hungarian validation of zimbardo time perspective inventory filehungarian validation of zimbardo...
TRANSCRIPT
Hungarian Validation of Zimbardo Time
Perspective Inventory
Gábor OROSZ University of Szeged,
Institute of Psychology 07/10/2011
First of all, the question arises:
„Why is time perspective an interesting topic?”
http://www.youtube.com/watch?v=A3oIiH7BLmg
Validity and reliability of translated questionnaires
• TP is an interesting topic – but how to measure it? • Criteria of questionnaire validation:
– Internal consistency – Cronbach’s alphas and test-retest reliability (several days –> years)
– Construct validity – firstly: Exploratory Factor Analysis (EFA); later: Confirmatory factor analysis (CFA) (same vs. different samples)
– Convergent validity: correlation with other variables
– Discriminant validity: Sensation seeking vs. Present hedonism
Criteria of validity and reliability
• EFA criteria: minimal loading .32 and no cross-loading above .32 (Tabachnik & Fidell, 2001)
• Cronbach’s alpha at least .7 but .8 is better (Nunnally, 1978) (however, on the basis on my experiences .9 or above alpha measures a very narrow psychological construct)
• CFA criteria: RMSEA ≤ .06, CFI ≥ .95, TLI ≥ .95 Hu & Bentler (1999)
• Inter-item correlation: between .15 and .50 (Clark & Watson, 1995)
Some trade-offs of validation 1. Cronbach’s alphas ↑ with the number of
items and ↓ with the number of factors 2. Generally, EFA factor structure becomes
clearer by dropping items 3. CFA is very sensitive to cross-loadings*
and lower factor loadings (↓ than .5) → only a few items meet this criteria
4. CFA is sensitive to the high N of factors → higher possibility of cross-loadings*
* Inappropriate covariances
The original questionnaire: ZTPI (Zimbardo & Boyd, 1999, JPSP)
Sample: NEFA = 606, NCFA = 361, Age = 19.9 Nitems = 56 Total explained variance = 36%
EFA & α PAST - PAST + PRES HEDON
PRES FATAL
FUTURE
N° OF ITEMS
9 10 15 9 13
EXP VAR. 4,5% 12,3% 8,9% 3,9% 6,3%
Cr α .80 .82 .79 .74 .77
CFA: χ2/df = 2.3 RMSEA = ?, CFI = ?, TLI = ?
ZTPI validations in different cultures I.
Nation Sample Exp. var.
EFA and reliability
Past pos.
Past neg
Pres hed
Pres fat
Fut. χ2/df RMSEA
CFI TLI Authors
USA N = 606 Age = 19.9
N = 56 36%
N of items: Exp. var.: Alpha:
9 4.5% .80
10 12.3% .82
15 8.9% .79
9 3.9% .74
13 6.3% .77
2.3 ? ? ? Zimbardo & Boyd (1999)
Italy N = 1507 Age = 34.7
N = 21 31%
N of items:
Exp. Var.: Alpha:
― ― 8
8.4%
.54
5 7.3%
.49
9 15.1%
.67 ? ? ? ? D’Alessio et al.
(2003)
France N = 419 Age = 21.5
N = 20 33%
N of items: Exp. var.: Alpha:
8 4.4% .70
9 8.1% .72
18 10.5% .79
7 3.7% .70
12 6.1% .74
2.04 .055 ? ? Apostolidis & Fieulaine (2004)
Mexico N = 300 Age = 31.8
N = 56 ―
N of items: Exp. var.: Alpha:
20 ?
.77
25 ?
.80
11 ?
.75 ? ? ? ? Corral-Verdugo
et al. (2006)
ZTPI validations in different cultures II.
Nation Sample Exp. var.
EFA and reliability
Past pos.
Past neg
Pres hed
Pres fat
Fut χ2/df RMSEA
CFI TLI Authors
Spain N = 756 Age = 40,1
N = 56 34%
N of items: Exp. var.: Alpha:
8 4.4% .70
14 11.2% .80
14 7.7% .79
9 4.0% .64
11 6.5% .74
? ? ? ? Diaz-Morales (2006)
Brazil N = 247 Age = 22.5
N = 38 31%
N of items:
Exp. Var.: Alpha:
6 ?
.60
7 ?
.60
9 ?
.55
6 ?
.46
10 ?
.67 ? ? ? ? Milfont et al.
