1 / 22

Department Psychology • Matthias Ziegler

Department Psychology • Matthias Ziegler. People fake! - So what?. Contents The BIG 5 – Knowldege and questions? Study design 3 questions General Conclusion. The BIG 5 – Knowldege and questions? Latent State Trait Theory (LST) Steyer, Ferring & Schmitt (1992)

winola
Download Presentation

Department Psychology • Matthias Ziegler

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Department Psychology • Matthias Ziegler People fake! - So what?

  2. Contents • The BIG 5 – Knowldege and questions? • Study design • 3 questions • General Conclusion

  3. The BIG 5 – Knowldege and questions? • Latent State Trait Theory (LST) Steyer, Ferring & Schmitt (1992) • up to 20% of variance in a questionnaire state or interaction (Deinzer et al. 1995) • Correlations between personality dimensions increase due to faking • Schmit & Ryan, 1993; Pauls & Crost, 2005 • Meta-analytical evidence for correlated dimensions • Mount, Barrick, Scullen & Rounds (2005) • true correlations up to ρ = .52 between N and C • Higher order personality factors (α&β, Digman, 1997)  There is a situational influence when measuring personality  How does that influence impact construct validity?

  4. The BIG 5 – Knowldege and questions? • BIG 5 prediciting job performance • C  r = .31 • Meta analysis Barrick, Mount & Judge (2001) • BIG 5 prediciting academic performance • Furnham & Chamorro Premuzic 12 % incremental validity to IQ  BIG 5 predict performance  Where does the predictive power come from? Trait or fake?

  5. The BIG 5 – Knowldege and questions? • What happens when people fake? • Models for faking from McFraland and Ryan (2001), Snell et al. (1999) • little empirical support/research • new model from Mueller-Hanson, Heggestad & Thornton (2006) • published after my project  • faking regarded equal between people (but Zickar, Gibby & Robie, 2004) • Study idea • qualitative analysis using cognitive interviews • Mixed Rasch Model (C) to detect different answer styles • explore differences between the classes

  6. Study design • Integration of LST Theory and ICE Design (Steyer, 2005) • 2 measurement times (LST) • NEO – PI – R twice • variance can be split into faking and personality • 2 groups (ICE) • CG  normal instructions twice • EG  2nd time concrete situation (test for student selection, good result, expert) • causal interpretation possible • Hypotheses • H1: A specific faking takes place in the EG  correlations between faked dimensions increase • H2: Controlling situational demand strongly diminishes correlations

  7. State 2: EG > CG State 1: CG = EG + CG: State 1 = State 2 State 1 State 2 (Fake) C11 C12 C21 C22 C31 C32 C41 C42 C51 C52 C61 C62 e e e e e e e e e e e e C LST Theory + ICE Design

  8. Results • 1st semester psychology students • NCG = 94 NEG = 92, about 70% females in both groups • demografics comparable • What was faked? Except for O all means differ substantially (and significantly) from time 1 to time 2 in the EG but not in the CG. Cohen‘s d for repeated measurement designs

  9. Results • What happened to the correlations? Time 1 Time 2 Above the diagonal are the correlations within the control group and beneath the diagonal within the faking group. * p < .05 ** p < .01 Correlations increase despite diminished variance!

  10. Results • fit indices of SEM • χ² = 3768.03 (2051), Bollen Stine p = .33 • CFI = .81; RMSEA = .067 (.064 - .071); SRMR = .138 • means and latent means between groups Groups differ significantly only in their amount of faking after controlling for situational demand!

  11. Results • What happened to the correlations? • Not part of the model  not necessary; inclusion does not improve model (neither does a higher order factor!) Correlations diminish drastically (E and A!) • significant state and trait variance (E!) • mostly substantial trait and state paths

  12. Conclusion I • faking had a causal effect on structure and means of the BIG 5 • specific faking took place causing highly inflated correlations (H1) and mean differences (except for Openness) • controlling the situational demand (H2) • both groups have the same means in personality dimensions • correlations diminished  uncorrelated BIG 5 structure in both groups • Extraversion and Agreeableness still share a lot of variance  explains troublesome SRMR • replication in larger and different (applicant) samples

  13. Next question • What predicts performance? • trait or state (faking)

  14. Very complex design • only faked facors were used • Pauls & Crost (2005) • within the CG loadings on the dimensions were set equal for each dimension • Allik & McCrae (2004) • Model fit • χ² = 3977.98 (2175), Bollen Stine p = .38 • CFI = .80; RMSEA = .067 (.064 - .070); SRMR = .137

  15. Results • What predicts performance? • Dimension variance drops  loadings only from 1 or 2 facets

  16. Results • What is faking? • correlations between faking variance and other measures

  17. Conclusion II • criterion validity as effect size remains stable • faking variance adds only little to the prediction • but positively • faking does influence construct validity • only few facets predict performance • faking is related with self efficacy beliefs • Question • What happens when people fake?

  18. Question 3 – What happens when people fake? • Qualitative analysis • N = 50 • 2 different faking strategies were used • slight faking and extreme faking • only relevant items were faked • unimportant items were answered honestly or neutrally Mixed Rasch Model • 3 class solution had best fit • regular respondents (4%), slight faker (69%), and extreme faker (27%)

  19. Differences between the classes • multinomial logistic regression with rr as reference categoy • no differences in criterion validity (R² = .02)

  20. Conclusion III • only important items are faked • 2 different faking styles • faking depends on trait, ability, age, and gender • no differences in criterion validity

  21. General Conclusion Model of Responding to Situational Demand

  22. Contact • Matthias Ziegler • Ludwig-Maximilians-University Munich • Department Psychology • Leopoldstraße 13 • 80802 München • phone: +49 89 / 2180 6066 • fax: +49 89 / 2180 3000 • Email: ziegler@psy.uni-muenchen.de Thank you for your attention!

More Related