
Figure 37: Profile Plot 27: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-dependent)
2.77
3.48 3.50
3.74
Estimated Marginal Means for Organizational Attractiveness
Testimonial-type At testimonial-source
125
The profile plot shows that, when positive employee testimonials are presented via
company-dependent sources, along with audio-visual corporate previews, estimated
marginal mean for organizational attractiveness is highest. In case of realistic
testimonials, whether presented through company dependent channels or whether the
preview-mode is text or video based does not create any significant difference in
attractiveness perception. It is observed that, if positive testimonials are to be
propagated through company-dependent channels, in such case corporate previews
should not be presented in text based mode, since it significantly lowers attraction of
the employer.
Profile Plot 28. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 38 demonstrates the marginal means of organizational attractiveness, given
testimonial-type, preview-mode and testimonial-source (company-neutral) as
treatment variables.
Figure 38: Profile Plot 28: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-neutral)
3.47
3.58
3.62
3.67
Estimated Marginal Means for Organizational Attractiveness
Testimonial-type At testimonial-source
126
The profile plot shows that, when positive employee testimonials are presented via
company-neutral sources, along with text based corporate previews, estimated
marginal mean for organizational attractiveness is highest. Perception of
attractiveness is slightly less if the same is provided with audio-visual corporate
previews. It is further observed that, if realistic testimonials are to be propagated
through company-neutral channels, in such case corporate previews should not be
presented in audio-visual based mode, since it significantly lowers attraction of the
employer.
Profile Plot 29. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 39 demonstrates the marginal means of intention to apply, given testimonial-
type, preview-mode and testimonial-source (company-dependent) as treatment
variables.
Figure 39: Profile Plot 29: DV- Intention to Apply; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
2.82
3.77
3.80 3.97
Testimonial-type
Estimated Marginal Means for Intention to apply
At testimonial-source = Company Dependent
127
The profile plot shows that, when realistic employee testimonials are presented via
company-dependent sources, along with text-based corporate previews, estimated
marginal mean for intention to apply is highest.
It is further observed that, if positive testimonials are to be propagated through
company-dependent channels, in such case corporate previews should not be
presented in text based mode, since it significantly lowers attraction of the employer.
Profile Plot 30. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 40 demonstrates the marginal means of organizational attractiveness, given
testimonial-type, preview-mode and testimonial-source (company-neutral) as
treatment variables.
Figure 40: Profile Plot 30: DV- Intention to Apply; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-neutral)
The profile plot shows that, when realistic employee testimonials are presented via
company-neutral sources, along with audio-visual corporate previews, estimated
3.83
3.67
4.18
3.85
Estimated Marginal Means for Intention to apply
Testimonial-type At testimonial-source
128
marginal mean for organizational attractiveness is highest. In case of realistic
testimonials, whether the preview-mode is text or video based does not create any
significant difference in attractiveness perception. It is further observed that, if
positive testimonials are to be propagated through company-neutral channels, in such
case corporate previews should not be presented in text based mode, since it
significantly lowers attraction of the employer.
129
5.1.5 Phase III Results (Study 1)
The experimental design for phase 3 can be represented as under:
Intention to apply= β0 + β1 (Preview-mode) + β2 (Testimonial-type) + β3
(Testimonial-source) + β4 (Preview-mode x Testimonial-type) + β5 (Testimonial-type
x Testimonial-source) + β6 (Preview-mode x Testimonial-source) + β7 (Preview-mode
x Testimonial-type x Testimonial-source) + β8 (Perceived Quality) + β9 (Perceived
Quality) + β10 (Organizational Attractiveness) + Error
Analysis of covariance (ANCOVA) was conducted using SPSS 16 with the dependent
variable intention to apply, organizational attractiveness, perceived credibility and
perceived quality as covariates and preview-mode (two factors-text/ audio-visual),
testimonial-type (two factors-positive/ realistic) and testimonial-source (two factors-
company dependent/ company neutral) as fixed factors.
The results revealed that perceived quality [F (1, 446) = 1.528, p > .10] and perceived
credibility [F (1, 446) = .058, p > .10] are not significant covariates of intention to
apply, while organizational attractiveness [F (1, 446 = 268.439, p < .001, effect size =
.376] is significant predictor of intention to apply. The treatments also show
significant main effect with testimonial-type showing a moderate effect size (.084).
The two way interaction between preview-mode and type of testimonial is not
significant [F (1, 446) = 2.848, p >.05] and that between preview-mode and
testimonial-source is significant only at 95 % confidence level [F (1, 446) = 5.535, p
<.05, effect size = .012].
130
TABLE 14: Test of Significance for Dependent Variable Intention to Apply
[Study 1]
Predictor variable F Value Sig. Partial Eta
Squared
Observed
Power
Perceived Quality 1.528 .217 .003 .234
Perceived Credibility .058 .810 .000 .057
Organizational Attraction 268.439 .001 .376 1.000
Preview-mode 11.044 .001 .044 .612
Testimonial-type 40.870 .001 .084 1.000
Testimonial-source 8.423 .004 .02 .525
Previewmode X Testimonial-type 2.848 .092 .006 .391
Previewmode X Testimonial-source 5.535 .019 .012 .651
Testimonial-type X Testimonial-source .171 .680 .000 .070
Previewmode X Testimonial-type X
Testimonial-source
12.515 .001 .177 1.000
The two way interaction between testimonial-type and testimonial-source [F (1, 446)
= .409, p > .05] is not significant at 95 % confidence level. At the same time, it is
observed that the three way interaction between preview-mode, testimonial-type and
testimonial-source [F (1, 446) = 12.515, p < .001, effect size = .177] has significant
131
effect on intention to apply. Therefore, the results suggest that organizational
attractiveness is a significant covariate of job-seekers‘ intention to apply for jobs. The
treatments have significant main effects on intention to apply though the three way
interaction between the treatment variable sis is considerably more significant with a
better effect size.
5.1.5.1 Profile Plot Analysis
Profile Plot 31. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type
Figure 41 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables.
Figure 41: Profile Plot 31: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
The profile plot shows that for realistic employee testimonial and audio-visual corporate
preview, estimated marginal means for intention to apply is highest. Intention to apply is
lower than the above case in both the following situations:
3.73
3.45
3.87
3.97
Estimated marginal means of intention to apply Testimonial-type
132
a. Realistic employee testimonial and text based preview
b. Positive employee testimonial and video based preview.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive and preview is text-based.
Profile Plot 32. Dependent Variables-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 42 demonstrates the marginal means of intention to apply, given testimonial-
source and preview-mode as treatment variables.
Figure 42: Profile Plot 32: DV- Intention to Apply; IV-Testimonial-source,
Preview-mode
The profile plot shows that for audio-visual preview and company-independent source
of employee testimonial, estimated marginal means for intention to apply is highest.
Intention to apply is lower than the above case and approximately same in both the
following situations:
a. text based preview and company neutral testimonial-source
3.69 3.66
3.69
4.00
Estimated marginal means of intention to apply Testimonial-source
133
b. video based preview and company dependent testimonial-source.
Profile Plot 33: Dependent Variable- Intention to Apply; Treatment Variables-
Testimonial-type, Testimonial-source
Figure 43 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables. The profile plot shows that for positive
employee testimonial and company-dependent source of employee testimonial,
estimated marginal means for intention to apply is highest.
Figure 43: Profile Plot 33: DV- Intention to Apply; IV-Testimonial-type,
Testimonial-source
Intention to apply is lower than the above case in both the following situations:
a. Realistic employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company neutral testimonial-source
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-dependent.
4.00
3.67
3.85
3.50
Estimated marginal means of intention to apply
Testimonial-source
Testimonial-type
134
Profile Plot 34: Dependent Variable- Intention to Apply; Treatment Variables-
Testimonial-type, Preview-mode at Testimonial-source (Company-dependent)
Figure 44 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables.
Figure 44: Profile Plot 34: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode at Testimonial-source (Company-dependent)
The profile plot shows that for realistic employee testimonial and text based preview,
estimated marginal means for intention to apply is highest. Intention to apply is lower
than the above case in both the following situations:
a. Realistic employee testimonial and video preview
b. Positive employee testimonial and video preview
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive and preview is text based when the testimonial source is
company dependent.
3.68
3.23
3.73
3.97
Testimonial-type
Estimated marginal means of intention to apply
At Testimonial-source=company dependent
135
Profile Plot 35: Dependent Variable- Intention to Apply; Treatment Variables-
Testimonial-type, Preview-mode at Testimonial-source (Company-neutral)
Figure 45 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables.
Figure 45: Profile Plot 35: DV- Intention to Apply; IV-Testimonial-type, Preview-
mode at Testimonial-source (Company-neutral)
The profile plot shows that for realistic employee testimonial and video preview,
estimated marginal means for intention to apply is highest when employee
testimonials are advertised through company neutral source. Intention to apply is
lower than the above case in both the following situations:
a. Realistic employee testimonial and text preview
b. Positive employee testimonial and video preview.
3.77
3.57
4.22
3.80
Estimated marginal means of intention to apply
At Testimonial-source=company-neutral Testimonial-type
136
5.1.6 Phase IV Results (Study 1)
A covariance based structural equation model (SEM) is estimated to assess the
hypothesized relationship. The fit of the structural model is acceptable, with χ2
(52)=
238.498 (p < .01), goodness of fit index (GFI)= .958, comparative fit index (CFI)=
.931, incremental fit index (IFI)= .924 and root mean square error of approximation
(RMSEA)= .063. CMIN/df is calculated to be 4.587. Based on these results, the
measurement model indicates an acceptable model fit of the data (Hair et al., 1998).
Consistent with the different cut-off criteria provided in the literature (Joreskog, 1993;
Hu and Bentler 1999), it is concluded that the hypothesized causal model is within the
acceptable range of all the fit statistics.
Please refer to Table 15 for structural path model‘s estimates and a summary of the
results of the hypothesis tests. In H10a it is proposed that perceived quality positively
affects job-seekers‘ perception of organizational attraction (β1a = .197, p = .014) and
in H10b it is proposed that perceived quality positively affects their intention to apply
(β2a = .155, p = .024). Both these hypotheses receive support at 95 % confidence
interval. Our next two hypotheses (H11a and H11b) are not supported because perceived
credibility is not found to have a positive effect on organizational attractiveness (β2a =
-.024, p = .795) and intention to apply (β2b = .046, p = .318). The hypotheses H4a, H4b
and H4c also do not get any support as preview-mode, testimonial-type and
testimonial-source are found to have no significant effect on organizational
attractiveness (p > .05). Organizational attractiveness on the other hand significantly
influences intention to apply (β6 = .577, p = .001), providing support for H14. In order
to evaluate whether the impact of perceived quality on job-seekers‘ intention to apply
is fully mediated by organizational attractiveness, a path from perceived quality to
137
intention to apply is created and the chi-square difference between the two models is
evaluated.
