THE RESPONSE ENTROPY INDEX
The response entropy (RE) index calculates the balance of proportions of values endorsed by participants on Likert-scaled surveys, and is proposed as a new method for flagging careless response patterns. In two studies (publication in press) the RE index performed competitively with other methods of careless response detection. Unlike other methods of careless response detection, the RE index had the added benefit of avoiding conflating culturally diverse response patterns with careless responding. Current recommendation for this tool are to use it to flag suspicious protocols, however, manual inspection should precede decisions to remove protocols from datasets.
Click the following link to be taken to the calculator: https://sagenm.shinyapps.io/REICalculator/
THE TRIANGULATED RACIALIZATION INDEX (TRI)
The TRI computes the geometric area of a triangle determined by stereotype endorsement in reference to three racialized groups (e.g., Asian, Black, and White) mapped onto a three dimensional (e.g., body, mind, and self-interest) field. Conceptually, this measure determines the extent to which these racial groups are triangulated in relation to one another; operationally, this is represented by greater distances between vertices in the three dimensional field. In a forthcoming study (Social Cognition), among a sample of Asian (n = 64), Black (n = 73), and White (n = 165) adults, regression analyses partially supported predictions that the triangulation-based metric would be a stronger predictor of prejudice than single-group referenced stereotypes. Mediation analyses supported the theoretical perspective that triangulation relates to prejudice because people who simultaneously endorse stereotypes of Black and Asian people at extreme ends of bipolar continuums have relatively fixed views about the nature of race (i.e., racial essentialism).
Both the specific script I used for computing the TRI with my current three-factor model with Black, Asian, White referent groups and a “shell” version of this script that can be easily formatted for other social group applications are available at:
https://osf.io/2vrnx/
A sample dataset is also available for the shell script to aid in formatting data sets and testing the scripts. This shell script is specifically suited for a three factor stereotype model with three referent groups. The following are my procedural recommendations for developing a multidimensional stereotype measure:
1. Develop a pool of stereotype items relevant to the social groups of interest. Possible sources of item pools could include previously published research on stereotypes (as was the case in this current study) or qualitative, open-ended exploration of stereotypes. As mentioned, the 12-item pool of stereotypes used in the current model was likely too small and resulted in questionable internal reliabilities. I would recommend a minimum of 20.
2. Assess the endorsement of these items in relation to the multiple social referent groups of interest and in relation to a more general referent group such as “people in the entire world.” Following procedures by Chang and Demyan’s (2007), participant should be asked to estimate the percentage of members of each referent group whom they believed to possess each stereotype trait.
3. Conduct an exploratory factor analysis using only the items in reference to the “people in the entire world.” Including items in reference to specific social groups will likely result in factors clustering around the social groups themselves and not the stereotype dimensions. This can be done by either directly specifying three factors, or, if more than three factors emerge consider dropping the factors with the fewest items in order to retain the three strongest factors.
4. Compute factor scores for each social group for each dimension, comprising 9 separate scores for each participant. These 9 scores comprise the 3 coordinate points (x, y, z) for each of the three referent groups on a three dimensional field.
Determine the area of the triangle created from the three coordinate points in the three dimensional field.This can be done easily by accessing the syntax for the shell scripts (available in both SPSS and R) in the link above. In the shell scripts, A, B, and C are placeholders for the three referent groups and x, y, and z are placeholders for the stereotype dimensions.Values should be entered only in the highlighted fields.For example, stereotype dimension x for the referent group A should be entered in the A_x field.