Publication Date
2016
Document Type
Thesis
Committee Members
Nathan Bowling (Committee Member), Gary Burns (Committee Member), David LaHuis (Advisor)
Degree Name
Master of Science (MS)
Abstract
Insufficient Effort Responding (IER) is prevalent enough in self-report data to cause issues with construct validity. There are many ways to detect IER, but they are less than ideal as they each detect different forms of IER. I compared an Item Response Theory (IRT) approach consisting of the lz person-fit statistic and the Person Fluctuation Parameter (PFP) to longstring, non-consecutive longstring, even-odd split, and psychological synonyms indices. I simulated 3200 samples with one of four types of random responding: consecutive responding, non-consecutive patterned responding, random responding following a normal distribution, and random responding following a uniform distribution. Also, I generated an additional sample that consisted of all types of IER examined within this study. I found that the IRT methods are able to detect IER considerably better than the other indices, excluding using the longstring method to detect consecutive responding. As such, they are robust enough to detect most forms of IER. I conclude that using IRT approaches after removing the obvious IER cases with the longstring index is the best way to detect IER.
Page Count
50
Department or Program
Department of Psychology
Year Degree Awarded
2016
Copyright
Copyright 2016, all rights reserved. My ETD will be available under the "Fair Use" terms of copyright law.
