CHAPTER 4 METHODOLOGY 101
4.4 RESEARCH INSTRUMENTS 105
4.4.2 DESIGN OF THE QUESTIONNAIRE 107
There are eight steps in designing a questionnaire as explained by Bissett (1994). First, the decision on what data are needed is important, for the survey questions need to facilitate the collection of the required data. This depends on the objectives of the study and the ultimate outcomes visualised. Based on this first step, the exact items of required information need to be drawn up in the second step. The third step consists of designing the questions itself. The format depends on the method used: postal, telephone, direct or any other. Open questions are desirable for qualitative work. Open questions can be useful to determine the types of data to be collected in step 1. Closed questions can elicit dichotomous response (yes/no), rating on a given scale or other forms. The choice of the question and the scale depends on whether
the variable is categorical or continuous. Forcing responses into categorical types should be avoided as far as possible. Likert scale can be used for rated answers. The next step is to compose words with brevity. Precise, simple, non-technical language is preferable, especially for a wide audience. Each question should deal with only a single idea, without being leading or biased. To design the layout and the presentation in the next step, conversational tone is suggested.
An introduction to the purpose of the study, involved organisations and an individual confidentiality guarantee are required. The word “questionnaire” should be avoided and “form” is more acceptable. Arrange the order of items and questions to ensure that the respondent is kept interested and not bored. Clear print and colour are important. At the end, the respondent should be thanked for spending valuable time to complete the questionnaire.
It is preferable to think about coding in advance, although it is not always possible to predict the complete range of answers. Preparing a first draft and pre-testing among a close circle and again with experts before piloting is the next desirable step. This is followed by piloting and evaluation. These points are discussed in detail elsewhere. The actual survey is ready to be started only now. Many other authors have also given similar ideas about designing a questionnaire.
These eight steps were applied in this study. In this study, the appropriateness was tested by discussing with experts and conducting a pilot study. The questions should be intelligible and easily understandable, for which it is better to use the language of the respondents. The questions were framed in simple English with clear response scales. An Arabic translation was done with the help of an approved official translator. Either version could be used by respondents as convenient to them.
The questions should be unambiguous in that they should mean the same thing to the researcher and the respondent. The discussions with experts, the pilot study and translation into Arabic facilitated the removal of these problems.
The questions should be unbiased for the respondent. The questions may look quite unbiased to the researcher, but the way they have been framed, preference for one specific type of response may be directly or indirectly indicated. This will defeat the very purpose of the research. An equal chance for any type of response needs to be
ensured. The questions were framed to ensure that there is no indication of a specific response bias.
If the answer depends on the memory of the respondent, there can be a recall bias, meaning only certain palatable events are always recalled in preference to unpleasant ones. This problem can occur only if MICE participants were surveyed for post-MICE experience after the lapse of a certain period. But in this study, the MICE experience was evaluated immediately during the MICE event, at the venue itself. Omni-competency of the questions also needs to be ensured. That is, the question should be capable of coping with any type of response. This can be achieved by inclusion of options like “Other” or “any other” or a section for comments may enhance the chances of omni-competency. However, this also may provide for non- answers whereby the question is avoided. This survey used a Likert scale of response that ensured an answer, while the extent of the range of questions ensured a comprehensive set of questions.
There should not be any ambiguity or overlap in the coding system. This needs to be checked and ensured. The categories must be exhaustive and mutually exclusive. It is better that the answers are self-coding. In this study, the coding was checked for their clarity. Some questions were self-coding by nature as in the case of demographic details, but the majority of questions used a Likert scale to meet this requirement.
Finally and very importantly, ethical requirements must be met. Many organisations insist on this for all research work done by their staff and they have clear procedures to ensure this. Ethical problems occur when the questions contain invasive, potentially hazardous or security or privacy elements. Both confidentiality of the identity of the respondents and the data gathered from them need to be protected from all types of threats using suitable procedures. Usually, these are guaranteed when requesting for participation in a survey. This study was ethically cleared by a competent regulatory authority.
The importance of cognitive aspects in the pre-test stage of questionnaire design was highlighted by Willis (2004) and (2008). Errors in writing the questionnaire is one of the many possible non-sampling errors. The authors stressed asking the right questions, properly phrased and correctly ordered. Whatever be the media of
questionnaire administration, the importance of its proper construction cannot be underestimated. The questionnaire provides a standardised interview across different participants. Asking the same questions differently to different people can lead to responses which are difficult to interpret, especially in the case of large samples. All the respondents should be able to read the question for the same intended meaning.
Instead of viewing respondents as mere sources of information, they need to be considered as living persons. Long questions, long sentences, difficult phrases and long, complex and boring questionnaires should be avoided. Familiarity with the context is essential to correctly word the questions. Thus, a questionnaire needs to be written in such a way that the required data are collected to answer the research questions as objectively and completely as possible without irritating the respondents with minimum risk of errors. In this study, the basic structure and variables of study were determined based on other works discussed in the previous Literature Review and then adapted to the study context. The design of questions was determined by reported findings in the literature review and the specific context of this study as already discussed at several places above.