4.2. Survey development
Given the potential for poor responses that can arise from lengthy and complex surveys, considerable attention was given to refining the visual appearance of the survey. The design was modelled on the manufacturing futures survey developed by De Meyer et al. (1989)which has been tested extensively (Miller et al., 1992). The survey was pilot tested with groups of managers and management accountants, to refine the design and focus the content.
To reduce any misunderstanding that can rise from unfamiliar terminology or ambiguities, the survey asked individuals to consult with others in their organization when this would result in more informed responses. Additionally, respondents were asked to indicate if they did not understand the meaning of any questions in the survey, and to include any additional comments of relevance in their responses. There was no evidence to indicate misunderstanding of the survey items.
The survey was designed to preserve the anonymity of respondents. Surveys were not prenumbered for identification, and respondents were not required to identify themselves or their company. The mailed survey package included an introductory letter explaining the purpose of the research, a copy of the survey, and two postage-paid envelopes—one for returning the survey anonymously, and the second to allow respondents to request a copy of the survey results. A reminder letter was posted three weeks after the initial mail-out. Two months after posting the follow-up letters, preliminary results were sent to all people who requested them.
The two mailings resulted in 78 usable responses, or a response rate of 56%. Eight firms returned the surveys unanswered indicating that it was company policy not to respond to voluntary surveys. To examine for non-response bias, the responses from the first 20% of returns and those from the last 20% (which would have included mainly respondents to the second mailing) were compared, to test if responses differed between the two groups. Levels of significance were determined for each item using t-tests. No differences were identified, providing some support for the absence of a non-response bias.
The possibility exists that respondents’ functional background may introduce bias. Table 2 reports that 86% of respondents were from accounting and 14% were non-accounting. An examination of the mean responses, between these groups, for variables listed inTable 3, did not reveal any significant differences.
4.2. Survey developmentGiven the potential for poor responses that can arise from lengthy and complex surveys, considerable attention was given to refining the visual appearance of the survey. The design was modelled on the manufacturing futures survey developed by De Meyer et al. (1989)which has been tested extensively (Miller et al., 1992). The survey was pilot tested with groups of managers and management accountants, to refine the design and focus the content.To reduce any misunderstanding that can rise from unfamiliar terminology or ambiguities, the survey asked individuals to consult with others in their organization when this would result in more informed responses. Additionally, respondents were asked to indicate if they did not understand the meaning of any questions in the survey, and to include any additional comments of relevance in their responses. There was no evidence to indicate misunderstanding of the survey items.The survey was designed to preserve the anonymity of respondents. Surveys were not prenumbered for identification, and respondents were not required to identify themselves or their company. The mailed survey package included an introductory letter explaining the purpose of the research, a copy of the survey, and two postage-paid envelopes—one for returning the survey anonymously, and the second to allow respondents to request a copy of the survey results. A reminder letter was posted three weeks after the initial mail-out. Two months after posting the follow-up letters, preliminary results were sent to all people who requested them.
The two mailings resulted in 78 usable responses, or a response rate of 56%. Eight firms returned the surveys unanswered indicating that it was company policy not to respond to voluntary surveys. To examine for non-response bias, the responses from the first 20% of returns and those from the last 20% (which would have included mainly respondents to the second mailing) were compared, to test if responses differed between the two groups. Levels of significance were determined for each item using t-tests. No differences were identified, providing some support for the absence of a non-response bias.
The possibility exists that respondents’ functional background may introduce bias. Table 2 reports that 86% of respondents were from accounting and 14% were non-accounting. An examination of the mean responses, between these groups, for variables listed inTable 3, did not reveal any significant differences.
การแปล กรุณารอสักครู่..
