Cognitive Interviewing: lessons from forensic psychology for qual research

On January 22nd, 1996, Connie Babin took the witness stand and identified Shareef Cousin as the man who shot and killed her date, Michael Gerardi, during an armed robbery. When asked by the prosecutor whether she was positive, Babin replied with absolute, unwavering certainty: “Yes. I will NEVER forget that face”.

Three years later, Cousin was exonerated. Evidence placing him at a basketball game across town at the time of the murder proved that Babin had identified the wrong man.

Sadly, the Cousin case is not unique. The unreliable nature of eyewitness accounts is a well-documented phenomenon. And in the vast majority of cases, false information comes down not to deliberate deception, but rather the limitations of human memory. Ask a witness to a crime what happened and you tend to get a fraction of what they know. Likewise, the information you do get can be tainted by external influence during the interview.

Forensic psychology has spent decades working on this problem. Market research has paid surprisingly little attention.

The cognitive interview

In the mid-1980s, psychologists Ronald Fisher and Edward Geiselman developed a technique called cognitive interviewing as a response to this issue. It had been noted that the poor quality of testimony largely came down to the way police officers questioned willing witnesses: interrupting, jumping between topics, and asking closed questions that cut off free recall.

The method rests on how memory functions. Recalling events is not a perfect playback, but rather a reconstructive process where retrieval is heavily dependent on cues. The interviewer therefore plays an important role in facilitating conditions that allow for the interviewee to retrieve as much as possible, as accurately as possible.

The core techniques are simple:

1. Context reinstatement: Mentally reconstructing the physical, emotional and situational context of an event to help retrieve associated memories.

2. Report everything: Encouraging witnesses to recall all details, even those that seem trivial, incomplete or irrelevant.

3. Recall in different orders: Reconstructing events in a different sequence, such as working backwards, to access details that might be missed in a conventional chronological account.

4. Change perspective: Attempting to recall the event from a different vantage point, potentially revealing additional details. This should not, however, be an invitation to speculate about what another person experienced.

Layered on top of this are the principles of establishing rapport, using open questions, and letting the witness lead.

Fisher and Geiselman’s experiments reported that witnesses interviewed with this technique recalled roughly a quarter to a third more correct information than those given a standard police interview. Later meta-analyses have supported this, finding a large increase in correct details compared with standard interviews, with only a small rise in errors and no meaningful loss of overall accuracy (Köhnken et al., 1999; Memon, Meissner and Fraser, 2010).

Applications in qualitative market research

Compare this with a typical qualitative interview or focus group where the discussion guide tends to move briskly through a list of topics. Questions are often leading (“How important is price?”), abstract (“What do you look for in a pet food?”) or invite rationalisation (“Why did you choose that particular brand?”). Respondents are asked to report on their motives, which psychologists have long known to be poorly accessible. Nisbett and Wilson’s classic 1977 paper argued that people frequently cannot report the true causes of their own behaviour and instead construct reasonable-sounding explanations. Skilled moderators know this. The standard toolkit, however, has few systematic answers to it.

Leading questions matter enormously too. Psychology experiments (Loftus and Palmer, 1974) have shown that changing a single verb in a question altered witnesses’ estimates of a car’s speed or whether there was broken glass on the ground following an accident. Qualitative research is full of equivalent verbs. So what can moderators learn from forensic psychology? There are mainly three lessons worth borrowing: The first is to anchor in specific events rather than general attitudes. A witness is asked about this incident, not about crime in general. Qualitative research can do the same: ask about the last time a pet owner visited the vet, what happened and in what order, rather than how they feel about veterinary care. Reasons emerge from the account instead of being invented on demand. The second is context reinstatement. People recall decisions better when guided back through the circumstances: where they were, who else was involved, what else was on their mind. This is particularly valuable for choices that were low-attention at the time, such as routine purchases.

 The third is restraint. The cognitive interview treats the interviewer as a facilitator, not a questioner. It asks for fewer interruptions, longer silences and open prompts. In a commercial setting, where there is pressure to cover the whole guide in the time available, this is the hardest lesson to apply and probably the most valuable.

As with everything, there are also limitations to be aware of. The cognitive interview was designed to recover facts about discrete events, while qualitative research often seeks meanings, attitudes and intentions, which are not stored as memories in the same way. The technique also carries risks if applied carelessly: more probing can encourage confabulation, and the original forensic research takes care to guard against it. Borrowing the method means borrowing its safeguards.

Conclusion

Market research has long been preoccupied with sample size, speed and the sophistication of analysis. Yet the quality of the underlying conversation often receives less scrutiny, despite everything that follows depending on it. Forensic psychology offers a useful perspective on how much the interview itself can shape the evidence it produces. Perhaps it is time qualitative research paid closer attention to how questions are asked, not just what answers they produce.