The Scottish Government

« Previous | Contents | Next »

Listen

SMALL BUSINESS SERVICE NATIONAL OMNIBUS SURVEY SCOTLAND: AUTUMN 2002

CHAPTER THREE: METHODOLOGY

3.1 SAMPLE FRAME

Respondents were selected at random from a sample that in turn had been randomly selected from the D&B database of enterprises in Scotland. SIC code and proprietorship were then used to split the sample in the same proportions as the population of small businesses in Scotland ¹. The sample frame contained 4,000 organisations and was structured as shown in Appendix 2.

The sample provided by D&B was flagged on their database so that the organisations could be excluded from future surveys for the Scottish Executive.

The Dun and Bradstreet database is thought to provide a representative sample of businesses in the micro, small and medium size groups. However, the sample of sole traders is likely to have an overrepresentation of more established firms and those selling to the public.

3.2 QUESTIONNAIRE

The survey was carried out by telephone and the questionnaire was completed by the interviewer using CATI software. The questionnaire is shown in Appendix 1. The interviews took, on average, 20 minutes.

The order of multiple choice prompted answers and multiple question sets was rotated each time the questionnaire was completed to avoid any bias due to the order of questions.

3.3 RESPONSE RATE FOR THE SURVEY

In total 1,756 organisations were contacted and asked to participate:

• 1,002 completed interviews
• 224 refused to participate - a refusal rate of 14% (2001 - 18%)
• 181 were out of size band quota
• 126 were not SMEs
• 88 were no longer in business
• 135 were appointments: respondents who agreed to participate and made an appointment for an interview but were not needed to meet quotas.

3.4 SIZE OF BUSINESS

The sizes of the businesses that participated in the survey are shown below:

Table 1 Respondents by business size

3.5 WEIGHTING

The results for each of the four size categories have been weighted on the basis of the number of organisations in the category. The source for the estimate of the population is businesses registered on the IDBR in 2001. The weighting factors used are:

Table 2 Weighting factors

The responses were analysed to establish both qualitative and quantitative data. The quantitative analysis was carried out using SPSS 11.0 - this is specialist software that is particularly effective for analysing survey data. The use of SPSS ensures that all analysis will be documented, transparent and repeatable.

The data was analysed to provide statistical results for each question. The use of syntaxes allows question responses to be analysed against each other where this is felt to add to an understanding of the objectives. Quantitative data and analysis is reported in the text of the report and through the use of tables and charts.

All the results are analysed by industry sector and size of the business, businesses with zero employees have been removed from the results except where results are specifically analysed by size of the business. As they are by far the most numerous category their inclusion would dominate any overall results.

3.6 QUALITATIVE DATA

Verbatim responses are analysed using Voxco Interviewer - this is a specialist interviewing package that allows complex questionnaires to be designed and used. This anecdotal information is reported in summary format.

Some qualitative data have been used in this report in the form of comments cited from respondents to the survey. On each occasion, these comments are selected as being of particular interest or as reflecting a general viewpoint expressed by a number of respondents in the survey. They should not be taken as representative of all responses to the relevant question in the survey.

3.7 CONFIDENCE IN THE RESULTS

The confidence interval for the results in this report describes the probability that the result lies within a particular band. Normally 95% confidence intervals are reported - this is the band within which there is a 95% probability that the true result lies. Confidence intervals are a function of the size of the sample for a question. The following indicative confidence intervals should be borne in mind when reading this report:

• For a sample of 1,000 there is a 95% probability that the true result is within 3.2% of the survey result.
• For a sample of 400, such as for larger industry sectors, there is a 95% probability that the true result is within 5% of the survey result.
• For a sample of 100, such as for smaller industry sectors, there is a 95% probability that the true result is within 10% of the survey result.

The exact number of responses to a question from the survey is indicated with n. The number for n is not always the same as not all respondents answered all questions. Some questions were missed in some interviews due to time constraints on the part of the respondent. These questions were between 187 and 245 inclusive. 29 of the 1,002 interviewees missed some questions (although not necessarily all of those between 187 and 245).

3.8 CHARACTERISTICS OF RESPONDENTS ²

3.8.1 Gender

The gender of the respondents that participated in the survey is shown below:

Table 3 Respondents by gender

3.8.2 Ethnic group

3% (n=35) of respondents considered themselves to be members of an ethnic minority group. Seventeen of these 35 regarded being from outside Scotland a reason to feel members of an ethnic minority group (Greek, Italian, Armenian).

3.8.3 Age

14% of respondents were between 25 and 35 years old. 29% of respondents were between 36 and 45 years old, and 46% were between 46 and 60. 1% (n=10) of the respondents were younger than 25 and 10% (n=99) were over 60.

3.8.4 Agreement to participate in further research

86% (n=863) of respondents would be willing to participate in further research for the Scottish Executive. More than half of these (n=585) use the Internet.

3.8.5 Business activity

The activity of businesses that participated in the survey is shown below:

Table 4 Respondents by industry sector

The proportions for business activity of respondents for the UK as a whole are similar to those in Scotland.

« Previous | Contents | Next »