Date of this Version
Al Baghal, MT (2012) Numeric Estimation and Response Options: An Examination of the Measurement Properties of Numeric and Vague Quantifier Responses, Ph.D. Dissertation
Many survey questions ask respondents to provide responses that contain quantitative information. These questions are often asked requiring open ended numeric responses, while others have been asked using vague quantifier scales. How these questions are asked, particularly in terms of the response format, can have an important impact on the data. Therefore, the response format is of particular importance for ensuring that any use of the data contains the best possible information. Generally, survey researchers have argued against the use of vague quantifier scales. This dissertation compares various measurement properties between numeric open ended and vague quantifier responses, using three studies containing questions with both formats. The first study examines uses new experimental data to compare accuracy between the measures; the second and third use existing data to compare predictive validity of the two formats, with one examining behavioral reports, the other examining subjective probabilities. All three studies examine the logical consistency between measures, and the potential correlates related to improved measurement properties. Importantly, these studies examine the influence of numeracy, a potentially important but rarely examined variable. The results of the three studies indicate that vague quantifiers may have better measurement properties than numeric open ended responses, contrary to many researchers’ arguments. Studies 2 and 3 are most clear about this increased strength; in both of the studies, using a number of tests, the predictive validity of vague quantifiers was consistently greater than that of numeric open ended responses, regardless of numeracy level. Study 1 shows that at that generally, vague quantifiers result in more accurate data than numeric, but this finding depends on other factors, such as numeracy. Therefore, numeracy was infrequently found to be important, but at times did have an impact on accuracy. Further, in the three studies, it was found that the two formats were logically consistent when translations between the questions were directly asked for, but inconsistency occurred when there was not a direct translation.
Advisor: Robert Belli