Questionnaire Research
The types of information sought when surveying individuals or objects, such as firms, usually include evidence on demographic and socio-economic variables.
In addition, evidence is sought on opinions or beliefs related to behaviors, experiences activities and attitudes.
The philosophical attitude that underpins the use of a questionnaire for the purposes of evidence collection is that there exists a generalisable public opinion that is available to be tested through the use of these sorts of questions.
Questionnaire Design
The point of departure in the design of a questionnaire is a clearly defined problem and explicit terms of reference and objectives.
There should be no ambiguity about the purpose of the study. That is, it is important to be clear about the phenomenon to be described and/or explained and the hypotheses to be tested. Once this has been achieved it will be possible to identify and define the concepts to be measured and how these are to be measured. At this point the first draft of the questionnaire can be designed.
The draft questionnaire is the product of qualitative research. This qualitative component is likely to include a search of the academic, trade and professional literature as well as the use of interviews, brainstorming and focus groups. Internalisation of how others have undertaken questionnaire-based research can be beneficial.
The use of existing questionnaires or questions from them is permissible but it is important to establish where the copyright for these resides.
Pre-testing the Questionnaire
Pre-testing of the questionnaire needs to be undertaken before it is finally administered.
Approaches to pre-testing can be fairly informal where one consults friends, colleagues, experts and people of diverse opinions, or it could be more formal, involving a pilot study which is a replication, on a small scale, of the main study.
Ideally the research candidate should attempt to ensure that the measuring instrument is pre-tested as far as possible.
Types of Questions
Questions can be either open ended or closed ended.
The type of question chosen has implications for the type of evidence that can be obtained and therefore on the method of analysis of the evidence.
Open-ended Questions
Open-ended questions are typically used in exploratory studies where the researcher is not in a position or is not willing to pre-specify the response categories.
The response is in the form of a narrative which has to be analysed qualitatively, but which may be converted into a form suitable for quantitative analysis.
A popular technique for analysing narrative is content analysis.
A disadvantage of open-ended questions is that they require the respondent to be articulate and willing to spend time on giving a full answer to the question.
Questions of this type are typically used in personal interview surveys involving small samples.
Closed – ended Questions
Closed-ended questions are typically used in quantitative studies.
The assumption is that detailed knowledge is available on the attributes of interest and therefore it is possible to pre-specify the categories of response. These can be pre-coded so as to be amenable to computer analysis using statistical packages such as SPSS and SAS, which are two of the more popular systems used to analyse questionnaire- based evidence.
Closed-ended questions are difficult to design but simplify the collection and analysis of evidence making the task of the respondent easier. Such questions are typically used in studies involving large samples.
Measurement Considerations
Questionnaire responses can be quantified by assigning numbers to the responses according to a given set of rules.
This is what is understood by measurement.
While those variables measured at the interval or ratio level are referred to as quantitative variables.
Nominal Scales
These numbers are no more than labels, and no ordering is implied.
Therefore, the only meaningful quantitative analysis that can be performed on such evidence is to determine the frequency (or relative frequency) of occurrence of responses in each of the categories.
Ordinal Scales
Numbers have no meaning and it is meaningful to compute such non-parametric statistics as the median, quartiles and rank correlations.
Interval scales
Pearson’s correlation coefficient.
Rating scales, such as the questions in Part B in Appendix B, are strictly.
However, in practice, especially in the marketing area, these are treated as being measured at the interval level.
In fact, the more categories.
Structure of the Questionnaire
Questionnaires usually comprise sections.
Typically these sections provide information through asking questions of the following types.
Background Questions
Background questions provide demographic and socio-economic information on the individual or firm.
At the individual level these include evidence on age, gender, occupation, income, education level, while at the level of the firm it can include evidence on the industry in which the firm operates, the number of staff employed, their turnover and position in the company. All of part A of the questionnaire in Appendix B collects typical background information.
