Recent decades have seen increased interest in finding out what motivates tourists to behave the way they do. In common with consumer behaviorists in general, however, there has been difficulties in choosing appropriate methods. Thus, a number of different approaches have been adopted. In an effort to ascertain just how interchangeable the more commonly used methods are, a survey was undertaken. This three-part instrument purportedly measures tourist motivation, but the sections do so in different ways.

First, representing the stream using an indirect approach, respondents were asked to write about both a positive and a negative holiday experience. The exemplar scenarios (provided as a guide to the level of information desired) replicated those given in Pearce and Caltabiano’s (1983) study. While the depth of information attained with this method was good, analysis was problematic. Responses were content analyzed with descriptions coded in terms of Maslow’s (1943) hierarchy of needs theory. Initial inter-coder reliability was low with some difficulty apparent in determining which was the dominant theme of the description. Once resolved by a further two judges, it was obvious that the level of data obtained from such a method limited the potential for comparison with other, quantitatively based instruments (discussed later).

A second stream of research has seen tourist motivation researchers ask respondents to rate the importance of various reasons for travel. Shoemaker’s (1989) set of reasons was selected as representative here, partly because it appeared to comprise all major dimensions of motivation, and also because its relative brevity made it appropriate for inclusion in a self-completion mail questionnaire. The only change made to Shoemaker’s list was the omission of the item “to play golf”, as this was felt to be specific to the older pleasure market on which Shoemaker’s study focused. With respondents being asked to circle a number between one and five, interpretation and subsequent analysis of responses was not a concern.

The third instrument included was Cossens, J. 1989 Positioning a Tourist Destination: Queenstown—A Branded Destination? Unpublished dissertation, University of Otago, New Zealand..Cossens’ (1989) list of 16 destination attributes. His scale was developed after a comprehensive review of the destination choice literature and self-completion mail surveys were sent to a similar population. As with the scale measuring the importance of different reasons for traveling, respondents were asked to indicate the importance of each attribute in their decision. From such ratings, underlying motives have then been inferred.

To test the assumption that scales such as Shoemaker’s and Cossens’ are both measuring the same construct (tourism motivation), canonical correlation analysis was employed. If the two scales are substitute measures of the same concept, the canonical redundancy analysis should show a high proportion of the variance in the attributes being explained by the set of canonical variables extracted from the reasons scale and vice versa. The 13 canonical variables extracted from the attributes set of data explained 88% of the original variance. (This is due to a smaller number of canonical variables being extracted because of there only being 13 items in the reasons scale.) What is more important is that the canonical variables extracted from this scale explain only 24% of the variance in the attributes data set. That is, the scale of reasons does not appear to be measuring the same construct as the scale of attribute importance. In a similar fashion, the total variance of the reasons data explained by the canonical variables extracted from the attributes set was 23%. Again the implication is that measuring the importance of various destination attributes is not a substitute for obtaining importance ratings of reasons as to why they go on holiday.

To enable the level of association between the content analysis responses and the other two methods to be tested, cluster analysis was undertaken for both the attributes measure and the reasons (separately). The resulting clusters were then cross-tabulated with the Maslow-based motivational categories and Cramer’s V used as an indicator of association. At R<0.05, the only significant association observed was between the positive experiences categories and the five reasons-based clusters. While significant (P=0.01), the level of association was weak (Cramer’s V =0.17). The only other relationship to near significance was that of the motivations inferred from the negative responses and the reasons for going on holiday. Cramer’s V was again low (0.16 and P=0.09).

Therefore, the results obtained lend support to the proposition that it is wrong to compare results obtained from attribute importance studies with those based on ratings of reasons’ importance. In other words, there is no empirical basis on which to assume they are measuring a similar construct. Additionally, motives inferred from indirect questioning are not strongly associated with the findings of the other two instruments, despite all having been used to measure tourism motivation in the past. While such a finding does not lead one any further towards identifying the “best” way by which to measure tourism motivation, it does sound a note of caution regarding the comparison of results obtained by different methods. If motives are truly to be understood, the search must continue for a theoretically sound, validated instrument.