Setting aside those hotels that did not return any survey forms, and therefore,
although indicating that they would, effectively did not participate in the research:
618 survey forms were distributed;
234 survey forms were returned;
o representing a response rate of 37.9 per cent^9
The hospitality leadership articles that had been identified for this research were
examined to identify typical response rates for postal surveys within organisations
in this area of social science. Eight studies published during the previous 10
years that had employed a postally-administered survey were identified (Chiang
and Jang 2008; Kozak and Uca 2008; Clark et al. 2009; Patiar and Mia 2009;
Asree et al. 2010; Gill et al. 2010; Kim, B. et al. 2010; Zopiatis and Constanti
2010). The response rates for these studies ranged from 19 to 66 per cent with a
mean response rate of 40 per cent. The response rate achieved for this survey
(37.9 per cent) is close to this average.
Assessment of the sample
The 59 hotels that agreed to participate had the following star ratings: (i) three
star n = 12; (ii) four star n = 36; (iii) five star n = 5; and (iv) no star rating
identified n = 6. Of the 27 that returned survey forms, six (22.2 per cent) had a
three star rating, nineteen (70.4 per cent) had a four star rating and two (7.4 per
cent) did not mention a star rating on their web sites. None of the five star rated
hotels (n = 5) that were contacted ultimately participated in the survey.
To assess whether or not star rating had any effect on the propensity of hotels to
participate, contingency tables were constructed - by cross-classifying hotels’ star
ratings and participation status (did / did not participate) - to test for the
independence of the two variables using the chi-squared statistic. Using the three
star rating categories (3, 4 and 5 star) prevented a meaningful analysis since the
low number of 5 star hotels created too many cells in the contingency table (>20
per cent) with expected values less than 5 (thus violating one of the assumptions
of the contingency table chi square test). To counter this issue, two alternative
contingency tables were constructed. In the first, the 5 star hotels (none of which
participated in the research) were removed from the analysis resulting in a finding
9
Owing to the way in which the pilot survey was administered it was not possible to
calculate a similar figure for the pilot.