Attitudinal Questions
Attitudinal questions provide information on the strength of feeling or opinion about objects, issues, activities and interests. For example, one may wish to determine the attitude of respondents towards privatisation, their jobs, management, internal marketing, computer-assisted learning techniques and so on. Attitudes can be measured through the use of single-item rating scales and multiple-items rating scales. Single-item scales are applied when a single question is used to measure the construct of interest whereas multiple-items scales are applied when two or more questions are used to measure the construct. The most popular approach is to use a 5,7 or 9 point rating scale.
Single-item Scale
In practice a minimum of three items is normally required.
Activity and Usage Questions
Activity and usage questions provide information on the extent of involvement in activities such as water sports, radio listening, surfing the Internet and so on.
Sequencing of Questions
It is generally agreed that the best way in which to order the questions is to place general questions first, followed by specific questions and then attitudinal questions. Hard questions should be placed fairly early and interspersed with easy questions. Further, there is a need to ensure that the questions are structured in such a way that the respondent will find it easy to answer questions within a topic, and also not be burdened with questions that are irrelevant to him or her.
Funnel Questions
Funnel questions are used to provide a sequence within a particular topic using a set of questions in which each question is related to the previous question with successive questions having a progressively narrower focus.
Filter Questions
Filter questions are used to exclude the respondent from being asked questions that are irrelevant to him or her.
Questionnaire Administration
Methods for collecting evidence fall into two categories: self-completion and interviews.
Self-completion methods include mailed and computerised questionnaires. Interview methods include personal or face-to-face interviews, and telephone interviews. There are advantages and disadvantages associated with each of the methods.
Mailed Questionnaires
Mailed questionnaires allow one to obtain a large sample with wide coverage, at a relatively low cost. It allows the respondent to complete the questionnaire in his or her own time, thereby ensuring that the responses are free from possible interviewer influence.
It will invariably guarantee anonymity to the respondent and offer incentives such as a copy of the final report. A good covering letter can contribute significantly to increasing the response rate.
Preferably not taking more than 20 minutes to complete. These factors are also important determinants of the response rate.
For a large survey, a response rate of 60 per cent is seen to be exemplary. Response rates as low as 1 per cent have been reported.
Computer-Administered Questionnaires
Disadvantages of this approach include the sample being restricted to users of the network and the complexities of designing and programming the questionnaire, which is likely to require considerable time and money.
Telephone Interview Questionnaires
It is possibly the most used method at present.
Despite most households having telephones, a sizeable number have unlisted numbers- in excess of 60 per cent in some areas. Omission of these from the sampling frame can result in serious response bias error.
This problem has, to a large extent, been overcome by the use of random digit dialing.
It is important that the interviewee be convinced that the approach is genuine and that the interview will be short. The latter requires that the questions be brief, simple and focused. If open-ended questions are used, it is strongly advised that such questions do not exceed fifteen words and that the interviewee not be excepted to give lengthy answers.
For closed-ended questions the number of precoded options presented to the interviewee should not exceed five and these, should in no way be ambiguous.
Response rates for telephone interviews are typically in the range from 35 per cent to 75 per cent.
Personal Interview Questionnaires
It provides an opportunity to probe complex issues in a relaxed atmosphere. Response rates are higher than for other approaches, typically between 50 and 80 per cent.
A Checklist for Using Questionnaires
The following is a step-by-step guide to conducting a survey in a firm.
Determine sample size and sample frame Calculate the response rate and then use this to decide on the required sample size. The mathematics required for this is discussed in Chapter 12.
Produce questions focus groups of six to ten informants to identify the key issues to be addressed by the survey and from these develop a list of appropriated questions.
Questionnaire layout of an appropriate scale, structure, layout, length and wording.
Conduct a pre-test Perform a pre-test using the questionnaire to determine initial responses. If a pilot study is used this may encompass between 50 or so individuals.
Revise the questionnaire Using the results of the pre-test and or pilot study, revise the questionnaire so that it focuses more closely on the key issues.
Collect results It may be possible to discard partially complete questionnaires in favour of complete ones, depending on the response rate achieved.
Edit and code result The results of the questionnaires should be coded appropriately, in order to make analysis and interpretation easier. Sometimes some editing has to be performed.
Analyse and interpret The coded results should be analysed to determine the overall results of the survey. Careful analysis will reveal whether the survey was successful.
No comments:
Post a Comment