Table 15: Moderating Effect of Preview-mode, Testimonial-type and
Testimonial-source and Mediating Effect of Organizational Attractiveness
[Study 1]
Hypothesized Path Standardized path coefficients
Hypothesis β S.E. C.R.
Perceived Quality → Attractiveness
H 10a .197** .092 2.467
Perceived Credibility→ Attractiveness
H11a -.024 -.259 .795
Preview-mode →Attractiveness
H4a -.034 -.017 -.389
Testimonial-type → Attractiveness
H4b -.027 .087 .563
Testimonial-source → Attractiveness
H4c .038 .087 .851
Perceived Quality→ Intention to Apply
H10b .155** .045 1.217
Perceived Credibility → Intention to apply
H11b .046 .046 .999
Attractiveness → Intention to Apply H14 .577* .039 14.384
MOD1 (Quality*Preview-mode) Attractiveness H19a .048 .087 .862
MOD2 (Quality*Testimonial-type)Attractiveness H19b .035 .087 .563
MOD3 (Quality*Testimonial-source)Attractiveness H19c -.015 .086 -.262
MOD4 (Credibility*Preview-mode)Attractiveness H20a .058 .085 1.007
MOD5 (Credibility*Testimonial-type)Attractiveness H20b .135** .077 2.425
MOD6 (Credibility*Testimonial-source)Attractiveness H20c .092** .079 1.559
Fit Statistics
χ2 (df) 238.498 (52)
p .000
CMIN/df 4.587
GFI .958
CFI .931
IFI .924
RMSEA .063
**p<.05; *p<.01
Notes: β represents standardized path coefficient
138
Figure 46: Structural Model with Path Coefficients for Study 1
It is observed that the independent variable (perceived quality) has significant
relationship (βc =.155, p = .001, S.E. = .045) with the dependent variable intention to
apply and the mediating variable organizational attractiveness (βa = .197, p = .001,
S.E.a = .092); and the mediating variable also significantly predicts the dependent
variable (βb = .577, p = .001, S.E.b = .039). These results show that the hypothesized
model satisfy the basic conditions for mediation as proposed by Barron and Kenny
MOD 3 MOD 2
Perceived
Credibility
Organizational
Attractiveness
Intention to
Apply
Perceived
Quality
MOD 1
Preview-
mode
Testimonial-
type
Testimonial-
source
MOD 4 MOD 5 MOD 6
.197** .155**
. 577*
.046 -.024
-.034
-.027
.038
.048
.035
-.015
.058
.135**
.122**
139
(1986). Afterwards, a Sobel‘s test (Sobel, 1982) was carried out to test the nature of
mediation. To calculate Sobel‘s test statistic, the following formula (Sobel, 1982) was used:
Where,
a= regression coefficient for the relationship between the independent variable and the
mediator
b= the regression coefficient for the relationship between the mediator and the dependent
variable
SEa = the standard error of the relationship between the independent variable and the
mediator and
SEb = the standard error of the relationship between the mediator variable and the
dependent variable
Sobel‘s test for significance = 2.11922306 (> 1.96), with one-tailed probability=
.01703581**, two-tailed probability= .03407162**. The test statistic was significant
at 95 % level of significance, indicating partial mediating effect (Preacher and Hayes
2004) of organizational attractiveness on the relationship between perceived quality
and intention to apply. Thus we received partial support for our hypothesis 21 for
study 1.
Next, in order to evaluate whether the impact of perceived credibility on job-seekers‘
intention to apply is fully mediated by organizational attractiveness, a path from
perceived credibility to intention to apply is created and the chi-square difference
between the two models is evaluated. It is observed that the independent variable
(perceived credibility) has no significant relationship with the dependent variable
140
(intention to apply) and the mediating variable (organizational attractiveness); though
the mediating variable significantly predicts the dependent variable. These results
show that the hypothesized model does not mediate the relationship between
perceived quality and intention to apply (Barron and Kenny). Therefore, H 22 for
study 1 received no support.
In order to test for moderating effect of preview-mode, testimonial-type and
testimonial-source on the relationship between perceived quality and firm
attractiveness, interaction terms are created for perceived quality with respect to
preview-mode, testimonial-type and testimonial-source (MOD1, MOD2 and MOD3).
Similarly, interaction terms are created to test the moderating impact of -mode,
testimonial-type and testimonial-source on the relationship between perceived
credibility and organizational attractiveness (MOD4, MOD5 and MOD6). As has
been observed, perceived quality has significant main effect on organizational
attractiveness. But no significant interaction effect is observed with respect to the
moderating variables created by perceived quality and the treatment variables. Hence
none of H19a, H19b and H19c received any support for study 1.
As for moderating impact of preview-mode, testimonial-type and testimonial-source
on the relationship between perceived credibility and organizational attractiveness, it
is observed that preview-mode (p = .314) does not demonstrate any significant
moderating effect on the relation between perceived credibility and firm attraction.
However, testimonial-type (p = .015) and testimonial-source (p = .019) are observed
to have significant moderating impact on the credibility-firm attraction relationship.
At the same time, perceived credibility does not have any main effect on
organizational attractiveness (p = .318). Hence H20a is not supported but H20b and H20c
receive complete support.
141
5.1.7 Phase V Results (Study 1): Post – Hoc Test Using Scheffe’s Procedure
Post-hoc tests were conducted with preview-mode (text/ audio-visual), testimonial-
type (positive/ realistic) and testimonial-source (company-dependent/ company-
neutral) on intention to apply, using Scheffe‘s test (Klockars and Hancock 2000). The
findings gave indications that in terms of intention to apply, perceptions are different
for those jobseekers in Group A who have viewed the website with treatment
condition 1 (positive employee testimonial, text based preview and company
dependent testimonial source) compared to Groups E, F, G, and H which correspond
to treatment conditions 5, 6, 7 and 8 (mean differences significant at p < .05).
Perceptions are different for those jobseekers in Group B who have viewed the
website with treatment condition 2 (realistic employee testimonial, text based preview
and company-dependent testimonial source) compared to Groups F, G, and H which
correspond to treatment conditions 6, 7 and 8 (MDs significant at p < .05). However,
perceptions are not significantly different when job-seekers under Groups A, C, D and
E are compared with those in Group B (mean difference significant at p < .05). With
respect to Group C (treatment 3), Groups E, F, G and H contributed to different
perception formation for intention to apply at 95 % confidence level but Groups A, B
and D are not able to create any difference in jobseekers‘ intention to apply at 95 %
confidence level. As for those exposed to website corresponding to treatment 4 in
Group D, perception of applying for job is different compared to those who viewed
treatments 6, 7 and 8 in Groups F, G and H (MD significant at p < .05). Intention to
apply for Group E job-seekers is markedly different from that for job-seekers in
Group G and Group H at 95 % confidence level, and that for Group F job-seekers is
significantly different from Groups A, C and D at 95 % confidence level.
142
Table 16: Post-hoc Analysis Using Scheffe’s Procedure [Study 1]
Referent group (I) Comparison group (J) Mean Difference (I-J) Std. Error Significance
Group A Group B .23385 .10701 .687
Group C .16436 .09771 .900
Group D .00617 .10251 1.000
Group E .10403** .09855 .033
Group F .00239* .11739 .000
Group G .03812* .12008 .000
Group H .17162** .11739 .011
Group B Group A -.23385 .10701 .687
Group C -.06949 .09372 .999
Group D -.22768 .09872 .621
Group E -.12982 .09460 .966
Group F -.23146* .11409 .006
Group G -.19573* .11686 .002
Group H -.06223* .11409 .000
Group C Group A -.16436 .09771 .900
Group B .06949 .09372 .999
Group D -.15819 .08855 .866
Group E -.06033* .08394 .001
Group F -.16197** .10542 .037
Group G -.12625* .10841 .007
143
Group H .00726* .10542 .000
Group D Group A -.00617 .10251 1.000
Group B .22768 .09872 .621
Group C .15819 .08855 .866
Group E .09786 .08948 .991
Group F -.00378* .10988 .000
Group G .03194* .11275 .000
Group H .16545** .10988 .043
Group E Group A -.10403 .09855 .993
Group B .12982 .09460 .966
Group C .06033 .08394 .999
Group D -.09786 .08948 .991
Group F -.10164 .10620 .996
Group G -.06592* .10917 .000
Group H .06759* .10620 .000
Group F Group A -.00239* .11739 .000
Group B .23146*** .11409 .066
Group C .16197** .10542 .037
Group D .00378* .10988 .000
Group E .10164 .10620 .996
Group G .03573 .12644 1.000
Group H .16923*** .12388 .067
Group G Group A -.03812* .12008 .000
144
Group B .19573* .11686 .002
Group C .12625 .10841 .987
Group D -.03194* .11275 .000
Group E .06592* .10917 .000
Group F -.03573* .12644 .000
Group H .13350 .12644 .993
Group H Group A -.17162** .11739 .045
Group B .06223* .11409 .000
Group C -.00726* .10542 .000
Group D -.16545** .10988 .043
Group E -.06759* .10620 .000
Group F -.16923** .12388 .067
Group G -.13350 .12644 .093
* p<.01, **p< .05
Group G job-seekers have different intention to apply compared to those in Groups A,
B, D, E and F. Finally, Group H is found to create difference in intention to apply at
95 % confidence level for all other groups except Group G. Observing the mean
differences for intention to apply, it may be concluded that treatment 7 (audio-visual
preview, positive testimonial type and company–neutral testimonial source) and
treatment 8 (audio-visual preview, realistic testimonial type and company–neutral
testimonial source) is the best web design to trigger the highest level of intention to
apply.
