response processes in sjt performance: the role of general and specific knowledge
DESCRIPTION
Response Processes in SJT Performance: The Role of General and Specific Knowledge. 27 th Annual Conference of the Society for Industrial & Organizational Psychology April 27, 2012. Matthew T. Allen. James A. Grand. Kenneth Pearlman. - PowerPoint PPT PresentationTRANSCRIPT
RESPONSE PROCESSES IN
SJT PERFORMANCE:
THE ROLE OF GENERAL AND
SPECIFIC KNOWLEDGE
James A. Grand Matthew T. Allen
Kenneth Pearlman
27th Annual Conference of the Society for Industrial & Organizational Psychology
April 27, 2012
The views, opinions, and/or findings contained in this presentation are solely those of the authors and should not be construed as an official Department of the Army or Department of
Defense position, policy, or decision, unless so designated by other documentation.
2
• Turning a critical eye towards SJT construct validity & its assumptions• What do SJTs measure?– An alternative take on an old
question– A response process model for SJT
performance
• Predictions of the response process model– Empirical support?
• Implications for interpretation & validity of SJTsResponse Processes in
SJT Performance
Presentation Overview
3
Acknowledgement
• The authors would like to thank the U.S. Army Research Institute for the Behavioral and Social Sciences (ARI) for allowing the use of their data for this research• More information about the
study can be found in:Knapp, D. J., McCloy, R. A., & Heffner, T. S. (Eds.) (2004). Validation of measures designed to Maximize 21st-century Army NCO performance (TR 1145). Arlington, VA: U.S. Army Research Institute for the Behavioral and Social Sciences.
4
Validity Evidence for SJTs
• Established evidence of criterion validity between SJT and job performance (cf., Chan & Schmitt, 2002; Clevenger et al., 2001; McDaniel et al., 2001)– Estimates in low .20s (corrected validities near mid-.30s)
(Chan & Schmitt, 2005)
• Agreement on construct validity is less certain...First-order Constructs
• Multiple, distinguishable dimensions• Specific a priori subscales
• Oswald et al. (2004)• Career orientation• Perseverance• Multicultural appreciation
Second-order Constructs
• Singular, high-level dimension
• Broad focal target
• Lievens et al. (2005)• Interpersonal skills
• Mumford et al. (2008)• Team role knowledge
“Practical Intelligence”
• Tacit knowledge or “common sense”• Everyday reasoning
• Sternberg et al. (2002)• “Practical know-how”
• Chan & Schmitt (2005)• Contextual knowledge
5
A Test in Name Only
• Virtually all perspectives approach and treat SJT measurement in a manner consistent with Classical Test Theory
• SJTs are NOT tests! (at least in the traditional sense of the word)– Low-fidelity simulations (Motowidlo et al., 1990)– Measurement methods capable of capturing a variety of
constructs (Chan & Schmitt, 2005; McDaniel & Nguyen, 2001)
X = T + EObserved Score = True
Score + Errorij
ij
rkrk
)1(1
6
So What Do SJTs Measure?
“SJT performance clearly involves cognitive processes. [...] Addressing basic questions about these underlying cognitive processes and eventually understanding them could provide the key to explicating constructs measured by SJTs”
(Chan & Schmitt, 2005)
“So far, there does not exist any theory about how people answer SJTs or about what makes some SJT items more or less difficult than others.”
(p. 1044, Lievens & Sackett, 2007)
7
So What Do SJTs Measure?
• Rather than conceptualize SJTs as though they measure a static construct or “true score,” SJTs capture sophistication of a respondent’s reasoning process
• By their nature, SJTs capture similarity between respondent reasoning and that implied by keyed responses
?≠
8
A Response Process Model of SJT Performance
• Generic dual-process accounts of human reasoning, judgment, and decision-making (Evans, 2008)
System 1
• Implicit, intuitive, and automatic reasoning
• Decisions guided by general heuristics , which are informed by domain experiences
• High capacity, low effort processing
System 2
• Systematic, rational, and analytic reasoning
• Decisions guided by controlled, rule-based evaluations and conscious reflection
• Low capacity, high effort processing
9
A Response Process Model of SJT Performance
• Dual-process accounts have been applied in a variety of perceptual, reasoning, and decision-making tasks (see Evans, 2008)– Extensions of dual-process model serve as foundation for
much of judgment & decision-making literature (e.g., Gigerenzer et al., 1999; Kahneman & Frederick, 2002, 2005)
Central Tenets of Dual-Process ModelsBecause of limits on our cognitive capacity and information processing...• System 1 reasoning is primary determinant of
judgment/decision-making in most situations• System 2 reasoning is typically engaged to evaluate
the quality of decisions or in attempts to consciously contrast alternatives
10
Dual-Process Accounts as a Response Process Model of SJT Performance
• Two predictions based on dual-process account relative to SJT performance:
– Beliefs about the general effectiveness of various behaviors, dispositions, or approaches serve as baseline heuristic for reasoning across many situations (cf., Motowidlo et al., 2006; Motowidlo & Beier, 2010)» “It is good to be thorough and conscientious in one’s work.”
