Researchers are often interested in asking questions that require respondents to think about time. Sometimes these questions involve counting the number of events in a specific time period, or asking respondents to report their feelings at a certain point in time in the past. However, these types of questions are prone to error from poor respondent recall. Sometimes respondents have no memory of specific events or are unable to piece the memories they do have together clearly. It sometimes helps respondent retrieve information from memory when questions clearly frame a specific time period of interest. This approach can also help ensure that all respondents are thinking about the same time period when answering the question.
Provide time references to improve recall.
In the last year, how many times have you visited a primary care physician, meaning a doctor you see for general medical issues and check-ups?
This question is problematic. What does the last year mean? Some respondents might interpret this as meaning the current calendar year, so only since January. If the interview is in August, then the respondent is only thinking about eight months. Other respondents might interpret it as meaning a full twelve month period. All respondents should have the same time frame in mind when thinking about their responses. Providing a clear time frame in the question can help standardize the time frame for respondents and can even aide the recall process for some respondents. Consider instead one of the following two alternatives:
In the last year, meaning since August 1, 2009, how many times have you visited a primary care physician, meaning a doctor you see for general medical issues and checkups?
We’d like you to think about the last year, starting with August 1, 2009. During this last year, how many times have you visited a primary care physician, meaning a doctor you see for general medical issues and check- ups?
Shorter reference periods are better for frequent or less important behaviors. Longer reference frames are fine for infrequent behaviors, especially if they are major events.
How often have you felt worried about the economy in the last month?
People may worry about the economy a lot, so each distinct time they did so may run together and not be that memorable. A question like this needs a shorter time period. Asking about a year or a month may be too long of a time period to yield reliable respondent estimates.
How many times have you had a heart attack in the last ten years?
Heart attacks are memorable and major events, so a question like this can take a longer time frame and still yield reliable estimates.
Avoid asking too many recall items.
Recall questions are effortful for respondents because they require sorting through memories and compiling them into informed responses. Asking too many recall items can tire respondents, harming the quality of their estimates as they start to satisfice by guessing rather than really thinking about their responses. There is no rule of thumb to follow for what constitutes “too many” recall questions, but researchers should always pretest their surveys to see if the sheer number of such items is taxing respondents too much.
General phrasing may suffice.
Sometimes researchers are just interested in an average rather than a precise count of the occurrences of some event. In these instances researchers may want to use general phrasing in their questions rather than asking for precise counts during very specific periods of time. These questions are less taxing for respondents and may be useful substitutes when there are too many specific recall items in a survey. For example:
How many cigarettes do you smoke in a typical day?
On average, how many times do you purchase groceries during a normal week?
How good is the economy these days? How good is the economy lately?
Author: Patrick R. Miller, DISM Survey Research Associate.