Whether we like it or not, surveys have become a staple of modernity. Be it that they are customer satisfaction surveys, be it that they are census, opinion polls, household surveys, attitude surveys or surveys for research purposes.
Typically surveys are used to:
- get feedback from "audience"
- collect information/statistics about an "audience" group
- understand the "audience" needs, challenges and opportunities in an effort to make informed decisions
- assess the impact of an intervention and/or activity on an "audience"
Should you need to design a survey, you may wish to take a few minutes and consider the following:
- What's purpose of the survey? Why are you designing and launching it?
- What type of data/information and feedback you wish to collect?
- Who is your audience? (demographics, social and economic status, occupation, literacy rate)
- Why is it important to reach out to your audience?
- How will you be using the results, outcome and data from your survey?
- What is your plan to share the survey results with your audience?
- Is the survey a one-off, a follow-up to a previous survey, part of a research project?
- Are you undertaking the survey on-behalf of a third party?
Survey goal and purpose (the why)
Once you have clearly identified and articulated the goal and purpose of your survey, make sure you write it down, because this is what you need to use as your survey's introductory text! Include the survey deadline along with how and when you'll be sharing the survey results in your introductory text.
Know your audience (the who)
Surveys are a communication tool, as such it is important to know who is the target audience. Here are some points to consider:
- know who you wish to reach, as this will determine the questions you will be asking, the timing and format in which you will be sharing the survey
- provide clear instructions, including what you will be doing with the results and how long it will take them to complete it. Be honest, if it takes 20 minutes to complete the survey, say so, as otherwise you may risk putting off your audience and end up having fragmented and unusable data
- make sure your survey is tailored to the literacy level and language group of your audience
- if you opt for on-line survey, make sure your audience has access to the appropriate technology
The questions (the what)
Once you've figured out why you are doing the survey and who is your target audience, you need to compile your questions. One overarching tip for formulating powerful questions, is to KNOW what you will do with the responses. If you cannot figure out how you will use the response of your question, either reformulate the question, or opt to drop it. You may find the following suggestions useful:
- craft clear and concise questions (preferably in plain English and jargon free)
- ask one question at a time. Do not stack your question and do not AND/OR in your question
- use multiple choice, true/false, checklist and rating scale as these make your compilation task easier and you do not run the risk of having to interpret the response
- use even number for rating scale type questions (for example: 1-4 where 1 is poor and 4 is excellent). This way you will encourage the respondent to provide a meaningful answer as opposed to settling for the middle ground. This said, where appropriate provide N/A (not applicable) option. Make sure you assign numeric values to your rating/scale questions. This will facilitate the compilation
- where applicable and appropriate consider asking the respondent to provide the occurrence of an activity as opposed to simply asking them to give you an approximation such as never, seldom, often, always
- keep open-ended questions to a minimum (structured questions make compilation work easier and you do not risk falling into the 'interpretation' trap)
- group questions logically and if appropriate breakup your survey in logical sections
- figure out which questions are of utmost importance - for which you require an answer or else your survey will be void - and make those mandatory
- include demographics (such as gender, age etc) so that you can disaggregate the results. If your survey is anonymous, make sure the demographic questions are in-line with your anonymity framework
- use validation questions as appropriate
Format (the how)
Once you know who is your audience, you will be able to decide whether to opt for electronic, print or interview format (in person, telephone). In determining the format, consider the following:
- access to technology
- literacy rate. In case of low literacy rate, you may opt for a pictorial version of the survey or conduct a face-to-face or phone interview
- respondent's cultural context and make sure you are gender sensitive
If you opt for on-line or print format (mailed or manually distributed), remember the eye wants it share as well. Make sure your survey is well-formatted and visually appealing. In case of print survey, allow enough space between questions and allocate adequate space for response to open-ended questions.
Timing (the when)
The timing of a survey can contribute to higher response rate. Knowing your audience will help you decide when is the best time to launch your survey. For example, if you were to survey farmers, you would try to avoid peak harvest time, as you know they will be busy in the fields and have other priorities.
I am adding the survey deadline under this heading. Decide how long you'll be running your survey. Seven to10 days seems to be the norm. Send a reminder four and two days before the survey's deadline.
Put yourself in the shoes of the respondent and think of the frustration in completing a survey that does not work, or has unclear questions.
This is why it is really important to field test your survey before launching it. By field testing, I am not just talking about making sure the technical and mechanical part works. More importantly, the field test is to assess if all your questions are clear, make sense and relevant to your audience.
For field testing, choose people who were not involved in the design process. If you can afford the luxury of having someone from your audience group, go for it and have them complete the survey. That would be the best litmus test.
While it makes total sense to aim for 100% response rate, conventional wisdom says that average response rate for on-line surveys is 30-40% and 60-70% for mailed ones.
You would hope that all your respondents have duly completed the survey. This is why it is important to decide which are your mandatory questions so that you avoid the risk of getting partial responses which could jeopardize the validity of your efforts.
Read carefully the answers to the open-ended questions. To the best of your ability, try to stay as objective as possible. This is why it is best to keep these types of questions to the bare minimum.
Once you've compiled the results and you get a good understanding of what the results are telling you, share it with your respondents. You may do so in a narrative form complementing it with graphs and charts.
In sharing the results, depending on the type of survey, let your respondents who were diligent enough to complete the survey know how and when you'll be taking action.
While I know the above is far from being comprehensive. Nonetheless, I hope you find it useful. I encourage you to also check out The University of Wisconsin Survey Center for valuable resources and guidance on how to design and implement surveys.