Traylor, Zachary Kyle (2018-10). Device-Engendered Cognitive Ability Score Differences on Unproctored Internet-Based Assessments: The Role of Selective Attention. Master's Thesis. Thesis uri icon

abstract

  • Although a volume of literature suggests that the device used to complete unproctored Internet-based tests (UIT) affects observed test scores, there have been a limited number of attempts to provide a psychological explanation for why this occurs. One such exception is Arthur, Keiser, and Doverspike's (2018) Structural Characteristics/Information Processing (SCIP) model, which provides a psychological explanation regarding the conditions under which one would expect UIT device types (e.g., desktop computer, smartphone, tablet) to affect test scores. The model proposes that systematic error is introduced via construct-irrelevant cognitive load attributable to the additional information-processing demands elicited by the UIT device's structural characteristics. While conceptually sound, there has been only one empirical examination of the propositions advanced by the model to date. Consequently, the primary objective of the present study was to test the SCIP model's propositions regarding selective attention, the information-processing demand elicited by permissibility, the associated structural characteristic of UIT devices. Two hundred sixty-one participants completed measures of general mental ability (GMA), personality, and selective attention. Participants were randomly assigned to one of two conditions differing in terms of the testing (1) environment and (2) device used to complete the GMA test and personality assessment (i.e., a busy, outdoor location [smartphone condition] or a quiet, indoor location [desktop condition]). Scores on the GMA test did not differ as a function of the testing device and environment, however, in accordance with the tenets of the SCIP model, it appears that test takers in the smartphone condition experienced a greater degree of selective-attention demands while completing the GMA test. All of the observed results are interpreted within the context of using an undergraduate student sample, low testing stakes, and random assignment instead of the self-selection of participants into conditions. Implications and limitations of the present study as well as recommendations for future research are discussed.

publication date

  • December 2018