Armstrong and Overton (1977) suggest several methods for evaluating the
potential impact of non-response on survey findings: (i) researchers’ subjective
estimates of the likely nature of the non-response bias based on the
characteristics of the non-respondents (see e.g. Kirchner and Mousley 1963; and
Vincent 1964 for using this technique based on socio-demographic variables); (ii)
extrapolation, which is based on the assumption that ‘late responders’ are more
like non-responders than ‘early responders’ and then examining differences in
data between early and later responders; and (iii) comparing respondents’
characteristics with known characteristics for the population of interest.
Regarding the use of subjective estimates of the effect of non-response
bias, there is not a great deal of literature in the hospitality leadership area to
guide such a process. In the case of nationality, Testa (2002, 2004, 2009) has
found evidence for greater levels of high-quality leader-member relations between
supervisors and staff with the same nationality. Elsewhere, however, Wong and
Chan (2010) found no evidence for nationality influencing employees’ leadership
perceptions. With respect to respondent age, none of the identified hospitality
leadership articles have included age (neither employee or supervisor) as an
independent variable in their respective analyses.
Owing to the way that this survey was administered it is not possible to
distinguish between early and late responders and therefore the extrapolation
approach for assessing non-response bias is not a feasible option. The method of
administration involved the survey forms being sent to a hotel contact person who
then distributed the forms to the potential respondents. The forms were then
either returned directly to Bournemouth University by respondents, or were
returned firstly to the hotel contact who then returned them to the University.
Because it is not known to the researcher when each respondent received the
survey form, nor when that respondent completed and returned it (was it
returned directly or via the hotel contact?) it is not possible to determine early
from late respondents.
The most efficacious method for determining the characteristics of survey non-
response was to examine the characteristics of the sample and compare these
with known population values. Known (approximate) population values were
identified using the current People 1st’s Industry Profiles for the hotels and
restaurants sector (People 1st 2011a, 2011b) (figures for the waiter category
were obtained directly from People 1st (2011c)). All of these data are derived