Choice sets are a central component of high involvement destination selection models. A taxonomy of vacation destination choice sets was developed by Crompton (1992) and a set of propositions relating to choice sets was offered by Crompton and Ankomah (1993). This research note reports on analyses that tested five of the propositions or corollaries of them. The study was conducted at Sun/Lost City in South Africa, which is the biggest entertainment theme park of the continent. Data were collected using personal interviews from a sample of 302 respondents. In an effort to ensure that only highly-involved tourists were sampled, visitors from Gauteng and the North-West Province, which are the proximate areas to Sun City, were excluded.

The early consideration set consists of those places considered by a potential tourist to be possible destinations for a vacation. In this study, this was operationalized by asking respondents, Think back to when you first thought about going on a holiday at this time. Back at the beginning, to which other places, apart from Sun/Lost City did you think of going? When the respondents finished giving their list of places, they were prompted further by asking, Is that all? Were there any other places that came to mind when you first thought about going on vacation? If more were named they were added to the list. The late consideration set was derived by the interviewer asking, Among the places you named that were destinations you initially considered, which of them were close to being selected before you made your final decision to come to Sun/Lost City?

Proposition 1 : Respondents preference rankings of vacation destinations will correspond positively with the order in which they are mentioned in unaided recall questions.

Respondents were given a 20 item instrument that measured destination attributes which were prefaced by the instruction, My impression is that (name of destination) has . . . The first destination respondents named in their late consideration set was inserted in the space. After completing it, they were handed the same set of items with the second destination name they gave in their late consideration set list inserted. There were no significant differences (0.05 level) in the responses to the two destinations on any of the 20 attribute items. Hence, this proposition was rejected.

Proposition 2: The average number of destinations that individuals will seriously consider in their late consideration set when making vacation decisions will not exceed four.

Proposition 3: Perceived importance and/or perceived risk of the destination decision will have no effect on the size of the late consideration set.

Perceived importance and perceived risk were operationalized with items taken from the Laurent-Kapfener (1985)Involvement Scale. The two-item importance scale had a Cronbach alpha reliability of 0.67, while the three-item risk probability scale alpha was 0.74. The average late consideration set size of respondents whose scores were above the median on the perceived importance scale was 3.6, while the average set size of those scoring below the median was 3.4. A t-test indicated this difference was significant at the 0.01 level. However, when a similar procedure was undertaken on the risk probability scale, contrary results emerged. Respondents perceiving low risk had an average set size of 3.7, while the mean set size of those reporting high risk was 3.4 (significant at the 0.001 level). The inconsistency of these findings, appears to reinforce the conclusion of Crompton and Ankomah (1993) that empirical studies have not shown any consistent statistical relationship between perceived importance and perceived risk, with size of the late consideration set.

Proposition 4: The ratio of size of late consideration set to size of early consideration set is likely to be between 0.6 and 0.9.

The means for the two sets were 4.02 and 3.54, respectively. Hence, the ratio was 0.882 and the proposition was supported.

Proposition 5: A destination in which a potential tourist has invested more effort in seeking information has a greater probability of being in the late consideration set and of being selected as the final choice, than destinations for which the invested information effort was lower.

After respondents had listed the destinations that constituted their early consideration set, they were asked, How much effort did you invest in actively seeking out information about (name of destination). Was it no effort, very little effort, little effort, reasonable effort, quite a bit of effort, great effort, or a very great effort? They were shown successive sheets with this seven-point scale and each of the destinations they had listed in the early consideration set was inserted on a separate sheet.

The mean search effort score for Sun/Lost City, which all the respondents ultimately selected, was 2.75, while the average of that invested in all the other destinations given in the early consideration set was 3.66. Thus, contrary to the proposition, results indicated that the search effort for competitive destinations that were not selected in the early consideration choice set was significantly (0.01 level) more extensive than it was for Sun/Lost City. This suggests that Sun/Lost City visitors were already familiar with the destination, and because it was so well known to them no additional information was needed. However, a significantly (0.001 level) greater amount of effort was invested in acquiring information about destinations included in the late consideration set (mean 3.81) that for those which appeared only in the early consideration set (mean 3.30). This finding was consistent with the proposition.