145
5.2 Results from Study 2
5.2.1 Sample Demographics for Study 2
In the second study, another 2x2x2 full factorial between-group design (Trochim
2008) is implemented to test the validity of hypothetical models 1, 2, 3 and 4 using an
web-based experimental survey with job-seekers who are already employed for the
past 1-2 years who have registered their details in various third-party job portals such
as Naukri.com, Monster.com etc. and have been actively involved in online job
search. Taking cue from the India Online Landscape Survey, job-seekers with either
or both engineering and MBA backgrounds and with less than 2 years of experience
have been identified as the target population. At first, email id of registered users at
Naukri and Monster, two of India‘s largest job portals have been identified using
search key terms ‗engineering‘, ‗B.E.‘, B.Tech.‘, ‗MBA‘, PGDM‘, ‗1-2 years
experience‘ and with addresses bearing names of metro cities (Kolkata, Mumbai,
Delhi NCR, Hyderabad, Chennai, Bengaluru, Pune and Ahmedabad). The profile of
the registered users could be accessed through the co-operation of third-party
recruiters who have employer login access to these websites.
In order to maintain the confidentiality of the individuals, only a list of email-ids was
provided by the third-party recruiters, who have either or both MBA and engineering
degrees, have worked for at most 1-2 years and who are located in the various metro
cities in India. This helped to ensure that the list consisted of those urban Indian job-
seekers who occupy a significant portion of the active job-seeker population
registered in various E-recruitment websites as per the India Online Landscape
Survey (2012). The final list included 1425 email-ids with respondent location
mentioned with corresponding email ids. Random allocation technique was used to
146
divide the total sample into eight different groups and emails were sent to the
respondents to take part in the study 100 at a time.
The respondents were then contacted through a formal email briefing them about the
purpose of the study and then requesting them to take part in the survey. In total, 361
recipients responded to the invitation and agreed to take part in the study with
response rate of 25.33 %. The response rate is coherent with the expected standard for
web-based surveys (Nulty, 2008). Then using the online random number generator
(research randomizer), the respondents were assigned to eight different groups of
approximate size of 45 members each labeled from 1 to 8. Eight different
questionnaire surveys were created using an online survey development company
‗Webs‘ (www.webs.com) with each survey containing a different link to eight types
of websites different in terms of the treatment conditions.
The online survey questionnaires with different website manipulations were then
randomly sent to the randomly selected groups to achieve completely randomized
experiment. For example, those respondents in group 1 were directed to a treatment 3
website which had audio-visual preview-mode, positive employee testimonial
propagated through a company dependent platform. Similarly, respondents in group 4
visited a website of ECS which had text-based corporate preview, realistic employee
testimonial and company-neutral source. Like this, eight groups of respondents
received eight different treatments.
The online questionnaire was designed with a very simple interface so that
participants can understand the necessary actions to be action to complete the survey.
All questions were mandatory, i.e., unless respondents answered all the queries, the
questionnaire could not be submitted. The completed responses which were submitted
got recorded in an excel file. Hence in study 2, there was no case of missing data. The
147
final sample size of 361 can be considered adequate for running experimental designs
with a statistical power of 80 % and effect size of 0.5 (Cohen 1988).
Table 17: Demographic Characteristics for Sample of Study 2
Demographic
Variable
Category Frequency Percentage
Gender Male 229 63.4
Female 132 36.6
Age <20 yrs nil 0
20-25 yrs 271 75.1
26-30 yrs 90 24.9
Educational
qualification
Graduate 161 44.59
Postgraduate 184 50.96
Professional courses 16 4.432
Undergraduate 0 0
Educational
Specialization
Engineering (BE/ B Tech) 124 34.35
MBA 168 46.54
MCA/BCA 29 8.033
MALLB/ BALLB 16 4.432
148
Others 24 6.65
Work Experience < 1 yr 54 15.0
1-2 yrs 165 45.7
2-3 yrs 107 29.6
3-4 yrs 35 9.7
Involvement in online
job search
< 1 yr 85 23.54
1-2 yrs 138 38.23
2-3 yrs 107 29.64
3-4 yrs 20 5.54
> 4 yrs 11 3.04
Registration in online
job portals
Nil 0 0
Naukri 355 98.37
Monster 305 84.48
Timesjobs 188 52.07
Shine 79 21.88
Others 20 5.54
149
Location Delhi NCR 56 15.52
Mumbai 54 14.95
Kolkata 49 13.57
Chennai 46 12.74
Pune 36 9.97
Bengaluru 42 11.63
Ahmedabad 33 9.14
Hyderabad 45 12.46
As far as sample demographics are concerned, approximately 64 % respondents of
study 2 are male and 75 % are in the age group of 20-25 years. Nearly 83 % of these
individuals have less than one year of experience. Respondents are mostly graduate
(44.5 %) and postgraduate (51%) with MBA and B.E./B.Tech degrees accounting for
80 % of the educational qualification which approximates the actual population data
obtained in the India Online report. Over 60 % respondents reported work experience
in the range of 1-2 years while another 30 % reported work experience of 2-3 years.
Of high interest is the fact that nearly 68 % respondents replied that they have been
actively involved in online job search process in the past 1-3 years. 98 % respondents
replied that they have registered profiles in Naukri.com while 90 % replied that they
have profile in Monster.com, followed by Timesjobs.com at a distant third (52%).
150
5.2.2 Data Analysis for Study 2
5.2.2.1 Descriptive Statistics
Overall, the means gave indications that the respondents in our sample had fairly high
perceptions of quality of the website (M = 3.515, SD = 1.1082) and credibility (M =
3.368, SD = 1.1352). Perceived organizational attractiveness (M = 3.798, SD = .9071)
and intention to apply (M = 3.471, SD = 1.0464) were also high in the individuals.
The correlation matrix (Table 18) gave indications that perceived credibility is
positively correlated with perceived quality (r = .420), organizational attractiveness (r
= .386) and intention to apply (r = .537). Perceived quality is also positively
correlated with attractiveness (r = .356) and intention to apply (r = .651).
Organizational attractiveness is found to have a positive correlation with intention to
apply (r = .431). The reliability coefficients have been reported in the diagonals for
each construct. Cronbach Alpha values are all above 0.9 (Nunally 1978) indicating
good reliability of the scales.
Table 18: Descriptive Statistics and Correlation Diagnostics [Study 2]
Name of construct M SD 1. 2. 3. 4.
1.Perceived Credibility 3.3684 1.13529 .948** .820 .386 .937
2. Perceived Quality 3.5152 1.10826 .820 .961** .356 .851
3. Organizational Attractiveness 3.7978 .90713 .386 .356 .939** .431
4. Intention to Apply 3.4709 1.04638 .937 .851 .431 .930**
* Correlation is significant at the 0.01 level (2-tailed).
**Cronbach α reliability coefficients reported in the diagonal, N=361
151
5.2.2.2 Manipulation Checks (Study 2)
Once again, three separate manipulation checks were conducted with the overall
sample to verify the following three premises- to check the perceived difference in
intention to apply for the audio-visual-preview and text preview, to check the
perceived difference in intention to apply for positive and realistic testimonial and to
check the perceived difference in intention to apply for company-dependent and
company neutral testimonial sources.
Table 19: Results from Manipulation Check 1 [Study 2]
Descriptive Statistics t-test for equality of means
Manipulation
(Preview-mode)
N
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Text 180 3.8389 .75608 .0563 Equal variance
Assumed
Equal variance
not assumed
-1.881
-1.881
359
359
.041**
.041**
Audio-visual 181 3.9890 .76004 .0565
*p<.05
The mean intention to apply for text preview is found to be 3.839 and that for audio-
visual previews is 3.989, and the mean difference is significant at 5 % level of
significance (refer to table 19). This indicates that the manipulation of preview-mode
was comprehended properly by the respondents. The mean intention to apply for
positive testimonial is found to be 3.395 and that for audio-visual previews is 4.037,
and the mean difference is significant at 1 % level of significance (refer to table 20).
This indicates that the manipulation of testimonial-type was also comprehended
properly by the respondents. The mean intention to apply for company dependent
152
testimonial-source is found to be 3.6201 and that for audio-visual previews is 3.9725,
and the mean difference is significant at 99 % level of confidence.
Table 20: Results from Manipulation Check 2 [Study 2]
Descriptive Statistics t-test for equality of means
Manipulation
(Testimonial-type)
N
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Positive 172 3.395 1.0235 .07804 Equal variance
Assumed
Equal variance
not assumed
-6.730
-6.647
359
319
.000*
.000*
Realistic 189 4.037 .78123 .05683
*p<.01
This provides evidence that the manipulation of testimonial-source was
comprehended properly by the respondents.
Table 21: Results from Manipulation Check 3 [Study 2]
Descriptive Statistics t-test for equality of means
Manipulation
(Testimonial-source)
N
Mean SD Std.
Error
Variance t value Df Sig.
(2 tailed)
Company-dependent 179 3.6201 .94859 .0709 Equal variance
Assumed
Equal variance not
assumed
-3.757
-3.753
359
351
.000*
.000*
Company-neutral 182 3.9725 .83057 .0616
*p<.01
153
5.2.3 Phase I results (Study 2)
The experimental design for phase 1 (study 2) can be represented as under:
Perceived Quality + Perceived Credibility + Organizational Attractiveness +
Intention to apply= β0 + β1 (Preview-mode) + β2 (Testimonial-type) + β3
(Testimonial-source) + β4 (Preview-mode x Testimonial-type) + β5 (Testimonial-type
x Testimonial-source) + β6 (Preview-mode x Testimonial-source) + β7 (Preview-
mode x Testimonial-type x Testimonial-source) + Error
A multivariate analysis of variance (MANOVA) was conducted using SPSS 16 with
the dependent variable block including the constructs of perceived credibility,
perceived quality, organizational attractiveness and intention to apply and preview-
mode (two factors-text/ audio-visual), testimonial-type (two factors-positive/ realistic)
and testimonial-source (two factors- company dependent/ company neutral) as fixed
factors. The Levene‘s test for homogeneity of variance was found to be insignificant
(p > .05) in case of each dependent variable which helped in meeting the necessary
assumptions to conduct a MANOVA (refer Table 22).
TABLE 22: Levene’s test for MANOVA [Study 2]
Name of variable F df1 df2 Significance
Perceived Quality 5.093 7 353 .145
Perceived Credibility 3.398 7 353 .072
Organizational Attractiveness 4.877 7 353 .099
Intention to apply 5.443 7 353 .067
The results from the MANOVA analysis indicate that the multivariate test of
significance (Wilk‘s Lambda) is significant at 95 % confidence level and preview-
mode, testimonial-type and testimonial-source, as well as the two-way and three-way
interaction terms all have significant F values. The multivariate tests indicate presence
154
of significant main and interaction effect of the treatment variables (preview-mode,
testimonial-type and testimonial-source) on the dependent variables. The observed
power for the above variables (greater than 0.8) also indicates strong possibility of
main and interaction effect (Hair et al. 1998).
