Before starting on the steps for designing surveys, it is important to mention, that they differ from questionnaires.
Questionnaires are not used for statistical purposes, whether surveys come to conclusion about a certain group of people.
A questionnaire might be sent to suppliers to identify which ones need stricter controls. There is no intention here to quantify or analyse the data.
Surveys, are a common way of gathering data for research study. Designing it is a difficult task and requires skill and knowledge. You can only start designing surveys if you have already done the preliminary research planning, set objectives of your study and defined the research questions. The survey will help you to achieve the objectives and answers.
Whatever the purpose, you will need to have a well-defined and structured survey. It needs to give you the information for analysis later. Therefore, the survey questions will need to be design the way, so that it is easy to categorise and obtain information.
Surveys can be made using Word document, or online tools i.e Survey Monkey or Google forms. Surveys are not easy to design. It requires more than common sense.
We need to write accurate questions, trial run, distribute and manage the returns. You must find the answer possible to deal with when they return.
The questions
Your research plan will identify important areas for investigation. Review the research objectives and decide which questions will meet the objectives. You may need several attempts for wording of each question. Remove ambiguity, make them clear and easy to understand. Ensure respondents can understand what you are asking, the questions are free of jargon and decide the types of questions you need to ask.
If your project, for example, requires identifying the number of respondents concerned with particular matter, you may need them to answer "Yes" or "No" questions. Then ask followup questions to get more information.
There are different types of questions you may ask, and each one of them has its advantages and limitations.
Open questions - we invite the respondent to comment. These type of responses can produce useful information, but analysis can prove difficult. In some cases, content analysis might be required, but well structured open questions can be useful.
List - we offer the respondent a list of item to choose. For example, we ask for qualifications and list them to select.
Category - we ask respondent to choose category options. For example, age categories.
Ranking - we invite respondent to put something in order of frequency or importance.
Quantity - the response is a number (exact or approximate).
Table or grid - we invite respondents to answer more than one questions at the same time.
Scale - various types of scales are offered.
When designing our questions we should ensure that they are clear and easy to understand. Words can mean different things to different people. We need the questions to be understood by all participants in the same way. The questions need to be precise and unambiguous. Sometimes it might be worth to ask two separate questions for more clarity and to ensure we obtain the results we need. Precision of wording is important.
We must remember, that "satisfaction" cannot be observed and we must find ways to measure it.
If respondents are upset or offended they may abort the survey. We should not make assumptions about the respondents and allow them to answer the questions as they want. We should also be mindful of questions where respondents are asked to remember things. Our memories can play tricks and it might be frustrating for participants to look for information elsewhere to answer the survey questions.
We also need to consider their knowledge. We wouldn't want to ask new employees about something they have had no chance to discover.
Double questions should also never be asked. For example, asking a participant whether they like both service and cleanliness in a hotel. This would be rather misleading. It is easier to ask those separately for service and cleanliness.
We need to avoid leading questions to avoid influencing respondents to answer questions a certain way.
Offensive and sensitive questions need thorough consideration. If any question may cause offence, they should be removed. If there is a need for asking sensitive questions, it might be worth adding them towards the end of the survey. If the respondent abandon the survey at this point, we may still be able to retrieve answers from previous questions. Age is often considered as a sensitive question. It might be better, therefore to indicate age category.
The layout
A well-presented survey must look tidy. Recipients should be encouraged to read and answer questions as poorly designed documents might put them off. There are some ground rules when it comes to the layout:
Instructions must be clear ( capitals and different fonts),
Add spacing between questions,
Keep response boxes towards the right side of the document as it is easier to record responses and analyse them,
Allow space on the right side for coding,
Evaluate your survey critically before distributing,
Take care over the order of questions. Leave sensitive questions to the last. Start with easy questions,
Promise anonymity and confidentiality,
Do not ask for names.
The sample
The number of respondents for questions will highly depend on the time you have. You will need to find out how many respondents in total are affected. It is often unlikely that you will be able to ask all to complete your survey.
Different techniques can be used to produce samples. The sample needs to be, as far as possible, the best representation of the whole research population. The make-ups of samples are important to be explained in the research report. For example, you may want to include 50% of your study population. A random sample will give each individual an equal chance to respond. But then, everyone randomly selected might not be able to participate. It would be wise to have reserve names specified. For example, the 10th person refused and we ask the 11th one.
We might also need to select a representatives of groups of population, i.e appropriate proportion of men and woman. Thus, you may include every second man and every second woman in your sample.
The trial
You need to test the survey before distributing to all recipients. We must check that all the questions and instructions are clear. The pilot run can be given to family or friends or the closest person who might best represent the study participants. The trial will indicate on how long it takes to answer the questions and clarify any issues.
You may ask the trial respondents the following questions:
How long does it take to complete the survey?
Are the instructions clear?
Are the questions easy to understand?
Are any questions offensive or sensitive?
Any important topics omitted?
Is the layout clear and encouraging?
The distribution
We need to ensure that we obtain necessary consent before distributing the survey. Written approval should always be obtained.
Once approvals are in place, you will need to decide how to distribute the survey and what to do with non-responses. There are certainly advantages of distributing the surveys personally. One of them is that the surveys can be completed immediately. You are able to establish personal contact with the respondents and explain the study.
If this is not possible, the most common way used is e-mails.
The rights and responsibilities
Respondents should be provided with a written statements of rights and responsibilities as well as the purpose of the research. You need to make it clear that an official approval has been granted. You need to explain on where the information about surveys will be used, how they will be stored and who will have access to them. You need to secure the compliance with GDPR.
To do this, you may require a letter that will be sent with your survey.
When sending the survey it is useful to give the return date. 2 weeks is generally sufficient. Give the precise date.
Once the surveys are distributed try and keep the record on when these were sent and returned. Some effort should also be made to encourage the responses. A good idea might be a follow-up request a week before the deadline. For some large projects follow-up reminders are often sent 4 to 5 times.
Summary
Surveys require skill and commitment. A well-designed survey can help you to obtain answers and meet the project objectives. Below is the summary of important steps to be taken. This is to ensure you can benefit from it to the highest potential.
Ensure that you have gained approval to proceed with the survey. Check requirements for ethics and obtain written confirmations. This might take some time, but its better to know where we stand. Allow extra time for this.
Review your research plan and decide on what you need to know. List all the information you require. Try not to include irrelevant information.
Ensure a survey is the best way to obtain the information. Other methods might be more appropriate.
If survey is the best, begin to word questions. You might want to write them on separate cards and order them later.
Check the wording of each question. Remove any assumptions or ambiguity. Keep language simple. Do not use words that people might not understand. Unless you are dealing with a professional group, do not use technical language.
Decide on the types of questions. Should these be open questions, a list, category, scale or grid?
Order the questions once you are satisfied with them. Leave difficult topic and sensitive matters until the end.
Write down instructions to the survey. You need to be clear on how you want respondents to answer the questions.
Decide if you need space for coding on the survey.
Review the survey appearance and layout. Use design tools that make it look attractive.
Consider your sample. Decide on a sample that is close to your final study population. Explain in the report the choice of sample.
Trial your survey. Ideally send it to people similar to the sample population.
Try out different methods of analysis from the pilot run.
Make necessary adjustments based on the trial run comments. Consider on whether the survey does not take too long to complete. Eliminate any irrelevant questions.
Decide on how the survey would be distributed. Ensure consent is obtained and assure confidentiality and anonymity.
Define the survey deadline. Keep records on when surveys were distributed and returned.
Consider on what you are going to do with non-respondents. Send reminders to all respondents to guarantee anonymity.
Begin to record responses as soon as these are returned.
Comments