WooCommerce Post-Purchase Survey: 7 Questions, When to Ask and What to Do Next Recall bias
Glossary 
Mike Taylor Updated: Nov 11, 2026 Reading time ≈ 4 min
- Topic
- Survey bias
Contents
Recall bias is a systematic error that occurs when people remember past events, behaviors or exposures inaccurately or incompletely, so their answers about the past are skewed in a predictable direction. It affects surveys that ask "how often" or "when", and retrospective studies such as case-control studies in epidemiology, where people with a condition may remember past exposures differently from people without it. Knowing how it works helps you word questions so memory has less room to distort the answers.
Recall bias is a type of response bias. In epidemiology it is classed as information (measurement) bias. Unlike lying, it is usually unconscious: the person believes they remember correctly.
Why recall bias happens
- Forgetting. Routine and distant events fade first. The longer the period you ask about, the more people estimate instead of remembering.
- Telescoping. People misplace events in time. Forward telescoping moves an older event into the period you asked about ("in the last 6 months"), which inflates counts.
- Salience. Vivid, emotional or recent events are recalled more easily and are over-reported compared with ordinary ones.
- The present colors the past. People's current situation, mood or beliefs shape what they recall. Someone who is now ill searches their memory harder for possible causes.
- Estimation shortcuts. Instead of counting, people answer from a "typical week" or round to convenient numbers like 5, 10 or 20.
- Social norms. Memory drifts toward what is approved, which overlaps with social desirability bias.
Examples of recall bias
| Question | What tends to go wrong |
|---|---|
| "How many times did you see a doctor in the past 12 months?" | Recent or serious visits are remembered, routine ones are forgotten, and older visits get pulled into the 12-month window |
| "How often do you eat vegetables?" | People describe an ideal or typical week, not what they actually ate |
| "How satisfied were you with your visit in March?" | The answer reflects the overall impression of the brand and later experiences more than the visit itself |
| Case-control study of birth defects: "Did you take any medication during pregnancy?" | Mothers of affected children often search their memory more thoroughly than mothers of healthy children, which can create a false association |
Differential and non-differential recall bias
- Differential: one group remembers differently from another, as in the case-control example. This can create an association that does not exist or exaggerate a real one.
- Non-differential: everyone misremembers in a similar way. It adds error and, for a simple yes/no exposure, usually pulls the estimated association toward zero.
In customer and employee surveys, recall bias becomes differential when you compare groups with different experiences, for example customers who complained against those who did not.
How to reduce recall bias in a survey
- Shorten the reference period. "In the past 7 days" gives more accurate answers than "in the past year". If you need a yearly figure, ask about a short period and scale up, or ask several times.
- Use landmarks. Anchor the period to a memorable date: "Since New Year's Day…" or "Since you started this job…".
- Ask about specific episodes. "Think about your last order" works better than "How are your orders usually?".
- Break big questions down. Instead of "How much did you spend on travel last year?", ask about flights, hotels and trips separately.
- Ask close to the event. A short survey sent right after a purchase, support case or visit leaves little time for memory to change. See triggered surveys.
- Use diaries or records when accuracy matters: receipts, calendars, app data, or a short daily log instead of one retrospective question.
- Keep the hypothesis neutral. In research, do not tell participants which exposure you suspect, and ask cases and controls the same questions in the same way.
- Pretest the questions. Cognitive interviewing shows how people actually reconstruct their answers and which periods they can handle.
How to spot it in your data
- Heaping: answers cluster on round numbers (10, 20, 50), a sign of estimation instead of recall.
- Mismatch with records: where you have purchase logs or visit data for some respondents, compare their answers with the records.
- Drift with time: answers about the same event change depending on how long ago it happened.
Common mistakes
- Asking about a long period without any anchor or breakdown.
- Treating self-reported frequencies as exact facts in the report.
- Using leading wording such as "How often did you enjoy our service?", which shapes both memory and answer.
- Putting a long "think back over the past year" block at the end of the survey, when respondents are already tired.
In the methodology section, name the reference period and say that answers are self-reported: "Questions on past behavior were limited to the previous 30 days to reduce recall bias."
Updated: Nov 11, 2026 Published: May 31, 2026