TABLE 23: Multivariate Test of Significance for Perceived Quality,
Perceived Credibility, Organizational Attractiveness and Intention to
Apply [Study 2]
Name of Variable Test of
Significance
Value F Value Significance Partial Eta
Squared
Observed
Power
Preview-mode Wilk‘s
Lambda .668 43.514 .000 .332 1.000
Testimonial-type Wilk‘s
Lambda .812 20.304 .000 .188 1.000
Testimonial-source Wilk‘s
Lambda .512 83.532 .000 .488 1.000
Preview-mode X
Testimonial-type
Wilk‘s
Lambda .944 5.152 .000 .056 .967
Preview-mode X
Testimonial-source
Wilk‘s
Lambda .969 2.802 .026 .031 .815
Testimonial-type X
Testimonial-source
Wilk‘s
Lambda .934 6.150 .000 .066 .987
Preview-mode X
Testimonial-type X
Testimonial-source
Wilk‘s
Lambda .984 1.438 .021 .086 .847
Table 24 shows the effect of treatment variables on each dependent variable separately.
The univariate results indicate that preview-mode does not have any main effect on
perceived quality [F (4, 350) = 2.962, p = .085] and perceived credibility [F (4, 350) =
4.536, p = .087] of E- recruitment platforms for the jobseekers. However, preview-mode
is found to have a significant main effect on organizational attractiveness [F (4, 350) =
115.475, p = .001, effect size = .247] and intention to apply [F (4, 350) = 76.596, p =
.001, effect size= .178]. Testimonial-type has a main effect on perceived credibility [F (4,
155
350) = 79.264, p = .001, effect size = .184], organizational attractiveness [F (4, 350) =
42.579, p = .001, effect size = .108] and intention to apply [F (4, 350) = 39.848, p = .001,
effect size = .101]. Testimonial-source is noticed to have significant main effect on
perceived quality [F (4, 350) = 329.482, p = .001, effect size = .483] and organizational
attractiveness [F (4, 350) = 92.234, p = .005, effect size = .207].
The two-way interaction effect of preview-mode and testimonial-type is significant for
only perceived credibility, that too at 95 % confidence level [F (4, 350) = 6.520, p = .011,
effect size = .058]. Two-way interaction between preview-mode and testimonial-source is
significant only on perceived quality [F (4, 350) = 10.147, p = .002, effect size = .068].
Effect of the two-way interaction between testimonial-type and testimonial-source is
again significant for only perceived quality [F (4, 350) = 4.589, p = .033, effect size =
.083].
The three-way interaction between preview-mode, testimonial-type and testimonial-
source is significant at 99 % confidence interval for perceived quality [F (4, 350) = 7.677,
p = .001, effect size = .085], perceived credibility [F (4, 350) = 8.473, p = .001, effect size
= .094], organizational attractiveness [F (4, 350) = 10.485, p = .001, effect size = .194]
and intention to apply [F (4, 350) = 9.327, p = .001, effect size = .152]. The effect sizes
(partial eta squared) for all significant effects range between .08 - .50 which indicates
small to moderate effect of the treatments on the dependent variables (Cohen 1988). The
findings suggest that all four dependent variables are perceived differently in case of
only the three-way interaction. This indicates that websites with different
combinations of preview-mode, testimonial-type and testimonial-source have a
possibility of being differently perceived by job-seekers while seeking job-related
information through internet compared to websites with either a single treatment or a
combination of any two treatments.
156
TABLE 24: Univariate Test of Significance for Perceived Quality, Perceived
Credibility, Organizational Attractiveness and Intention to Apply [Study 2]
Main & Interaction Effects
F Value
Partial
Eta
Squared
Observed
power
Preview-mode Perceived Quality
Preview-modePerceived Credibility
Preview-modeOrganizational Attractiveness
Preview-modeIntention to apply
2.962
3.536
115.745*
76.596*
.009
.018
.247
.178
.488
.542
1.000
1.000
Testimonial-type Perceived Quality
Testimonial-type Perceived Credibility
Testimonial-type Organizational Attractiveness
Testimonial-type Intention to apply
4.706
79.264*
42.579*
39.848*
.024
.184
.108
.101
.737
1.000
1.000
1.000
Testimonial-source Perceived Quality
Testimonial-source Perceived Credibility
Testimonial-source Organizational Attractiveness
Testimonial-source Intention to apply
329.482*
6.520
92.234*
4.433
.483
.019
.207
.013
1.000
.655
1.000
.388
Preview-mode X Testimonial-type Perceived Quality
Preview-mode X Testimonial-type Perceived Credibility
Preview-mode X Testimonial-type Organizational Attractiveness
Preview-mode X Testimonial-typeIntention to apply
3.811
6.520**
3.799
1.415
.011
.058
.011
.004
.495
.821
.494
.220
Preview-mode X Testimonial-source Perceived Quality
Preview-mode X Testimonial-source Perceived Credibility
Preview-mode X Testimonial-source Organizational Attractiveness
Preview-mode X Testimonial-source Intention to apply
10.147*
1.667
.521
1.621
.068
.005
.001
.005
.888
.251
.111
.246
157
Testimonial-type X Testimonial-source Perceived Quality
Testimonial-type X Testimonial-source Perceived Credibility
Testimonial-type X Testimonial-source Organizational Attractiveness
Testimonial-type X Testimonial-source Intention to apply
4.589**
.366
1.438
.557
.083
.001
.004
.002
.870
.093
.223
.116
Preview-mode X Testimonial-type X Testimonial-source Quality
Preview-mode X Testimonial-type X Testimonial-source Credibility
Preview-mode X Testimonial-type X Testimonial-source Attractiveness
Preview-mode X Testimonial-type X Testimonial-source Intention to
apply
7.677*
8.473*
10.485*
9.327*
.085
.094
.194
.152
.852
.828
.929
.946
*p< 0.01, **p<0.05
5.2.3.1 Profile Plot Analysis
Profile Plot 36. Dependent Variable-Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-type
Figure 47 demonstrates the marginal means of perceived quality, given preview-mode
and testimonial-type as treatment variables.
Figure 47: Profile Plot 36: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type
3.37
3.05
3.11
3.73
Estimated Marginal Means of Perceived Quality Testimonial-type
158
The profile plot shows that for video-based preview and realistic employee
testimonial, perceived quality of website is highest. Perceived quality was lower than
the above case in both the following situations:
a. Video based preview and positive testimonial
b. Text based preview and realistic testimonial.
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is audio-visual and testimonial-type is positive.
Profile Plot 37. Dependent variable-Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-source
Figure 48 demonstrates the marginal means of perceived quality, given
preview-mode and testimonial-source as treatment variables. The profile plot shows
that for realistic testimonial posted through company-independent source of employee
testimonial, estimated marginal means for perceived quality of website is highest.
Perceived quality is lower than the above case in the situation where there is positive
testimonial and company neutral testimonial-source.
Figure 48: Profile Plot 37: DV- Perceived Quality; IV- Preview-mode, Testimonial-
source
However, the lowest marginal means were recorded for the scenarios where the
testimonial is positive or realistic and testimonial-type is company-dependent.
2.57 2.53
3.74
4.18
Estimated Marginal Means of Perceived Quality
Testimonial-source
Testimonial-type
159
Profile Plot 38. Dependent Variable-Perceived Quality, Treatment Variables- Preview-
mode, Testimonial-source
Figure 49 demonstrates the marginal means of perceived quality, given preview-mode
and testimonial-source as treatment variables. The profile plot shows that for audio-visual
previews and company-independent source of employee testimonial, estimated marginal
means for perceived quality of website is highest. Perceived quality is lower than the
above case in the following situations:
a. Text preview and company neutral testimonial-source
b. Video preview and company dependent testimonial-source.
c. Text preview and company dependent testimonial-source.
Figure 49: Profile Plot 38: DV- Perceived Quality; IV- Testimonial-type,
Testimonial-source
2.60
2.50
3.53
4.30
Estimated Marginal Means of Perceived Quality
Testimonial-source
160
Profile Plot 39. Dependent Variable-Perceived Credibility, Treatment Variables-
Testimonial-type, Preview-mode
Figure 50 demonstrates the marginal means of perceived credibility, given testimonial-
type and preview-mode as treatment variables. The profile plot shows that for video-
based preview and realistic employee testimonial, perceived credibility of website is
highest. Perceived credibility was lower than the above case in both the following
situations:
a. Text based preview and realistic testimonial
b. Video based preview and positive testimonial.
Figure 50: Profile Plot 39: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text based and testimonial-type is positive.
2.45
3.53
4.15
3.33
Testimonial-type
Estimated Marginal Means of Perceived Credibility
161
Profile Plot 40. Dependent Variable-Perceived Credibility, Treatment Variables-
Testimonial-type, Testimonial-source
Figure 51 demonstrates the marginal means of perceived credibility, given
testimonial-source and testimonial-type as treatment variables. The profile plot shows
that for realistic testimonial and company-neutral source of employee testimonial,
estimated marginal means for perceived credibility of website is highest. Perceived
credibility is lower than the above case in both the following situations:
a. Positive testimonial and company neutral testimonial-source
b. Realistic testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive and testimonial-source is company-dependent.
Figure 51: Profile Plot 40: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-source
3.38
4.15
2.77
3.42
Estimated Marginal Means of Perceived Credibility
Testimonial-source
Testimonial-type
162
Profile Plot 41. Dependent Variable-Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-source
Figure 52 demonstrates the marginal means of perceived credibility, given
testimonial-source and preview-mode as treatment variables. The profile plot shows
that for audio-visual preview and company-neutral source of employee testimonial,
estimated marginal means for perceived credibility of website is highest. Perceived
credibility is lower than the above case in both the following situations:
a. Video preview and company dependent testimonial-source
b. Text preview and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
preview is text based but testimonial-source is company-dependent.
Figure 52: Profile Plot 41: DV- Perceived Credibility; IV-Testimonial-type,
Testimonial-source
Profile Plot 42. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Preview-mode
Figure 53 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-type as treatment variables. The profile plot shows that
for video-based preview and realistic employee testimonial, perceived attractiveness
4.32
3.15
3.50
2.67
Estimated Marginal Means of Perceived Credibility
Testimonial-source
163
of website is highest. Perceived attractiveness was lower than the above case in both
the following situations:
a. Text based preview and realistic testimonial
b. Video based preview and positive testimonial.
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text based and testimonial-type is positive.
