Date of Award
Doctor of Philosophy
William R. Nugent
John C. New, Neil Greenberg, John E. Nolt
The purpose of this study was to investigate the extent to which the word “pet” provided valid results in the Lexington Attachment to Pets Scale (LAPS) (Johnson, Garrity, & Stallones, 1992). A multiple-groups confirmatory factor analysis, using maximum likelihood (ML) estimation, was conducted to test the hypothesized factor structures in the specified measurement model. Respondents to the Original version (i.e. used the word pet) and the Personalized version (i.e. embedded the canine’s name as typed by participants) made up the two groups. Features of an electronic survey were necessary for this personalization and systematic alternative assignment to the two LAPS versions. A snowball sampling method utilized electronic mail to invite self-selected participants meeting the following criteria: 18 years or older; lived in the United States; and at least one dog living inside/outside their home and for whom they are responsible at least some of the time.
According to collected human and canine demographics, the Original (n = 1,854) and Personalized (n =1849) groups appeared to be statistically equivalent (N = 3,703). The hypothesized measurement model generated a χ2 value of 4130.242, with 264 degrees of freedom and a probability of less than .001 (p <.001), suggestive of a lack of fit. However, goodness-of-fit indices were a consideration. Comparing the CFI (.95 vs. .862), TLI (.95 vs. .840) and RMSEA (<.05 vs. .063) cut off values with the results from this study respectively, reveals again, an inadequate fit. These results imply that the hypothesized measurement model was not consistent with the data and precluded specific tests of differential validity.
While the results of this study made it inadvisable to further examine the differential validity associated with linguistic differences in the scale items, critical information was nonetheless identified. According to the data in the current study, the LAPS conceptualization of “pet attachment” (sic) may not be valid. Until further research provides stronger evidence, use of this scale could produce results that lead to invalid inferences. The original LAPS conceptualization and quantification of human and canine relationships is still a work in progress. Moreover, psychometric work is critically necessary before using the LAPS.
Zaparanick, Tracy Leigh, "A Confirmatory Factor Analysis of the Lexington Attachment to Pets Scale. " PhD diss., University of Tennessee, 2008.