Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
The Detection of Invalid Scores in Non-cognitive Measurement: An Investigation of Person–Fit Statistics for Use with Social–Emotional Learning Assessments
Abstract
An area of non-cognitive measurement that has been receiving increasing attention is the measurement of social–emotional learning (SEL) skills in K-12 schools, with both educators and policy makers becoming aware of the importance of these skills for academic achievement. However, challenges in the measurement of non-cognitive skills such as SEL continue to exist in the form of validity threats often found in self–report measures such as a reference bias, social desirability, or random responses. One area of measurement that may have a potential for use in the detection of invalid responses in non-cognitive measurement is person–fit statistics. The current study investigated the detection power and error rates of the lpz, GPn, INFIT, OUTFIT, and their standardized versions using the Partial Credit Model under conditions similar to those found in SEL measures.The current study hypothesized that the nonparametric statistic, the GPn, would outperform the other person–fit statistics included in the study. The current study also investigated the detection rates of these statistics under varying conditions of sample size, test length, Type I error rates, and levels of both global and individual misfitting responses.Results from the simulation study showed that the GPn outperformed the all other statistics with tests containing four items, with the lpz showing the highest power and lowest Type II errors across all other conditions. An empirical study was completed and suggests that the probability of a student being classified as producing a significantly misfitting response vector differs by gender, socio-economic status, and ethnicity. The results of this study suggest that person–fit analyses may have utility in the detection of invalid scores in the large-scale measurement of social–emotional learning.
Subject Area
Quantitative psychology|Educational psychology
Recommended Citation
Trantham, Pamela Sue, "The Detection of Invalid Scores in Non-cognitive Measurement: An Investigation of Person–Fit Statistics for Use with Social–Emotional Learning Assessments" (2019). ETD collection for University of Nebraska-Lincoln. AAI27667025.
https://digitalcommons.unl.edu/dissertations/AAI27667025