Figure 53: Profile Plot 42: DV- Organizational Attractiveness; IV-Testimonial-type,
Previewmode
Profile Plot 43. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Testimonial-source
Figure 54 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and testimonial-source as treatment variables. The profile plot shows
that for realistic testimonial and company-neutral source of employee testimonial,
estimated marginal means for perceived attractiveness of website is highest. Perceived
attractiveness is lower than the above case in both the following situations:
3.40
4.00
3.727
2.245
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
164
a. Positive testimonial and company neutral testimonial-source
b. Realistic testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but testimonial-source is company-dependent.
Figure 54: Profile Plot 43: DV- Organizational Attractiveness; IV-Preview-mode,
Testimonial-source
Profile Plot 44. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-source
Figure 55 demonstrates the marginal means of organizational attractiveness, given
testimonial-source and preview-mode as treatment variables. The profile plot shows
that for realistic employee testimonial and company-independent source of employee
testimonial, estimated marginal means for perceived attractiveness of website is
highest. Perceived attractiveness is lower than the above case in both the following
situations:
3.45
4.10
2.53
3.25
Testimonial-source
Estimated Marginal Means of Organizational Attractiveness
Testimonial-type
165
a. text preview and company neutral testimonial-source
b. audio-visual preview and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the preview
is textual but testimonial-source is company-dependent.
Figure 55: Profile Plot 44: DV- Organizational Attractiveness; IV-Testimonial-
type, Testimonial-source
Profile Plot 45. Dependent Variable-Intention to Apply, Treatment Variables-
Testimonial-type, preview-mode
Figure 56 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables. The profile plot shows that for realistic
employee testimonial and audio-visual preview, estimated marginal means for
intention to apply is highest. Intention to apply is lower than the above case in both
the following situations:
a. positive employee testimonial and video preview
b. Realistic employee testimonial and text preview
4.20
3.40 3.42
2.75
Testimonial-source
Estimated Marginal Means of Organizational Attractiveness
166
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but preview is text based.
Figure 56: Profile Plot 45: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
Profile Plot 46. Dependent Variable- Intention to Apply, Treatment Variables-
Testimonial-type, Testimonial-source
Figure 57 demonstrates the marginal means of Intention to apply, given testimonial-
type and testimonial-source as treatment variables. The profile plot shows that for
realistic employee testimonial and company-independent source of employee
testimonial, estimated marginal means for intention to apply is highest. Intention to
apply is lower than the above case in both the following situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Realistic employee testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but testimonial-source is company-dependent.
3.40
4.11
3.60
2.77
Estimated Marginal Means of Intention to apply
Testimonial-type
167
Figure 57: Profile Plot 46: DV- Intention to Apply; IV-Preview-mode, Testimonial-
source
Profile Plot 47. Dependent variable- Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 58: Profile Plot 47: DV- Intention to apply; IV-Preview-mode,
Testimonial-source
Figure 58 demonstrates the marginal means of intention to apply, given testimonial-
source and preview-mode as treatment variables. The profile plot shows that for
audio-visual preview and company-independent source of employee testimonial,
4.15
3.60
3.33
2.75
4.30
3.45 3.40
2.71
Testimonial-type
Testimonial-source
Estimated Marginal Means of Intention to apply
Estimated Marginal Means of Intention to apply
Testimonial-source
168
estimated marginal means for intention to apply is highest. Intention to apply is lower
than the above case in both the following situations:
a. Text preview and company neutral testimonial-source
b. Video preview and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the preview
is text based but testimonial-source is company-dependent.
Profile Plot 48. Dependent Variable- Perceived Quality, Treatment Variables- Preview-
mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 59: Profile Plot 48: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type, at Testimonial-source (Company-dependent)
Figure 59 demonstrates the marginal means of perceived quality, given testimonial-
type and preview-mode as treatment variables when testimonial source is company-
dependent. The profile plot shows that for company-dependent source of employee
testimonial, if testimonial is realistic and preview of firm is audio-visual, estimated
2.77
2.51
2.50
2.67
Estimated Marginal Means of Perceived Quality
Testimonial-type
At Testimonial-source =
169
marginal means for perceived quality of website is highest. Perceived quality is lower
than the above case in both the following situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive, preview is text based and testimonial-source is company-
dependent.
Profile Plot 49. Dependent Variable- Perceived Quality, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 60: Profile Plot 49: DV- Perceived Quality; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-neutral)
Figure 60 demonstrates the marginal means of perceived quality, given testimonial-
type and preview-mode as treatment variables when source of advertisement is
company neutral. The profile plot shows that for company-neutral source of employee
4.62
3.72
4.00
3.54
Estimated Marginal Means of Perceived Quality
Testimonial-type
At Testimonial-source =
170
testimonial, if testimonial is realistic and preview of firm is audio-visual, estimated
marginal means for perceived quality of website is highest. Perceived quality is lower
than the above case in both the following situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
Profile Plot 50. Dependent Variable- Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 61 demonstrates the marginal means of perceived credibility, given
testimonial-type and preview-mode as treatment variables when testimonial source is
company-dependent.
Figure 61: Profile Plot 50: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
The profile plot shows that for company-dependent source of employee testimonial, if
testimonial is realistic and preview of firm is audio-visual, estimated marginal means for
3.72
3.20 3.30
2.00
Testimonial-type
Estimated Marginal Means of Perceived credibility
At Testimonial-source =
171
perceived credibility of website is highest. Perceived credibility is lower than the above
case in both the following situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the testimonial
is positive, preview is text based and testimonial-source is company-dependent.
Profile Plot 51. Dependent Variable- Perceived Credibility, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 62: Profile Plot 51: DV- Perceived Credibility; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-neutral)
4.50
3.50
3.90
2.70
Testimonial-type
Estimated Marginal Means of Perceived credibility
At Testimonial-source = Company Neutral
172
Figure 62 demonstrates the marginal means of perceived credibility, given
testimonial-type and preview-mode as treatment variables when source of
advertisement is company neutral. The profile plot shows that for company-neutral source
of employee testimonial, if testimonial is realistic and preview of firm is audio-visual,
estimated marginal means for perceived credibility of website is highest. Perceived
credibility is lower than the above case in both the following situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the testimonial
is positive, preview is text based and testimonial-source is company-independent.
Profile Plot 52. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-
dependent)
Figure 63 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables when source of ad is
company dependent. The profile plot shows that for company-dependent source of
employee testimonial, if testimonial is realistic and preview of firm is audio-visual,
estimated marginal means for perceived organizational attractiveness of website is
highest. Perceived attractiveness is lower than the above case in both the following
situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
173
However, the lowest marginal mean was recorded for the scenario where the testimonial
is positive, preview is text based and testimonial-source is company-dependent.
Figure 63: Profile Plot 52: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-dependent)
Profile Plot 53. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 64 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables when source of ad is
company neutral. The profile plot shows that for company-independent source of
employee testimonial, if testimonial is realistic and preview of firm is audio-visual,
estimated marginal means for perceived organizational attractiveness of website is
highest. Perceived attractiveness is lower than the above case in both the following
situations:
3.50
3.00
3.27
2.25
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
At Testimonial-source
174
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the testimonial
is positive, preview is text based and testimonial-source is company-independent.
Figure 64: Profile Plot 53: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-neutral)
Profile Plot 54. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-dependent)
Figure 65 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial source is company
dependent. The profile plot shows that for company-dependent source of employee
testimonial, if testimonial is realistic and preview of firm is audio-visual, estimated
marginal means for intention to apply is highest. Intention to apply is lower among
job-seekers than the above case in both the following situations:
4.50
3.53
3.62
3.00
Testimonial-type
Estimated Marginal Means of Organizational Attractiveness
At Testimonial-source
175
a. Realistic employee testimonial and text preview.
b. Positive employee testimonial and audio-visual preview
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive, preview is text based and testimonial-source is company-
dependent.
Figure 65: Profile Plot 54: Dependent Variable-Intention to Apply, Treatment
variables- Preview-mode, Testimonial-type at Testimonial-source (Company-
dependent)
Profile Plot 55. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 66 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial source is company
independent. The profile plot shows that for company-neutral source of employee
testimonial, if testimonial is realistic and preview of firm is audio-visual, estimated
3.50
3.20 3.18
2.25
Testimonial-type
Estimated Marginal Means of intention to apply
At Testimonial-source =
176
marginal means for intention to apply is highest. Intention to apply is lower among
job-seekers than the above case in both the following situations:
a. Positive employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive, preview is text based and testimonial-source is company-
independent.
Figure 66: Profile Plot 55: Dependent Variable-Intention to Apply, Treatment
Variables- Preview-mode and Testimonial-type at Testimonial-source
(Company-neutral)
4.60
3.55
3.90
3.22
Testimonial-type
Estimated Marginal Means of intention to apply
At Testimonial-source
177
5.2.4 Phase II result (Study 2)
The experimental design for phase 2 can be represented as under:
Organizational Attractiveness + Intention to apply= β0 + β1 (Preview-mode) + β2
(Testimonial-type) + β3 (Testimonial-source) + β4 (Preview-mode x Testimonial-
type) + β5 (Testimonial-type x Testimonial-source) + β6 (Preview-mode x
Testimonial-source) + β7 (Preview-mode x Testimonial-type x Testimonial-source) +
β8 (Perceived Quality) + β9 (Perceived Quality) + Error
A multivariate analysis of covariance (MANCOVA) was conducted using SPSS 16
with the dependent variable block including the constructs of organizational
attractiveness and intention to apply, perceived credibility, perceived quality as
covariates and preview-mode (two factors-text/ audio-visual), testimonial-type (two
factors-positive/ realistic) and testimonial-source (two factors- company dependent/
company neutral) as fixed factors. The Levene‘s test for homogeneity of variance was
found to be insignificant (p > .05) which helped in meeting the necessary assumptions
to conduct a MANCOVA (refer Table 25).
Table 25: Levene's Test of Equality of Error Variances for MANCOVA
[Study 2]
Name of Variable F df1 df2 Significance
Organizational Attractiveness 1.663 7 353 .117
Intention to apply 2.376 7 353 .072
The results from the MANCOVA analysis indicate that the multivariate test of
significance (Wilk‘s Lambda) is significant at 95 % confidence level and preview-
mode, testimonial-type and testimonial-source, as well as the two-way and three-way
interaction terms all have significant F values. The multivariate tests indicate presence
178
of significant main and interaction effect of the treatment variables (preview-mode,
testimonial-type and testimonial-source) on the dependent variables.
