Questionnaires
Various forms of questionnaires exist and the style of questions normally dictates the type of research that emerges. You may need to construct one later in your course if you do a piece of coursework.
Questionnaires can be conducted in a number of ways:
- Face-to-face - a researcher asks the questions. This allows the purpose to be fully explained and any problems in completing the questions addressed
- Self-completion - These often used at the end of a service. For instance, patients in a hospital may be asked about their treatment when they are discharged.
- Postal - these may follow the purchase of a product or a service provided in the home. It asks for feedback. Return rates are generally very low and the questionnaire often ends up in the bin, so firms often try to provide an incentive for completion, such as entry into a free draw, a discount on future purchases of some kind of voucher.
- Telephone - this is similar to a face-to-face questionnaire, except of course that the intended respondent can hang up!
- Internet - it is increasing practice for pop-ups to appear after placing an order or accessing particular websites.
Questionnaires can be structuredin that they rely mainly on closed questions with an occasional open, expanding style question or unstructured, which use more open questions.
Closed questions.
Closed questions have a limited number of preset answers, the most common of which is, 'Yes' or 'No'.
Questionnaires based on closed questions are popular as they are easy to quantify and turn into visual representations, such as graphs, which are easy to interpret. Structured questionnaires are easier and quicker to complete, but may not provide the same quality of information that unstructured questionnaires do, because they limit the responses and demand a specific answer. It is difficult to establish 'shades' of feeling. However, although opinions may be harder to detect, it is possible to record the strength of opinions by using techniques such as Likert scale responses. This provides a range of responses such as:
- Very good
- Good
- Satisfactory
- Poor
- Very poor
However, an open or unstructured approach will allow for a far greater range of response. Researchers can dig more deeply into issues and establish just why a person holds a particular opinion:
Closed question:
Do you buy this product? Yes or No √
Open question:
If 'No', please give your reasons:
However, the more open the questions the more difficult it becomes to classify responses. The results will become more subjective and therefore less easy to actually put into categories. Just imagine asking your friends what they wanted for school lunches. The response could be so varied and contain so many personal likes that the canteen could never produce such a range. Hence, in most school canteens the range is based on certain well-tried favourites.
Design of questionnaires
Before questionnaires are conducted, it is good practice to test a 'pilot' version. It is surprisingly difficult to prepare a questionnaire that is fully understood and collects the data required. This pilot allows for problems to be identified and questions removed or rephrased.
The design of a questionnaire is subject to some general principles:
- What are the key objectives - what information is required?
- Questions should be easy to understand and unambiguous and avoid industry jargon.
- Avoid asking two questions in one, e.g. 'do you own a car and is it a diesel?'
- Avoid 'leading' questions that encourage respondents into certain types of answer.
- Follow a logical sequence of questioning and limit the numbers of questions (a maximum of 15 is usually recommended in face-to-face questionnaires). Ask yourself how much time you would be willing to spend on a Saturday afternoon answering a a survey when you wanted to shop.
- Don't rely too much on respondents' memories.
- If offering a choice of responses, don't make one of them the obvious answer.
- Take care not to ask questions that are thought to be offensive.
- Keep to what the respondent knows.
- Put some 'control questions' in the design. These questions check for bias in both the way the researcher is asking the question and the answers they are receiving.
- Use a balance of open and closed questions.