|
Article Excerpt
This paper examines the effect of alternative scale formats on reporting the nature and extent of attitudes toward grocery supermarkets on bipolar semantic differential measurement scales. A traditional one-stage format and an alternative two-stage format were tested in two studies conducted in different countries. In general, the two-stage format generated the greatest percentage of extreme-position (i.e. greatest amount) responses across scales, indicating that the more usual traditional one-stage format is subject to a central tendency form-related error. A test of predictive ability showed that the two-stage format was a better predictor of shopping behaviour in one country, whereas the results for the other country were mixed, although for the most part the two-stage format did a better job of prediction. Consequently, a question can be raised about the etic attributes of this measurement scale.
Introduction
Marketing and cross-cultural researchers use a number of different types of rating scales or scale formats in measuring attitudes or components of attitudes. Included in this 'arsenal' of approaches are the semantic differential, Stapel scale, Likert scale, Thurstone differential scale, and direct rating scales. One of the more widely used formats is the semantic differential, particularly for measuring images of such objects as brands, products, retail stores and other types of organisations (Osgood et al. 1957). In actuality, anything that could possibly form an image in a person's mind can be an object of measurement using the semantic differential.
The purpose of this methodological study is to report on the use of the semantic differential in two different cultures in detecting amount, or extent, of belief that respondents have about their attitudes -- in this case, their images of specified objects. An alternative format is proposed and tested against the more traditional approach. The 'amount' issue is one of possible central tendency error in that there is reluctance on the part of respondents to give extreme responses.
Using respondents from two different cultures addresses some of the emic/etic issues in scale and measurement formats. An aspect of methodology is an emic when it is culture-bound. That is, it behaves in a specified way in one culture and one culture only. When it operates similarly in many cultures, it is considered to be culture-free and is an etic (Berry 1980, p. 11). By examining method in more than one culture, an aspect of external validity is assessed. In doing this, imposed etic validity is established by correctly predicting an outcome in a culture by using a test imposed from another culture (Berry 1980, p. 19). Of concern is form-related bias arising from development in the country/culture of origin, and adaptation to the specific national or cultural context of interest (Douglas & Craig 1983, p. 134).
Approaches to image measurement
Attitude has been defined as 'a psychological tendency that is expressed by evaluating a particular entity with some degree of favour' (Eagley & Chaiken 1993, p. 1). Classically, attitudes were assumed to have three components: (1) cognitive, (2) affective, and (3) behavioural. This classical tripartite view of attitude has been abandoned by most researchers (Sudman et al. 1996, pp. 123-124). Current theoretical models of attitudes suggest that there are two dimensions -- direction (e.g. positive or negative predisposition towards an object) and strength (Petty & Krosnick 1995). For example, a person may like or dislike a product and may hold this attitude with a varying degree of strength or intensity -- interpreted as either confidence, certainty, accessibility, or conviction (Berger & Alwitt 1996). An aspect of both attitude direction and strength is amount or extent, particularly when the attitude is based on belief that an object can be described by characteristics, thus forming an image. In this context direction and its amount are both cognitive judgerrients, whereas strength is an affective judgement.
A variety of approaches have been used to measure images. In addition to the semantic differential, these include the Fishbein model, multi-dimensional scaling, open-ended questions, Likert-type scales, and the Stapel scale. Although researchers have suggested that other scaling formats are useful, most image researchers continue to use some version of the semantic differential or horizontal bipolar adjective/phrase rating scale.
Three major versions of the semantic differential for measuring multi-object, multi-attribute perceptions have been widely used:
1. traditional semantic differential (TSD), for which an object is rated on all attribute dimensions before the next object is rated;
2. modified traditional semantic differential (MTSD), for which all objects are evaluated on a single attribute before. another attribute is introduced; and
3. graphic positioning scale (GPS), for which all objects are evaluated on the same scale via graphics reflecting their relative perceptual placement on a scale between polar opposites.
Golden and her colleagues introduced a fourth approach, the numerical comparative scale (NCS), which combines the desirable numerical properties of other horizontal bipolar adjective/phrase scales with...
|
