Skip to Main Content
 

Global Search Box

 
 
 
 

Files

ETD Abstract Container

Abstract Header

Face Validity and Decision Aid Neglect

Kajdasz, James Edward

Abstract Details

2010, Doctor of Philosophy, Ohio State University, Psychology.

Test makers are generally not concerned with how suitable a test item is perceived to be by the testee (face validity). Unlike test takers however, decision makers (DMs) may choose to disregard a decision aid if they believe it is unsuitable for its stated purpose. I present a scale adapted from Nevo (1985) to measure decision aid face validity (DAFV).

In three experiments, measures of DAFV obtained from one group of participants are shown to correlate with rates of decision aid reliance in another group of participants. Participants were asked to estimate the rent of an apartment. After making an initial estimate for the apartment, the participant was then presented with one of several possible DAs designed to estimate rent. The methodology of the aid was described along with the DA estimated rent. Participants were given the option to revise their initial rent estimate based on the new information provided by the DA if they desired. The FV of the aid (as measured during a previous pilot study) was correlated with the measure of DA reliance.

I also present research that integrates DAFV with decision maker confidence. The previous methodology is repeated except the apartments are located in four cities. Participants have high confidence in their ability to estimate the rent in the local city, but less confidence in their ability to estimate rent in far away cities. As shown in previous research, a negative correlation was observed between confidence and decision aid reliance. In addition, a hypothesized interaction was observed between confidence and DAFV. When participant confidence was high (i.e., local city), DA reliance tended to be low. When confidence was low (i.e. foreign city), DA reliance was high if DAFV was high. When DAFV was low, DA reliance tended to be low even when confidence was low.

Finally, how the performance of a DA is communicated was examined. The performance of a hypothetical decision aid was communicated in several different formats: 1) participants saw the mean accuracy of the DA, and the mean accuracy of the DM: “The decision aid was, on average, 70% accurate compared to 60% accuracy of unaided decision makers.” 2) participants were given mean DA accuracy only “The decision aid was, on average, 70% accurate.” 3) Participants were given the proportion of DMs beaten by the DA: “This decision aid performed better than 76% of human resource experts who did not use the aid.” Face validity was assessed for the aid using the various formats. Two experiments replicate the finding that “DA only” had consistently higher FV than “DA & DM” performance. This occurred even though the DA always performed better than the DMs. It seems participants were unimpressed by an aid that only outscored its expert decision maker by 10% or less, and punished the DA when this difference was made salient.

Hal R. Arkes, PhD (Advisor)
Thomas E. Nygren, PhD (Committee Member)
Michael L. DeKay, PhD (Committee Member)
122 p.

Recommended Citations

Citations

  • Kajdasz, J. E. (2010). Face Validity and Decision Aid Neglect [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1287176743

    APA Style (7th edition)

  • Kajdasz, James. Face Validity and Decision Aid Neglect. 2010. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1287176743.

    MLA Style (8th edition)

  • Kajdasz, James. "Face Validity and Decision Aid Neglect." Doctoral dissertation, Ohio State University, 2010. http://rave.ohiolink.edu/etdc/view?acc_num=osu1287176743

    Chicago Manual of Style (17th edition)