Careless responding detection revisited: Accuracy of direct and indirect measures.

To screen for careless responding, researchers have a choice between several direct measures (i.e., bogus items, requiring the respondent to choose a specific answer) and indirect measures (i.e., unobtrusive post hoc indices). Given the dearth of research in the area, we examined how well direct and indirect indices perform relative to each other. In five experimental studies, we investigated whether the detection rates of the measures are affected by contextual factors: severity of the careless response pattern, type of item keying, and type of item presentation. We fully controlled the information environment by experimentally inducing careless response sets under a variety of contextual conditions. In Studies 1 and 2, participants rated the personality of an actor that presented himself in a 5-min-long videotaped speech. In Studies 3, 4, and 5, participants had to rate their own personality across two measurements. With the exception of maximum longstring, intra-individual response variability, and individual contribution to model misfit, all examined indirect indices performed better than chance in most of the examined conditions. Moreover, indirect indices had detection rates as good as and, in many cases, better than the detection rates of direct measures. We therefore encourage researchers to use indirect indices, especially within-person consistency indices, instead of direct measures.

A comparison of conventional and resampled personal reliability in detecting careless responding.

Detecting careless responding in survey data is important to ensure the credibility of study findings. Of several available detection methods, personal reliability (PR) is one of the best-performing indices. Curran, Journal of Experimental Social Psychology, 66, 4-19, (2016) proposed a resampled version of personal reliability (RPR). Compared to the conventional PR or even-odd consistency, in which just one set of scale halves is used, RPR is based on repeated calculation of PR across several randomly rearranged sets of scale halves. RPR should therefore be less affected than PR by random errors that may occur when a specific set of scale half pairings is used for the PR calculation. In theory, RPR should outperform PR, but it remains unclear whether it in fact does, and under what conditions the potential gain in detection accuracy is the most pronounced. We conducted two studies: a simulation study examined the performance of the conventional PR and RPR in detecting simulated careless responding, and a real data example study analyzed their performance when detecting human-generated careless responding. In both studies, RPR turned out to be a significantly better careless response indicator than PR. The results also revealed that using 25 resamples for the RPR computation is sufficient to obtain the expected gain in detection accuracy over the conventional PR. We therefore recommend using RPR instead of the conventional PR when screening questionnaire data for careless responding.

On the Utility of Indirect Methods for Detecting Faking.

Indirect indices for faking detection in questionnaires make use of a respondent's deviant or unlikely response pattern over the course of the questionnaire to identify them as a faker. Compared with established direct faking indices (i.e., lying and social desirability scales), indirect indices have at least two advantages: First, they cannot be detected by the test taker. Second, their usage does not require changes to the questionnaire. In the last decades, several such indirect indices have been proposed. However, at present, the researcher's choice between different indirect faking detection indices is guided by relatively little information, especially if conceptually different indices are to be used together. Thus, we examined and compared how well indices of a representative selection of 12 conceptionally different indirect indices perform and how well they perform individually and jointly compared with an established direct faking measure or validity scale. We found that, first, the score on the agreement factor of the Likert-type item response process tree model, the proportion of desirable scale endpoint responses, and the covariance index were the best-performing indirect indices. Second, using indirect indices in combination resulted in comparable and in some cases even better detection rates than when using direct faking measures. Third, some effective indirect indices were only minimally correlated with substantive scales and could therefore be used to partial faking variance from response sets without losing substance. We, therefore, encourage researchers to use indirect indices instead of direct faking measures when they aim to detect faking in their data.