(2008)
Lith. N = 1244 Age = 30,9
N = 56 33%
N of items: Exp. var.: Alpha:
8 3.3%
.70
10 7.7%
.72
14 7.3%
.79
10 4.4%
.70
13 12%
.74 2.22 .044 .67 .65 Liniauskaite &
Kairys (2009)
Taiwan N = 420 Age = 20,5
N = 20
N of items: Exp. var.: Alpha:
4 ?
.68
4 ?
.76
4 ?
.68
4 ?
.49
4 ?
.68 2.01 .05 .93 - Gao (2011)
Hungarian validation of ZTPI I.
• Main goals: 1. Achieving an appropriate EFA factor
structure (e.i. the rule of .32) 2. Achieving appropriate internal consistency in
terms of higher alphas than .7 3. Achieving appropriate CFA (RMSEA, CFI, TLI)
• Not goals (yet) 1. Test-retest reliability 2. Convergent validity & discriminant validity
Hungarian validation of ZTPI II. • Translation:
– Five persons translated independently the original ZTPI to Hungarian (3 psychology MA students who knew the topic and have at least advanced language exams, and 2 English teachers who did not know the topic and who have MA in English)
– Then all of us discussed the inconsistencies until finding the best solution
– The final solution was given to a bilingual psychology student for backtranslation to Hungarian
– Finally, the seven persons discussed the final solution
Sample • 1364 persons
– 924 women and 405 men – Age between 14 and 86 (M = 32.19, SD = 14.70)
• Education: – 142 primary school – 53 vocational school – 399 high school degree – 712 higher education (BA, BSC, MA, MSC) – 51 postgraduate degree
• Place of residence: – 170 villages, 553 towns, 326 county towns, 309 capital
Measures • The translated version of ZTPI (56 items), 5-point Likert
scale (1 = very uncharacteristic; 5 = very characteristic) • Gender • Age • Marital status • Place of residence • Perceived financial status • Expected financial situation in 5-10 years • Level of education • Perceived health
Results 1: EFA & CFA • Principal Axis Factoring (PAF) extraction • Promax rotation (Kappa = 0) (Brown, 2006)
– Oblique solution instead of orthogonal: (1) it provides better results for the CFA, (2) TP subscales correlated in previous studies (Anagnostopoloulos & Griva, 2011)
– Respecting the rule of .32 of Tabachnik & Fidell (2001)
A 36 item solution emerged: Strong points • Five factors: scree test • Exp. Var: 45.9% • Bartlett and KMO: OK • Alphas are higher than .7
Weak points CFA results:
• RMSEA = .068 !!! (X < .06) • CFI = .75 !!! (X > .95) • TLI = .72 !!! (X > .95)
The final 17-item version
RMSEA = .039 CFI = .963 TLI = .954
36 items ↓
17 items
OK, but how the hell can I carry out a confirmatory factor analysis?
CFA – how to do it?
• The most important message of this presentation:
„USE YOUTUBE IF YOU DON’T KNOW HOW TO CARRY OUT A STATISTICAL ANALYSIS!”
http://www.youtube.com/watch?v=JkZGWUUjdLg
Potential problems… • It is not the best to use ML method if we don’t have the normal
distribution of our variables (generally it is the case…) • Modification indices can be very useful in the item selection, do
not overdose it (follow common sense as well) • Error covariances can be useful, however, we have to explain
why we put such error covariances (case of Vallerand) • It is very difficult to achieve a good model fit if we have 4-5
factors and 4-5 items per factor • This method pushes authors to create scales with few items
which can measure efficiently narrow psychological constructs – TP is a multidimensional phenomena and NOT a narrow
construct…
Are Competition and Extrinsic Motivation Reliable Predictors
of Academic Cheating?
