Editorial.(Editorial)

The first paper, by Ming-huei Hsieh, is an investigation into the importance of country-of-origin in car brand preference and uses a novel application of correspondence analysis to extract the information. The findings of the study are based on a large, international survey and support a number of the hypotheses proposed by the author, including that consumers on the whole have a preference for a car from their own country or region. There are a number of caveats to these findings relating to general attitude to the brand and the availability of international products in the home country. However, the general method of analysis is worth looking at in detail as it uses the numerical output of correspondence analysis to generate a variety of variables for further analysis rather than just producing a visual display.

The second paper by Rune Lines and Jon Denstadli looks at the issue of information overload in conjoint studies. A series of hypotheses, based on a literature review, are formulated to predict the outcome of using varying numbers of attributes, and an experiment is designed to test them. The results show that as more attributes are used there is evidence of a perceived greater experience of information overload and definite evidence of reduced reliability in the predictive ability of the models developed. However, these effects were reduced for those who had more knowledge of the product area or were more motivated to do the task. The authors end by discussing the conflict of, on the one hand, the need to adequately describe the proposition (pressure for more attributes) and, on the other, the need to limit the respondent task (pressure to reduce attributes). This leads to interesting conclusions about the role of this technique in new product development.

The third paper by Eunkyu Lee, Michael Hu and Rex Toh investigates the reasons for respondents dropping out of panels and relates this to their level of item non-response. In particular it considers the variables of length of time of participation in the panel and question difficulty. Item non-response was found to be related to question difficulty and to the rate of attrition from the panel--certainly in the initial stages--after which attrition rate showed a slowing over time. During the course of the panel's duration item non-response was found to be comparatively stable and this was ...