Supplemental Materials
Who Cares and Who Is Careless? Insufficient Effort Responding as a Reflection of Respondent Personality
by N. A. Bowling et al., 2016, Journal of Personality and Social Psychology
Inconsistency indices equate IER with the presence of self-contradictory responses.A participant, for instance, who answersstrongly agree to the job satisfaction item “I like my job” and answersstrongly disagree to the job satisfaction item “I am happy with my job,” has provided inconsistent responses to two essentially interchangeable items, thus displaying evidence of IER. As described below, we computed the individual reliability, psychometric synonyms, and psychometric antonyms indices—each of which is a type of inconsistency index—by calculating a series of within-person correlations for each participant (see Huang et al., 2012;Meade & Craig, 2012).
The individual reliability index assesses the consistency with which a given participant responds to different items from the same substantive measure. To compute the individual reliability index, we used eight substantive measures (trait PA, trait NA, core-self evaluations, interpersonal treatment by one’s coworkers, interpersonal treatment by one’s supervisor, job satisfaction, work centrality, and life satisfaction) and computed two vectors of scores for each participant (eight variablesper vector): (a) the mean score on the even-numbered items of each of the eight substantive measures, and (b) the mean score on the odd-numbered items of each of the eight substantive measures. A given participant’s individual reliability index score was the within-person correlation between that person’s even-number item means and the corresponding odd-numbered item means. So that a high score would reflect high probability of IER, we multiplied the individual reliability index by -1 (in the remainder of the current paper, we refer to this recoded measure as the Individual Unreliability IER index).
The psychometric synonyms index assesses the consistency with which a given participant responds to items that have a strong positive relationship in the sample as a whole, whereas the psychometric antonyms index assess the consistency with which a given participant responds to items that have a strong negative relationship in the sample as a whole. To compute the psychometric synonyms index, we first identified 30 item pairs that yielded the strongest positive correlationsamong our entire sample and arranged them into two vectors of scores (30 variables per vector). The psychometric synonyms index was the within-person correlation between these two vectors. Thus, a high score on the psychometric synonyms index indicates low IER. So that a high score would be indicative of high probability of IER, we multiplied the psychometric synonyms index by -1 (in the remainder of the current paper, we refer to this recoded measure as the Psychometric Synonyms IER index).
To compute the Psychometric Antonyms IER index, we first identified 30 item pairs that yielded the strongest negative correlationsamong our entire sample and arranged them into two vectors of scores (30 variables per vector). The Psychometric Antonyms IER index was the within-person correlation between these two vectors. A high score on the Psychometric Antonyms IER index naturally indicates high probability of IER, thus it was not re-coded.
Outlier indicesequate IER with the improbable combination of responses to different items within a single scale. Following Meade and Craig (2012), we computed a multivariable distance measure—Mahalanobis D—for each of the eight scales by regressing scale scores onto vectors of individual items on the scale (e.g., we regressed the T1 work centrality scale onto its 12 items). The resulting eight Mahalanobis D measures were standardized and averaged into a single Mahalanobis D IER index (α = .61 and .65 for T1 and T2, respectively). A higher score on the Mahalanobis D IER index indicates high probability of IER.