TABLE 26: Multivariate Test of Significance for Organizational Attractiveness
and Intention to Apply [Study 2]
Name of Variable Test of
Significance
Value F Value Sig. Partial Eta
Squared
Observed
Power
Perceived Quality Wilk‘s
Lambda
.937 11.733 .000 .263 .994
Perceived Credibility Wilk‘s
Lambda
.575 1.2952 .000 .425 1.000
Preview-mode Wilk‘s
Lambda
.979 3.719 .025 .051 .880
Testimonial-type Wilk‘s
Lambda
.999 .169 .045 .041 .876
Testimonial-source Wilk‘s
Lambda
.990 3.830 .022 .037 .881
Preview-mode X Testimonial-type Wilk‘s
Lambda
.997 .561 .041 .033 .843
Preview-mode X Testimonial-
source
Wilk‘s
Lambda
.997 .524 .023 .039 .836
Testimonial-type X Testimonial-
source
Wilk‘s
Lambda
.956 8.130 .000 .044 .858
Preview-mode X Testimonial-type
X Testimonial-source
Wilk‘s
Lambda
.990 4.187 .005 .145 .977
The observed power for the above variables (greater than 0.8) also indicates strong
possibility of main and interaction effect (Hair et al. 1998). Table 27 shows the effect
of treatment variables on each dependent variable separately. The univariate results
indicate that perceived website quality has a significant main effect on organizational
attractiveness [F (2, 350) = 12.634, p = .001, effect size= .065] and intention to apply
[F (2, 350) = 23.523, p = .001, effect size = .063] for the jobseekers.
179
Perceived credibility has a highly significant main effect on organizational
attractiveness [F (2, 350) = 246.455, p = .001, effect size = .413] and intention to
apply [F (2, 350) = 190.179, p = .001, effect size = .352]. It is also observed that
preview-mode exerts a notable main effect on organizational attractiveness [F (2, 350)
= 3.290, p = .005, effect size = .079] but has no significant main effect on intention to
apply (p = .097). Testimonial-type has no main effect only on intention to apply (p =
.112), and organizational attractiveness (p = .074). Testimonial-source is noticed to
have significant main effect on organizational attractiveness [F (2, 350) = 3.941, p =
.008, effect size = .073] but not on intention to apply (p = .068).
The two-way interaction effect of preview-mode and testimonial-type is neither
significant for organizational attractiveness (p = .412) nor for intention to apply (p =
.903). Two-way interaction between preview-mode and testimonial-source is also not
significant on organizational attractiveness (p = .398) and intention to apply (p =
.837). As for the two-way interaction between testimonial-type and testimonial-
source, significant effect is observed for organizational attractiveness [F (2, 350) =
3.419, p = .005, effect size = .081] and intention to apply [F (2, 350) = 3.369, p =
.004, effect size = .096]. The three-way interaction between preview-mode,
testimonial-type and testimonial-source is also observed to be significant for
organizational attractiveness [F (2, 350) = 7.321, p = .002, effect size = .122] and
intention to apply [F (2, 350) = 10.251, p = .001, effect size = .146) respectively. The
effect sizes (partial eta squared) for all significant effects are in the range .06-.5 which
indicates small to moderate effect of the treatments on the dependent variables (Cohen
1988; Kotrlik & Williams 2003).
180
TABLE 27: Univariate Test of Significance for Organizational Attractiveness
and Intention to Apply [Study 2]
Main & Interaction Effects F Value Partial
Eta
Squared
Observe
d power
Perceived Quality Attractiveness Perceived Quality Intention to apply
12.634* 23.523*
.065
.063 .943 .998
Perceived Credibility Attractiveness
Perceived Credibility Intention to apply
246.455*
190.179*
.413
.352
1.000
1.000
Preview-mode Attractiveness
Preview-mode Intention to apply
3.290**
.014
.079
.000
.840
.052
Testimonial-type Attractiveness
Testimonial-type Intention to apply
.160
.338
.000
.001
.068
.089
Testimonial-source Attractiveness Testimonial-source Intention to apply
3.941** .406
.073
.001 .802 .097
Preview-mode X Testimonial-type Attractiveness
Preview-mode X Testimonial-type Intention to apply
.674
.015
.002
.000
.130
.052
Preview-mode X Testimonial-source Attractiveness
Preview-mode X Testimonial-source Intention to apply
.717
.042
.002
.000
.135
.055
Testimonial-type X Testimonial-source Attractiveness
Testimonial-type X Testimonial-source Intention to apply
3.419*
3.369*
.081
.096
.954
.911
Preview-mode X Testimonial-type X Testimonial-source Attractiveness
Preview-mode X Testimonial-type X Testimonial-source Intention to
apply
7.321**
10.251*
.122
.146
.917
.977
*p<.001, **p<.01
The findings suggest that perceptions regarding attractiveness of an organization as a
potential employer as well as job aspirant‘s intention to apply for any vacancy in the
firm are perceived differently depending on the main effects of perceived quality,
perceived credibility, preview-mode and testimonial-source. Testimonial-type
accounts for only the difference in perception towards intention to apply. Two-way
interactions are significant for combination of treatments preview-mode and
testimonial-type, and for combinations of testimonial-type and testimonial-source.
The three-way interaction is also significant on job-seekers‘ perception regarding
181
organizational attraction and intention to apply. However, a combination of preview-
mode and testimonial-source is not found to be significant for influencing intention to
apply. This combination is found to trigger only difference in perception towards
attractiveness of a firm. This indicates that websites with different combinations of
preview-mode, testimonial-type and testimonial-source have a possibility of being
differently perceived by job-seekers while seeking job-related information through
internet compared to websites with either a single treatment or a combination of any
two treatments.
5.2.4.1 Profile Plot Analysis
Profile Plot 56. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Preview-mode
Figure 67 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-type as treatment variables. The profile plot shows that
for video-based preview and positive employee testimonial, perceived attractiveness
of website is highest.
Perceived attractiveness was lower than the above case in both the following
situations:
a. Video based preview and realistic testimonial
b. Text based preview and realistic testimonial
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text based and testimonial-type is positive.
182
Figure 67. Profile Plot 56: DV- Organizational Attractiveness; IV-Testimonial-type,
Previewmode
Profile Plot 57. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-source
Figure 68. Profile Plot 57: DV- Organizational Attractiveness; IV-Preview-mode,
Testimonial-source
Figure 68 demonstrates the marginal means of organizational attractiveness, given
preview-mode and testimonial-source as treatment variables. The profile plot shows
3.48
3.46
3.38
3.32
3.56
3.45
3.43
3.27
Estimated Marginal Means of Organizational Attractiveness
Estimated Marginal Means of Organizational Attractiveness
Testimonial-type
Testimonial-source
183
that for video-based preview and company-neutral source of employee testimonial,
estimated marginal means for perceived attractiveness is highest. Perceived
attractiveness is lower than the above case in both the following situations:
a. Text based preview and company neutral testimonial-source
b. Video based preview and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the preview-
mode is text-based but testimonial-source is company-dependent.
Profile Plot 58. Dependent Variable- Organizational Attractiveness, Treatment
Variables- Testimonial-type, Testimonial-source
Figure 69. Profile Plot 58: DV- Organizational Attractiveness; IV-Testimonial-type,
Testimonial-source
Figure 69 demonstrates the marginal means of organizational attractiveness, given
testimonial-type and preview-mode as treatment variables. The profile plot shows that
for realistic employee testimonial and company-independent source of employee
testimonial, estimated marginal means for perceived attractiveness of website is
3.53
3.42
3.38
3.32
Estimated Marginal Means of Organizational Attractiveness
Testimonial-type
Testimonial-source
184
highest. Perceived attractiveness is lower than the above case in both the following
situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Positive employee testimonial and company dependent testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-dependent.
Profile Plot 59. Dependent Variable-Intention to Apply, Treatment Variables-
Testimonial-type, Preview-mode
Figure 70 demonstrates the marginal means of intention to apply, given
testimonial-type and preview-mode as treatment variables.
Figure 70. Profile Plot 59: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
The profile plot shows that for realistic employee testimonial and text preview,
estimated marginal means for intention to apply is highest. Intention to apply is lower
than the above case in the following situation where there is realistic employee
3.495
3.479
3.447 3.445
Estimated Marginal Means of intention to apply
Testimonial-type
185
testimonial along with audio-visual preview. The scenarios which registered the
lowest intention to apply were:
a. Realistic employee testimonial and text preview.
b. Positive testimonial and audio-visual preview.
Profile Plot 60. Dependent Variable- Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 71 demonstrates the marginal means of Intention to apply, given testimonial-
type and preview-mode as treatment variables.
Figure 71. Profile Plot 60: DV- Intention to Apply; IV-Preview-mode, Testimonial-
source
The profile plot shows that for text based preview and company-neutral source of
employee testimonial, estimated marginal means for intention to apply is highest.
Intention to apply is lesser than the above case in the situation where there was audio-
visual corporate preview and company neutral testimonial-source. Intention to apply
was lowest for the following two situations:
a. Audio-visual preview and company dependent testimonial-source.
3.53
3.48
3.43 3.44
Estimated Marginal Means of intention to apply
Testimonial-source
186
b. Text based preview and company-dependent testimonial-source.
Profile Plot 61. Dependent Variable- Intention to apply, Treatment Variables-
Testimonial-type, Testimonial-source
Figure 72 demonstrates the marginal means of intention to apply, given testimonial-
type and testimonial-source as treatment variables. The profile plot shows that for
positive employee testimonial and company-independent source of employee
testimonial, estimated marginal means for intention to apply is highest.
Figure 72. Profile Plot 61: DV- Intention to Apply; IV-Testimonial-type,
Testimonial-source
Intention to apply is lesser than the above case in both the following situations:
a. Realistic employee testimonial and company dependent testimonial-source
b. Realistic employee testimonial and company neutral testimonial-source.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positively designed but testimonial-source is company-dependent.
3.51 3.49
3.47
3.385
Estimated Marginal Means of intention to apply
Testimonial-type
Testimonial-source
187
Profile Plot 62. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-
dependent)
Figure 73 demonstrates the marginal means of organizational attractiveness, given
testimonial-type, preview-mode and testimonial-source (company-dependent) as
treatment variables.
Figure 73. Profile Plot 62: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-dependent)
The profile plot shows that, when positive employee testimonials are presented via
company-dependent sources, along with audio-visual corporate previews, estimated
marginal mean for organizational attractiveness is highest. In case of realistic
testimonials presented through company dependent channels if the preview-mode is
video based, then attractiveness is higher than in the cases where testimonial is either
positive or realistic and preview is text based.
