Phantom behavioral assimilation effects: systematic biases in social comparison choice studies

Area of Science:

• Psychology
• Social Psychology
• Personality Research

Background:

• Social comparison theory (SCT) suggests individuals compare themselves to others.
• Previous research indicated upward comparison choices benefit subsequent performance (behavioral assimilation effect, BAE).
• Uncontrolled measurement error in pretest achievement was a potential confound in prior studies.

Purpose of the Study:

• To investigate the role of measurement error in the observed behavioral assimilation effect (BAE).
• To re-evaluate the magnitude of BAEs using methods that control for measurement error.
• To demonstrate the utility of latent-variable models in personality research with correlational data.

Main Methods:

• Studies 1 and 2 utilized pretest achievement data to examine the influence of measurement error on BAE.
• Study 3 employed simulated data and latent-variable models with multiple indicators to eliminate phantom BAEs.
• Studies 4 and 5 applied latent-variable models to existing datasets (Blanton et al., 1999; Huguet et al., 2001) to correct for measurement error.

Main Results:

• Uncontrolled measurement error in pretest achievement partially accounted for the apparent BAE in initial studies.
• Latent-variable models successfully eliminated phantom BAEs in simulated data.
• Re-analysis of existing data using latent-variable models yielded substantially smaller, yet still significant, positive BAEs.

Conclusions:

• The behavioral assimilation effect (BAE) is partially an artifact of measurement error in pretest variables.
• Latent-variable modeling provides a robust approach to correct for measurement error biases in personality research.
• Failure to control for measurement error positively biases the apparent effects of personality variables in correlational studies.