Gábor OROSZ University of Szeged
Institute of Psychology 07/10/2011
Theoretical roots of this question
• “Competition is perhaps the single most toxic ingredient in a classroom, and it is also a reliable predictor of cheating” (Anderman & Murdock, 2007, p. XIII)
• Competition has an overall negative impact on performance, problem solving, and personal relationships in comparison with cooperation (Deutsch, 1949, Johnson & Johnson, 1974, 1979, 1982; Johnson, Maruyama, Johnson, Nelson, Skon, 1981; Lewis, 1944a, 1944b; Qin, Johnson & Johnson, 1995)
• Competition undermines intrinsic motivation (Deci, Betley, Kahle, Adrams & Porac, 1981; Vallerand, Gauvin, Hallivell, 1986a, 1986b)
Paradigm shift of competition • Solid theoretical basis regarding competition’s positive
consequences: – It can improve performance, interpersonal
relationships, resource control and intrinsic motivation (Bornstein, Erev & Rosen, 1990; Carnavale & Probst, 1997; Charlesworth, 1996; Epstein & Harackiewicz, 1992; Erev et al., 1993; Fülöp, 1997, 1999, 2001, 2004; Harackiewicz, 1998; Hawley, 2003, 2006; Hurlock, 1927; Moede, 1914; Reeve & Deci, 1999; Ryckman, Hammer, Kaczor & Gold 1996; Sims, 1927; Tauer & Harackiewicz, 2004; Tassi & Schneider, 1997; Tjosvold, Johnson, Johnson & Sun, 2003, 2006; Young, Fisher & Lindquist, 1993; Wentzel, 1991; Whittemore, 1924; Julian, Bishop, & Fiedler, 1966; Rabbie, & Wilkens 1971; Reeve, Cole & Olson, 1986; Reeve & Vallerand, 1984; Vallerand & Reid, 1984)
– Furthermore, perceived classroom competition was positively related to self-reported cheating behavior (Smith, Ryan & Diggins, Taylor, 1972; Pogrebin & Dodge, 2002 Whitley, 1998)
The nature of paradigm shift of competition
New paradigm Competition is not
opposed to cooperation
Competition is a heterogenous phenomena
Competition is a situational and personality variable
Old paradigm Competition is opposed
to cooperation
Competition is a homogenous phenomena
Competition is a situational variable
First goal of the present study
• To assess the impact of individual-level and situation-level competition-related variables on academic cheating.
– Aggressively competing students will cheat more, than those students who have positive attitudes towards competition.
– Such classroom atmosphere in which the goal is achieving the recognition of the teachers will induce more cheating, than such competitive classroom atmosphere which promotes skill development.
Second goal of the study
– Previous studies found that mastery and intrinsic motivations reduce cheating, whereas performance goals and extrinsic motivations enhance it. (Anderman, Griesinger, & Westerfield, 1998; Anderman & Midgley, 2004; Anderman & Murdock, 2007; Murdock & Anderman, 2006)
– In these studies performance goal orientation and competitive pressures are interpreted as overlapping concepts – NO DISTINCTION BETWEEN THE TWO
• Distinction between the effects of motivation- and competition-related factors’ of cheating.
Third goal of the study • Comparing the relative importance of motivational
and competition-related variables with proximal variables of cheating behavior such as:
(1) attitudes towards cheating (3) risk of detection (2) guilt (4) possible punishments
We hypothesize that the importance of motivational and competition-related factors is overrated in the literature of academic cheating
More proximal factors have vastly larger impact of cheating than extrinsic motivation and competition
Factors that have effect on cheating I.
• Grade point average – negative (Kerkvliet, 1994; Kerkvliet & Sigmund, 1999; Leming, 1980; Newstead, Franklyn-Stokes & Armstead, 1996; Whitley, 1998; Straw, 2002)
• Attitudes towards cheating – positive (Bolin, 2004; Jordan, 2001; Jensen, Arnett, Feldman & Cauffman, 2001; Whitley, 1998)
• Guilt – negative (Diekhoff, LaBeff, Shinohara & Yasukawa,1999; Malinowski & Smith, 1985)
• Classroom competition – positive (Anderman & Murdock, 2007; Smith, Ryan & Diggins, 1972; Taylor et al., 2002)
• Self-developmental competition – negative (Orosz, 2010)
• Hypercompetitive traits – positive (Orosz, Jánvári, Salamon, 2011)
Factors that have effect on cheating II.
• Motivation: Vallerand et al., 1992 – Academic Motivation Scale – Intrinsic motivation to know – negative – Extrinsic motivation of external regulation – positive – Amotivation - positive
• Risk of detection – negative (Heisler, 1974; Leming, 1978; Corcoran & Rotter, 1987; Covey, Saladin & Killen, 1989; Whitley, 1998)
• Expected punishments – negative (Bunn, Caudill & Gropper, 1992; Cohran, Chamlin, Wood & Sellers, 1999)
Hypotheses I. H1a: Attitude of aggressive competition will be positively
correlated with academic cheating H1b: Attitude of self-developmental competition will be
negatively correlated with cheating H1c: Positive attitude towards competition will be unrelated
to cheating H1d: Such classroom climate, in which the goal of
competition is recognition by the teachers, will be linked with academic dishonesties
H1e: Competitive climate, which promotes self-development, will lead to lower prevalence of cheating
Hypotheses II.
