Attitude Toward the Product/Brand (General
SCALE DESCRIPTION: The scales consist of various bi-polar adjectives presumed to measure the subject's overall evaluation of the product or brand. The various versions of the scale are similar in that they are not specific to any particular product or brand under investigation although certain adjectives may not beappropriate in some cases. Note that some scale users have referred to their measures by other names such as product evaluation, (e.g., Muthukrishnan and Ramaswami (1999), Gürhan-Canli and Batra (2004), and product utility (Thompson et al. 2005). In Lane (2000), a version of this scale was used as a brand extension evaluation. Stafford and Day (1995) made slightly different use of the scale than most ofthe others by measuring attitudes toward a service rather than a good. Given the directions used by Gürhan-Canli and Maheswaran (2000), their scale had the sense of a country-of-origin evaluation of a class of products. One of the three uses of the scale by Ruth and Simonin (2003) was with an event (parade) sponsored by two companies. Attitude toward a hotel chain was measured by Posavac et al.(2004). SCALE ORIGIN: There is no common origin for these scales and many of them are unique in that the sets of items of which they are composed have been used as a set in just one or two studies. Some items have been used much more than others but good/bad is by far the most commonly used bi-polar adjective. Many of the scales have used favorable/unfavorable and/or pleasant/unpleasant. At the otherextreme, there are several items (e.g., #22 to #25) that appear to have been used just once. Versions of the scale in languages other than English have been reported such as Korean (Choi and Miracle 2004; Taylor, Miracle, and Wilson 1997) and Chinese (Zhang and Schmitt 2001). RELIABILITY: Reported internal consistencies have ranged from below .70 (Iyer 1988) to .98 (Kozup, Creyer, and Burton2003). However, the reliabilities have tended to be on the high side with most of them being greater than .80 if not .90. See last section for specific reliabilities for each study. VALIDITY: Little if any evidence of scale validity was provided in the majority of the studies. A few authors conducted some testing, however, of unidimensionality (e.g., Anand and Sternthal 1990; MacInnis and Park). Batraand Stayman (1990) performed confirmatory factor analysis on their tenitem scale and indicated that there were two factors, one more hedonic and the other more utilitarian. However, since use of the two scales separately led to findings not
Copyright 2010, GCBII Productions
significantly different from those of the combined items, the latter was not discussed any further in the article. Afactor analysis was performed by Bezjian-Avery, Calder, and Iacobucci (1998) on items composing their attitude toward the brand and attitude toward the ad scales (#59). Each set of items appeared to be unidimensional with no cross-loadings greater than .34. Darley and Smith (1993) conducted several tests to determine if the three multi-item measures they used (brand attitude, ad attitude, and adcredibility) were sufficiently representative of their respective latent constructs. Among the findings was that a three factor model fit the data better than a one factor model. This provides some evidence of the scale's discriminant validity. Miller and Marks (1992) performed a factor analysis of nine items expected to measure either attitude-toward-the-ad or attitude-toward-the-brand. All of theitems had loadings of .65 or higher on the expected factors and was used to support a claim of each scale's discriminant validity. COMMENTS: See also Debevec and Iyer (1986), Desai and Keller (2002), Health, McCarthy, and Mothersbaugh (1994), Holmes and Crocker (1987), Kamins and Marks (1987), Maheswaran (1994), Nyer (1997), Orth and Holancova (2004), Pham et al.(2001), Prakash (1992), Priluck...