– Domain experience/knowledge leads to development of more conditional, refined, and nuanced heuristics (Hunt, 1994; Klein, 1998; Phillips et al., 2004)» “It is good to be thorough and conscientious in one’s work, but you can
generally skimp on Task X and still do just fine.”– Thus, generalized heuristics/beliefs/temperaments become less
predictive of SJT performance as experience increases
As job experience increases, the predictive validity of domain-general heuristics on SJT performance will decrease
1
11
Dual-Process Accounts as a Response Process Model of SJT Performance
• Two predictions based on dual-process account relative to SJT performance:
– Common for respondents to identify/rate most and least effective/preferred SJT response options (McDaniel & Nguyen, 2001)
– Identifying most effective option should engage System 1 reasoning» Select most reasonable option based on intuitive heuristic, less effortful
processing– Identifying least/less effective option should engage System 2
reasoning » “Play out”/evaluate consequences of remaining options, more effortful
processing– Thus, identifying least/less effective option more g-loaded than
identifying most/more effective option
Cognitive ability will be more strongly related to assessment of less preferred SJTs options than more preferred options
2
12
Methods & Measurement
• Concurrent validation study on predictors of current and future expected job performance of Army NCOs (n = 1,838) (Knapp et al., 2004)– Primarily interested in predicting leadership
performance/potential– Sample:Rank N Relevant Experience
E4 433 Little to no leadership experience
E5 864 First opportunities for leadership experience, have received some leadership training
E6 541 More leadership experience across greater variety of situations
13
• Domain-general heuristic measures– Differential attractiveness: individuals
who more strongly endorse a trait/quality perceive behaviors which reflect that trait/quality as more effective (Motowidlo et al., 2006; Motowidlo & Beier, 2010)
– Temperament inventories» Assessment of Individual Motivation (AIM)
1. Multidimensional 38-item forced choice measure (α ≈ .60 all scales)
» Biographical Information Questionnaire (BIQ)1. Multidimensional 156-item self-report biodata
questionnaire (α ≈ .70 all scales)
• General cognitive aptitude (ASVAB)• 40-item SJT on
leadership/interpersonal skills (Knapp et al., 2002)– 5 response alternatives, SMEs rated all
options– Respondents chose most & least effective
options» Responses recoded to SME ratings
Response Processes in SJT Performance
Methods &
Measurement
14
Empirical Examination of Predictions from Dual Process Model
As job experience increases, the predictive validity of domain-general heuristics on SJT performance will decrease
1
AIM
-- D
epen
dabi
lity
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
E4 E5 E6
Correlation with Most Effective Response RatingLow Dependability High Dependability
4.75
4.85
4.95
5.05
5.15
5.25
5.35
5.45E4 E5
Mos
t Effe
ctiv
e Re
spon
se
Rati
ng
Regression Summary Main effect of
temperament Main effect of experience Significant interaction• Relationship stronger for
less experienced
Results consistent across
all scales & SJT scores
15
-0.10 0.00 0.10 0.20 0.30 0.40 0.50
Empirical Examination of Predictions from Dual Process Model
As job experience increases, the predictive validity of domain-general heuristics on SJT performance will decrease
1
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
E4 E6
Leadership
Physical Conditioning
Agreeableness
Work Orientation
Adjustment
Dependability
Interpersonal Skill
Leadership
Openness
Tolerance for Ambiguity
Social Maturity
Social Perceptiveness
AIM Scales
BIQ Scales
16
Cognitive ability will be more strongly related to assessment of less preferred SJTs options than more preferred options
2
Empirical Examination of Predictions from Dual Process Model
E4 E5 E60
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 Most Effective Least Effective
Corr
elat
ion
betw
een
Gene
ral C
ogni
tive
Apti
-tu
de a
nd S
JT R
espo
nse
Rati
ng
17
Conclusions & Implications
• Most research on SJT measurement, development, and validity has largely been atheoretical (but see Motowidlo & Beier, 2010)– Dual-process account appears to be a reasonable response process
model– Currently working on more explicit empirical examination (see
also Foldes et al., 2010)
• What does having a response process model buy us?– SJT construct validity: Constructs vs. Reasoning
» Could label it “practical intelligence,” but even that depends on...– Interpretation of SJT performance
» Who is selected as the “experts” holds significant importance » Extent to which respondents reason/process information in a manner
similar to “experts”– Response elicitation affects SJT interpretation
» Most likely option/ratings = more heavily influenced by heuristic reasoning» Least likely option/ratings = more heavily influenced by cognitive reasoning
RESPONSE PROCESSES IN
SJT PERFORMANCE:
THE ROLE OF GENERAL AND
SPECIFIC KNOWLEDGE
James A. Grand Matthew T. Allen
Kenneth Pearlman
27th Annual Conference of the Society for Industrial & Organizational Psychology
April 27, 2012
The views, opinions, and/or findings contained in this presentation are solely those of the authors and should not be construed as an official Department of the Army or Department of
Defense position, policy, or decision, unless so designated by other documentation.