The participants for this study will be undergraduate college students in introductory criminology and criminal justice classes at John Jay College of Criminal Justice in New York, NY. Based on preliminary power analyses, a minimum sample size of 159 will be necessary to detect a medium-sized effect (f = .25) with a power of .80 at an α level of .05. The participants will be randomly assigned using a computer based algorithm to watch one of three brief videos which depict a police traffic stop resulting from a minor traffic violation. Given the findings from previous studies, which highlight the influence of age, sex, and race in perceptions of police-citizen interactions (see Engel, 2005; Hurst …show more content…
Specially, he will use commands without politeness features such as merely saying license and registration or step out of the vehicle. The second video represents the procedural justice condition, in which the officer’s script from the control condition is modified to include key linguistic elements of procedural justice and accommodation theory. In contrast to the control condition, the officer will use commands with politeness features in accordance with those discussed Brown & Levinson (1987) such as reforming his commands for documentation into a respectful request May I have your license and registration. The third video will represent the overaccommodation condition where the officer will speak to the driver in a casual or informal manner without invoking his formal authority. To accomplish this, the script from the procedural justice condition will be further modified to include key elements of overaccommodation. Specifically the overaccommodation condition will incorporate features such as mitigated commands: If it isn’t too much trouble, why don’t you just give me your license and …show more content…
If all three dummy variables were included in the model simultaneously perfect multicollinearity would occur, thus to examine different contrasts between experimental conditions each condition must be tested twice wit the third condition excluded.
The indicators used to measure each latent outcome variable in the study are ordinal variables, consequently the models will be estimated using a weighted least squares estimator (WLSMV). Monte Carlo simulation research has shown that the WLSMV estimator performs well for models with categorical indicators (DiStefano & Morgan, 2014; Flora & Curran, 2004; Muthén, Du Toit, & Spisic, 1997). However, because WLSMV methods are designed for large sample size they have no guarantee to work well in studies with smaller samples (Asparouhov and Muthén 2010), all models will also be estimated with a Bayesian estimator using the Markov Chain Monte Carlo algorithm. Simulation research has shown that this estimator has a greater degree of reliability when working with CFA models with ordinal variables and small sample sizes (Asparouhov & Muthén, 2010; Liang & yang,