How we change what others think, feel, believe and do
What do you need to know?
Before you decide what to ask, you should consider what you really need to know. Many researches contain questions that are thought to be 'interesting' and maybe 'useful' but which do not lead you toward the knowledge you seek.
How should you ask?
There are two main methods of questioning: surveys and interviewing. Surveys allow you to get to a wider audience but lack the communication and flexibility of the 1-1 interview.
Beyond ensuring you have permission to ask and to use the results (suitably anonymized), plus ensuring the person is motivated to answer, the basic principle in each is to ask carefully crafted questions that result in honest and accurate answers.
Will you need to screen out people?
Screening questions check that you have the right people, for example if you are surveying a subset that is found in a larger general population (eg. mothers, computer users, high-earners).
How complex can the questions be?
Sometimes the questions you want to ask are quite complex in nature. Complex questions take longer to ask and also to answer, and may confuse the respondent, leading to false answers. They can also lead to respondents abandoning surveys.
If possible, complex questions should be broken down into a set of smaller, simpler questions, for example following a decision tree.
How specific should the questions be?
Questions can be very specific or very general or anywhere in between.
How do you go about choosing your holiday? (general)
How many hotels do you look at before you decide?
When you ask general questions you are likely to get a wide range of possible answers. If you can handle this, you may well need to ask fewer questions (which is usually desirable).
When you ask specific questions you get data from a limited set of possibilities. This can be particularly useful for quantitative analysis. A dilemma occurs when you need lots of such data and hence need to ask lots of questions. It can also be problematic if your respondents are unable to answer accurately.
In practice a balance of detail is often needed.
What sensitive questions are there?
Sensitive questions can offend and anger, and thus need to be avoided or very carefully worded. Sometimes a screening question may be needed ('Can I ask you about...').
How many questions do you need?
The number of questions should be sufficient to get you the data from which you can draw valid and reliable conclusions. However this number of questions is constrained primarily by time and the goodwill of how long the respondent is willing to invest in something that usually gives them nothing in return other than your undying gratitude.
Also remember that more questions means more work in analysis, so ensure you only ask what you need to ask.
How do you avoid bias?
Bias is very easy to introduce into questions. For example:
How useful do you think this will be?
This makes the person think about about how they will use the item in question, not whether or not the item might be useful at all. Its usefulness is presupposed by the question and applies an assumptive frame.
Do you think this might fail?
Do you think this will fail?
Will this fail?
Will this fail yet again?
People will not necessarily answer questions accurately or truthfully.
Here are additional tips for using questions:
And the big