Estimated Marginal Means of Organizational Attractiveness
Testimonial-type At Testimonial-source
3.52
3.35
3.28
3.26
188
Profile Plot 63. Dependent Variable-Organizational Attractiveness, Treatment
Variables- Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 74 demonstrates the marginal means of organizational attractiveness, given
testimonial-type, preview-mode and testimonial-source (company-neutral) as
treatment variables.
Figure 74. Profile Plot 63: DV- Organizational Attractiveness; IV- Preview-
mode, Testimonial-type at Testimonial-source (Company-neutral)
The profile plot shows that, when realistic employee testimonials are presented via
company-neutral sources, along with audio-visual corporate previews, estimated
marginal mean for organizational attractiveness is highest. In case of realistic
testimonials presented through company dependent channels with text based preview-
mode attractiveness perception is second highest, followed by the scenario where the
testimonial is positive but the preview is audio-visual. It is lowest for the scenario
3.57
3.52
3.47
3.37
Estimated Marginal Means of Organizational Attractiveness
Testimonial-type At Testimonial-source
189
wherein the preview is text based and testimonial is positively framed through a
company-neutral channel.
Profile Plot 64. Dependent Variable-Intention to apply, Treatment Variables-
Preview-mode, Testimonial-Type, Testimonial-source (Company-dependent)
Figure 75 demonstrates the marginal means of intention to apply, given testimonial-
type, preview-mode and testimonial-source (company-dependent) as treatment
variables.
Figure 75. Profile Plot 64: DV- Intention to Apply; IV- Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
The profile plot shows that, when realistic employee testimonials are presented via
company-dependent sources, along with textual corporate previews, estimated
marginal mean for intention to apply is highest. In case of realistic testimonials
presented along with video based preview-mode and in case of positive testimonials
with audio-visual previews, there is not much significant difference in intention to
apply. It is further observed that, if positive testimonials are to be propagated through
3.43
3.42
3.54
3.33
Estimated Marginal Means of intention to apply
Testimonial-type
At Testimonial-source
190
company-dependent channels, in such case corporate previews should not be
presented in text based mode, since it significantly lowers job-seekers‘ intention to
apply.
Profile Plot 65. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type, Testimonial-source (Company-neutral)
Figure 76 demonstrates the marginal means of intention to apply, given testimonial-
type, preview-mode and testimonial-source (company-neutral) as treatment variables.
Figure 76. Profile Plot 65: DV- Intention to Apply; IV- Preview-mode,
Testimonial-type, at Testimonial-source (Company-neutral)
The profile plot shows that, when positive employee testimonials are presented via
company-neutral sources, along with text based corporate previews, estimated
marginal mean for intention to apply is highest. Next best scenario for a high intention
to apply is when realistic testimonials are presented along with audio-visual preview-
mode.
3.571
3.517
3.445 3.458
Estimated Marginal Means of intention to apply
Testimonial-type At Testimonial-source = Company Neutral
191
5.2.5 Phase III Result (Study 2)
The experimental design for phase 3 can be represented as under:
Intention to apply= β0 + β1 (Preview-mode) + β2 (Testimonial-type) + β3
(Testimonial-source) + β4 (Preview-mode x Testimonial-type) + β5 (Testimonial-type
x Testimonial-source) + β6 (Preview-mode x Testimonial-source) + β7 (Preview-
mode x Testimonial-type x Testimonial-source) + β8 (Perceived Quality) + β9
(Perceived Quality) + β10 (Organizational Attractiveness) + Error
Analysis of covariance (ANCOVA) was conducted using SPSS 16 with the dependent
variable intention to apply, perceived credibility, perceived quality as covariates and
preview-mode (two factors-text/ audio-visual), testimonial-type (two factors-positive/
realistic) and testimonial-source (two factors- company dependent/ company neutral)
as fixed factors. The results revealed that perceived quality [F (1, 350) = 10.419, p =
.001, effect size = .059], perceived credibility [F (1, 350 = 7.779, p = .006, effect size
= .062] and organizational attractiveness [F (1, 350) = 372.766, p = .001, effect size=
.516] are all significant predictors of purchase intention. Among the treatments, only
preview-mode shows significant main effect [F (1, 350) = 4.119, p = .043, effect size
= .032]. The two way interaction between preview-mode and type of testimonial [F
(1, 350) = .449, p = .503] and that between preview-mode and source of testimonial-
type were found to be insignificant [F (1, 350) = .332, p = .565].
Only the two-way interaction between testimonial-type and testimonial-source were
found to be significant [F (1, 350) = 12.726, p = .001, effect size = .095]. The three
way interaction between preview-mode, testimonial-type and testimonial-source was
found to be significant at 95 % confidence interval [F (1, 350) = 5.493, p = .042,
effect size = .11].
192
TABLE 28: Test of Significance for Dependent Variable Intention to Apply
[Study 2]
Predictor variable F Value Partial Eta Squared
Observed Power
Perceived Quality 10.491* .059 .898
Perceived Credibility 7.779* .062 .894
Organizational Attraction 372.766* .516 1.000
Preview-mode 4.119** .032 .826
Testimonial-type .178 .001 .071
Testimonial-source .722 .002 .135
Preview-mode X Testimonial-type .449 .001 .103
Preview-mode X Testimonial-source .332 .001 .089
Testimonial-type X Testimonial-source 12.726* .095 .945
Preview-mode X Testimonial-type X
Testimonial-source
15.493** .110 .832
*p<.01, **p<.05
193
5.2.5.1 Profile Plot Analysis
Profile Plot 66. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-type
Figure 77 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables. The profile plot shows that for positive
employee testimonial and text preview, estimated marginal means for intention to
apply is highest, closely followed by the situation where preview is text based but
testimonial is realistic. The lowest marginal mean was recorded for the scenario
where the testimonial is positive and preview is audio-visual.
Figure 77. Profile Plot 66: DV- Intention to Apply; IV-Testimonial-type,
Preview-mode
3.535
3.517
3.438
3.35
Testimonial-type
Estimated Marginal Means of intention to apply
194
Profile Plot 67. Dependent Variable-Intention to Apply, Treatment Variables-
Preview-mode, Testimonial-source
Figure 78 demonstrates the marginal means of Intention to apply, given testimonial-
source and preview-mode as treatment variables. The profile plot shows that for text
preview and company-dependent source of employee testimonial, estimated marginal
means for intention to apply is highest. Intention to apply is lower than the above case
in the following situations:
a. Text preview and company neutral testimonial-source
b. Video preview and company dependent testimonial-source.
c. Video preview and company neutral source
Figure 78. Profile Plot 67: DV- Intention to Apply; IV-Testimonial-source,
Preview-mode
Profile Plot 68: Dependent Variable- Intention to Apply; Treatment Variables -
Testimonial-type, Testimonial-source
Figure 79 demonstrates the marginal means of intention to apply, given testimonial-
type and testimonial-source as treatment variables. The profile plot shows that for
realistic employee testimonial and company-dependent source of employee
3.56
3.42 3.47
3.40
Testimonial-source Estimated Marginal Means of intention to apply
195
testimonial, estimated marginal means for intention to apply is highest. Intention to
apply is lower than the above case in both the following situations:
a. Positive employee testimonial and company neutral testimonial-source
b. Positive employee testimonial and company dependent testimonial-source
However, the lowest marginal mean was recorded for the scenario where the
testimonial is realistic but testimonial-source is company-independent.
Figure 79. Profile Plot 68: DV- Intention to Apply; IV-Testimonial-type,
Testimonial-source
Profile Plot 69: Dependent Variable- Intention to Apply; Treatment Variables-
Testimonial-type, Preview-mode at Testimonial-source (Company-dependent)
Figure 80 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial source is company-
dependent. The profile plot shows that for realistic employee testimonial and text
preview presented via company-dependent source, estimated marginal means for
intention to apply is highest.
3.58
3.51
3.41 3.363
Testimonial-type
Testimonial-source
Estimated Marginal Means of intention to apply
196
Figure 80. Profile Plot 69: DV- Intention to Apply; IV-Preview-mode,
Testimonial-type at Testimonial-source (Company-dependent)
Intention to apply is lower than the above case in both the following situations:
a. Realistic employee testimonial and audio-visual preview
b. Realistic employee testimonial and text preview.
However, the lowest marginal mean was recorded for the scenario where the
testimonial is positive but preview is audio-visual given testimonial-source is
company-dependent.
Profile Plot 70: Dependent Variable- Intention to apply; Treatment Variables -
Testimonial-type, Preview-mode at Testimonial-source (Company-neutral)
Figure 81 demonstrates the marginal means of intention to apply, given testimonial-
type and preview-mode as treatment variables when testimonial source is company-
neutral. The profile plot shows that for realistic employee testimonial and text
preview, estimated marginal means for intention to apply is highest.
3.67
3.49
3.45 3.37
Testimonial-type
Estimated Marginal Means of intention to apply
At Testimonial-source =
197
Figure 81. Profile Plot 70: DV- Intention to apply; IV-Testimonial-type, Preview-
mode at Testimonial-source (Company-neutral)
Intention to apply is lower than the above case in both the following situations where
employee testimonial is realistic and preview is audio-visual and in the situation
where employee testimonial is positive and preview is audio-visual. However, the
lowest marginal mean was recorded for the scenario where the testimonial is positive
but preview is text based given testimonial-source is company-independent, though
the difference in mean is negligible.
3.61
3.41
3.38 3.36
Testimonial-type
Estimated Marginal Means of intention to apply
At Testimonial-source
198
5.2.6 Phase IV Results (Study 2)
A covariance based structural equation model (SEM) is estimated to assess the
hypothesized relationship. The fit of the structural model is acceptable, with χ2
(52) =
186.351 (p<.01), goodness of fit index (GFI) = .919, comparative fit index (CFI) =
.939, incremental fit index (IFI) = .923 and root mean square error of approximation
(RMSEA) = .046. CMIN/df is calculated to be 3.584. Based on these results, the
measurement model indicates an acceptable model fit of the data (Hair et al., 1998).
Consistent with the different cut-off criteria provided in the literature (Joreskog, 1993;
Hu and Bentler 1999), it is concluded that the hypothesized causal model is within the
acceptable range of all the fit statistics.