H2: Motivational pattern (IM, EM, AM) are separate from those of the competition-related individual and contextual variables
H3: the magnitude of the effects of motivational and competition-related variables on academic dishonesty is lower than those of the proximal individual (GPA, attitudes towards cheating, guilt) and situational variables (risk of detection, expected punishments)
Participants • 620 high school students (M = 264, F = 356)
• 19 classes from 7 schools – 2 schools upper third, 3 middle third, 2 lower third section of Hungarian high school ranking
• Age: 13-20 years, M = 16.66 years (SD = 1.51)
• Teachers were not present during data gathering
• 381 students filled in the questionnaire concerning competitive climate, 236 students did not fill in this scale
Measures - individual • Individual differences of competition:
– Aggressive competition scale: „I can be aggressive with my rivals” or “I’m often in conflict with my opponents”
– Self-developmental competition scale: “Competition helps me to improve my skills” or “Competition brings the best out of me”
– Positive attitudes towards competition scale: “I like the challenge of competition” or “Competition inspires me”
• Individual differences of motivation Vallerand et al. AMS: – IM to know – motivation to acquire knowledge – EM external regulation – learning due to only external pressures
and obligations – AM – the absence of motivation
Measures – proximal & situational • Two cheating vignettes: cheating sheet & copying
– self-reported cheating – acceptance – punishments – feeling of guilt – Perceived risk of detection
• Competition Climate Scale (CCS): – Constructive competition (CC) - it has positive impact on
students performance, creativity, interpersonal relationship – Destructive competition (DC) – it has negative impact on
relationships, the goal is achieving the recognition of teachers
Results - Descriptives Self-
reported cheating
Accept. Guilt Risk of
det. Exp.
punish.
Cheating sheets
no 24.6%
1 4.4% 40.5% 2.4% 14.4%
2 33.8% 37.1% 14.7% 82.6%
yes 75.4% 3 47.1% 17.5% 72.0% 2.5%
4 14.8% 4.9% 10.9% .5%
Copying no 38.1%
1 7.6% 35.3% 1.3% 12.1%
2 39.3% 33.2% 7.8% 86.2%
yes 61.9% 3 41.5% 21.1% 59.3% 1.2%
4 11.5% 10.4% 31.6% .5%
χ2/df = 1.786, CFI = .959, TLI = .952, RMSEA = .036
Results on the basis of the model I.
H1a: Aggressive competition has a positive indirect effect on SR cheating – proved
H1b: Self-developmental competition has a negative indirect effect on SR cheating – proved
H1c: Positive attitude towards competition is unrelated to cheating – proved
H1d: Such classroom climate, in which the goal of competition is recognition by the teachers, will be linked with academic dishonesties – not proved: DC climate is unrelated
H1e: Competitive climate, which promotes self-development, will lead to lower prevalence of cheating – not proved: CC climate is unrelated
Results on the basis of the model II.
H2: Motivational pattern (IM, EM, AM) are separate from those of the competition-related individual and contextual variables – proved: both CFAs and the model indicates that
H3: the magnitude of the effects of motivational and competition-related variables on academic dishonesty is lower than those of the proximal individual (GPA, attitudes towards cheating, guilt) and situational variables (risk of detection, expected punishments) – proved: (1) competitive climate (CC & DC) has no effect, (2) SD Comp & Aggr. Comp has small indirect effect, (3) extrinsic motivation no effect, BUT IM & AM have a serious effect!
Discussion “Competition is perhaps the single most toxic ingredient in a classroom, and it is also a reliable predictor of cheating” (Anderman & Murdock, 2007, p. XIII) – probably not true!
1. It is not the extrinsic motivation but the amotivation which counts!
2. Intrinsic motivation can prevent cheating!
3. Both motivational and competition-related effects are less important than proximal variables such as acceptance, GPA and guilt
− Are Competition and Extrinsic Motivation Reliable Predictors of Academic Cheating?
− Not really!!!
Practical implications • Eliminating competition from classroom is not the
best way to prevent cheating! (Tjosvold, Fülöp, etc) • Eliminating extrinsic motivation is not the best way
either in order to prevent cheating! (see Haraczkiewicz, Pintrich, etc.)
• Eliminating amotivation and increasing intrinsic motivation can be more useful! (enthusiasm studies)
• As teachers we have to create such environment during exams in which we prevent students from cheating: Risk of detection is more important than serious punishments (Houston)
Limitations & further directions • Hungarian educational context: in other contexts high
school competition can be more destructive which creates more conflicts, which induce individual cheating and prevent from collaborative cheating
• In Hungary a pretty large proportion of students cheated. It is surely different in other countries (i.e. France )
• It would be important to measure SR cheating in a more gradual way (occurrence of cheating per semester)