It was not possible to undertake non-response bias tests relating to comparing demographic data for respondents
and non-respondents as no data were available at the business unit level. Respondents provided details
relating to their business units (e.g. annual sales turnover, number of employees, types of business activities
undertaken) but this frequently differed from the information contained for the company as a whole
in the FAME database. Thus, it was inappropriate to compare respondents demographic data (based on
business unit responses) with those of non-respondents (based on the company as a whole recorded in
the FAME database).
The initial mailing of the questionnaire resulted in 83 usable responses and the first and second
reminders respectively resulted in 46 and 18 usable responses. Thus, after the second reminder stage
a total of 147 usable responses were received. The analysis of the respondents indicated that 22 of the
responding organizations (15%) were ABC users. Given that the research sought to examine the influence
of contextual factors on ABC adoption and sophisticated costing systems it was considered that the aims
of the research could not be achieved based on responses from only 22 ABC adopters. 10
In order to increase the number of respondents using sophisticated costing systems we obtained the
assistance of a UK firm of ABC consultants. They agreed to mail the questionnaire to their clients and
this process resulted in the receipt of a further 29 responses from ABC adopters. All of the 29 respondents
were included in the FAME database11.
The responses of the 22 ABC adopters obtained from the initial, first and second reminder mailings
were compared with the 29 ABC adopters obtained from the ABC consultants for all of the variables
used in this research. There were no significant differences in the responses between these two groups
of ABC adopters. A non-response bias test, based on the assumption that later respondents more closely
resemble non-respondents was undertaken, by comparing the responses of the initial mailing (N= 83)
with the responses from the reminder mailings (N= 64) in respect of all of the dependent and independent
variables used in the research.12 There was no evidence of non-response bias.