Please refer to Table 29 for structural path model‘s estimates and a summary of the
results of the hypothesis tests. In H10a it is proposed that perceived quality positively
affects job-seekers‘ perception of organizational attraction (β1a = .174, p = .004) and
in H10b it is proposed that perceived quality positively affects their intention to apply
(β2a = .101, p = .001). Both these hypotheses receive support at 99 % level of
confidence. Our next two hypotheses (H11a and H11b) are also supported because
perceived credibility is found to have a positive effect on organizational attractiveness
(β2a = .609, p = .001) and intention to apply (β2b = .154, p = .003). Out of the
hypotheses H4a, H4b and H4c, H4b do not get any support but H4a and H4c receive
support as preview-mode (β3 = .094, p = .021) and testimonial-source (β5 = .114, p =
.002) are found to be significant. As seen in study I, organizational attractiveness
once again significantly influences intention to apply (β6 = .750, p = .001), providing
support for H14. In order to evaluate whether the impact of perceived quality on job-
seekers‘ intention to apply is fully mediated by organizational attractiveness, a path
199
from perceived quality to intention to apply is created and the chi-square difference
between the two models is evaluated.
Table 29: Moderating Effect of Preview-Mode, Testimonial-Type and
Testimonial-Source and Mediating Effect of Organizational Attractiveness
[Study 2]
Hypothesized Path Standardized path coefficients
Hypothesis β S.E. C.R.
Perceived Quality → Attractiveness
H10a .174* .060 2.918
Perceived Credibility→ Attractiveness
H11a .609* .053 10.908
Preview-mode →Attractiveness
H4a .094** .059 2.303
Testimonial-type → Attractiveness
H4b .022 .057 .571
Testimonial-source → Attractiveness
H4c .114** .057 3.071
Perceived Quality→ Intention to Apply
H10b .101** .030 3.329
Perceived Credibility → Intention to apply
H11b .154* .039 2.931
Attractiveness → Intention to Apply H14 .750* .042 18.947
Perceived Quality*Preview-modeAttractiveness H19a -.302* .041 -2.447
Perceived Quality*Testimonial-typeAttractiveness H19b -.016 .041 -.385
Perceived Quality*Testimonial-sourceAttractiveness H19c .037 .055 .760
Perceived Credibility*Preview-modeAttractiveness H20a -.043 .045 -1.087
Perceived Credibility*Testimonial-typeAttractiveness H20b .101** .043 2.193
Perceived Credibility*Testimonial-sourceAttractiveness H20c .095** .042 1.160
Fit Statistics
χ2 (df) 186.351 (52)
p .000
GFI .919
CFI .939
IFI .923
RMSEA .046
**p<.05; *p<.01 Notes: β represents standardized path coefficient
200
Figure 82: Structural Model with Path Coefficients [Study 2]
It is observed that the independent variable (perceived quality) has significant
relationship (βc=.101, p=.001, S.E. = .03) with the dependent variable intention to
apply and the mediating variable organizational attractiveness (βa=.174, p=.001, S.E.a
= .060); and the mediating variable also significantly predicts the dependent variable
(βb=.750, p=.001, S.E.b = .042). These results show that the hypothesized model
satisfy the basic conditions for mediation as proposed by Barron and Kenny (1982).
Afterwards, a Sobel‘s test (Sobel, 1982) was carried out to test the nature of
Perceived
Credibility
Organizational
Attractiveness
Intention to
Apply
Perceived
Quality
MOD 1
Preview-
mode
Testimonial-
type
Testimonial-
source
MOD 4 MOD 5 MOD 6
.174* .101**
. 750*
.154* .609*
.094**
.022
.114**
-.302*
-.016
.037
-.043
.111**
.095**
MOD 2
MOD 3
201
mediation. To calculate Sobel‘s test statistic, the same Sobel‘s statistics used in study
1 is used.
Sobel‘s test for significance = 2.86249824 (>1.96), with one-tailed probability =
.00210158, two-tailed probability = .00420316. The test statistic was significant at 99
% level of significance, indicating partial mediating effect (Preacher and Hayes 2003)
of organizational attractiveness on the relationship between perceived quality and
intention to apply. Thus we received partial support for our hypothesis 21. Next, in
order to evaluate whether the impact of perceived credibility on job-seekers‘ intention
to apply is fully mediated by organizational attractiveness, a path from perceived
credibility to intention to apply is created and the chi-square difference between the
two models is evaluated. It is observed that the independent variable (perceived
credibility) has a significant relationship (βc = .154, p = .001, S.E.c = .039) with the
dependent variable (intention to apply) and its relation (βa = .609, p = .001, S.E.a =
.053) with the mediating variable (organizational attractiveness) is also significant.
The mediating variable also significantly predicts the dependent variable (βb = .750, p
= .001, S.E.b = .042). These results show that the hypothesized model satisfy the basic
conditions for mediation as proposed by Barron and Kenny (1982). Another Sobel test
was conducted to test the nature of mediation and it was found that Sobel‘s test for
significance = 9.66291469 (>1.96), with one-tailed probability= .00, two-tailed
probability= .00. The test statistic was significant at 99 % level of significance,
indicating partial mediating effect (Preacher and Hayes 2003) of organizational
attractiveness on the relationship between perceived quality and intention to apply.
Thus we received partial support for our hypothesis 22.
In order to test for moderating effect of preview-mode, testimonial-type and
testimonial-source on the relationship between perceived quality and organizational
202
attractiveness, interaction terms are created for perceived quality with respect to
preview-mode, testimonial-type and testimonial-source. As can be observed,
perceived quality has significant main effect on organizational attractiveness (p =
.001). Interaction effect of perceived quality and preview-mode is found to be
significant on organizational attractiveness (p = .004), but interaction effect of
perceived quality with testimonial-type (p = .700) and testimonial-source (p = .448) is
insignificant. Hence H19a is supported, but H19b and H19c received no support. As for
moderating impact of preview-mode, testimonial-type and testimonial-source on the
relationship between perceived credibility and firm attractiveness, interaction terms
are created for perceived credibility with respect to preview-mode, testimonial-type
and testimonial-source. Perceived credibility has main effect on organizational
attractiveness (p = .001). Preview-mode (p = .277) does not demonstrate any
significant moderating effect on the relation between perceived credibility and firm
attractiveness. However, testimonial-type (p = .028) and testimonial-source (p = .046)
are observed to have significant moderating impact on the credibility-firm
attractiveness relationship. Hence H20a is not supported but H20b and H20c receive
complete support.
203
5.2.7 Phase V Results (Study 2): Post-Hoc Comparisons
Once again, post-hoc tests were conducted using Scheffe‘s test (Klockars and
Hancock 2000). The findings gave indications that in terms of intention to apply,
perceptions are different for those jobseekers in Group A who have viewed the
website with treatment condition 1 compared to Groups C, D, G, and H which
correspond to treatment conditions 3, 4, 7 and 8 (MDs significant at p < .01).
However, perceptions are not significantly different when job-seekers under Groups
A, E and F are compared with those in Group A (p > .05). Perceptions are different
for those jobseekers in Group B who have viewed the website with treatment
condition 2 compared to Groups C, D, G and H which correspond to treatment
conditions 3, 4, 7 and 8 (MDs significant at p < .01) respectively. However,
perceptions are not significantly different when participants in groups A, E and F are
compared with those in Group B (p > .05). With respect to Group C (treatment 3),
Groups A, B, E, F and H contributed to different perception formation for intention to
apply at 99 % confidence level but Groups D and G are not able to create any
difference in jobseekers‘ intention to apply (p > .05).
As for those exposed to website corresponding to treatment 4 in Group D, perception
of applying for job is different compared to those who viewed treatments 1, 2, 5, 6
and 8 in Groups A, B, E, F and H (MDs significant at p < .01). Intention to apply for
Group E and Group F job-seekers is markedly different from that for job-seekers in
Group C, D, G and Group H at 99 % confidence level. Group G job-seekers have
different intention to apply compared to those in Groups A, B, E, F and H. Finally,
Group H is found to create difference in intention to apply at 95 % confidence level
for all other groups as well as having higher means compared to the other groups for
intention to apply,
204
Table 30: Post-Hoc Analysis Using Scheffe’s Procedure [Study 2]
Referent
group (I)
Comparison
group (J)
Mean Difference (I-J) Std. Error Sig.
Group A Group B -.01141 .15073 1.000
Group C -1.06409* .14858 .000
Group D -1.20319* .15585 .000
Group E -.18084 .15230 .985
Group F -.29129 .15230 .817
Group G -1.52986* .16246 .000
Group H -2.15839* .14490 .000
GroupB Group A .01141 .15073 1.000
Group C -1.05267* .14422 .000
Group D -1.19177* .15170 .000
Group E -.16943 .14805 .988
Group F -.27987 .14805 .827
Group G -1.51844* .15849 .000
Group H -2.14698* .14043 .000
Group C Group A 1.06409* .14858 .000
Group B 1.05267* .14422 .000
Group D -.13910 .14956 .997
Group E .88324* .14586 .000
Group F .77280* .14586 .000
Group G -.46577 .15645 .266
205
Group H -1.09431* .13812 .000
Group D Group A 1.20319* .15585 .000
Group B 1.19177* .15170 .000
Group C .13910 .14956 .997
Group E 1.02235* .15326 .000
Group F .91190* .15326 .000
Group G -.32667 .16336 .779
Group H -.95520* .14591 .000
Group E Group A .18084 .15230 .985
Group B .16943 .14805 .988
Group C -.88324* .14586 .000
Group D -1.02235* .15326 .000
Group F -.11044 .14965 .999
Group G -1.34902* .15998 .000
Group H -1.97755* .14211 .000
Group F Group A .29129 .15230 .817
Group B .27987 .14805 .827
Group C -.77280* .14586 .000
Group D -.91190* .15326 .000
Group E .11044 .14965 .999
Group G -1.23857* .15998 .000
Group H -1.86711* .14211 .000
Group G Group A 1.52986* .16246 .000
206
Group B 1.51844* .15849 .000
Group C .46577 .15645 .266
Group D .32667 .16336 .779
Group E 1.34902* .15998 .000
Group F 1.23857* .15998 .000
Group H -.62854** .15296 .020
Group H Group A 2.15839* .14490 .000
Group B 2.14698* .14043 .000
Group C 1.09431* .13812 .000
Group D .95520* .14591 .000
Group E 1.97755* .14211 .000
Group F 1.86711* .14211 .000
Group G .62854** .15296 .020
*p< .01, **p<.05
This suggests that the final treatment condition has the most desirable effect on job-
seekers‘ intention to apply. Only Group G can be considered to be the next best
website design with the mean difference with that of Group H being the lowest (MD =
.62854, p < .05).