Interpreting Eyewitness Confidence: What do you mean when you say, “pretty sure”?

5th August 2020

Pia Pennekamp

Crimes and eyewitnesses

When a crime happens, the police put in a lot of effort to find out who did it. If the crime had an eyewitness, they can usually help. The process of gathering eyewitness evidence is complicated, lengthy, and requires a lot of resources. For example, crime scenes need thorough investigation. In addition to physical evidence, interviews and/or interrogations often need to be conducted. Having an eyewitness look at a lineup can be an important step in the course of an investigation and eyewitnesses play a key role in the courts. Asking an eyewitness to view a lineup is common practice, particularly for serious crimes. However, this procedure is not without pitfalls (but more on this later).

Once the police have a suspect, eyewitnesses can help by looking at a lineup. A lineup is a commonly used procedure to collect eyewitness identification evidence (recommendations for best practices for lineup procedures). Eyewitnesses are generally called in to try and identify the culprit, if he is present, among fillers (i.e., people that are not suspected of having committed the crime) from a photo or video lineup (more on fillers here). At this point, the job of an eyewitness is to use their memory for the culprit to provide evidence that the suspect is either guilty or innocent. Ultimately, when an eyewitness makes an identification, they provide evidence that the suspect is guilty (by identifying him/her) or innocent (by not identifying him/her). They provide valuable information for investigations and when they testify in court, they are very likely to be believed (e.g. jury perceptions of eyewitness evidence). But we know that memory is fallible. The Innocence Project in the United States has revealed that erroneous eyewitness accounts were involved in 70% of 362 later exonerated convictions in the United States.

Eyewitness identification procedures

Different jurisdictions, police forces, and even individual officers, use a variety of methods to obtain eyewitness evidence. In addition to making a decision after viewing a lineup, an eyewitness may indicate how confident they are in that decision. For example, remember how I said that eyewitnesses are very likely to be believed in court? Well, one piece judges and jurors rely on to decide whether to believe an eyewitness is their confidence. However, the way that an eyewitness’ confidence in their lineup identification is collected varies a lot around the world. There is quite a bit of variability even within the United Kingdom. In England and Wales, eyewitnesses are asked to confirm their identification (PACE Code D, 2017). In Northern Ireland, the police collect the reason for the identification, the “words of recognition”, expressions of doubt, and the features of the image or person that led to recognition. In Scotland, eyewitnesses indicate “why” they identified the person they did.

There is also no standard for interpreting the level of confidence expressed by the eyewitness. Any comments the eyewitness makes are recorded, including spontaneous confidence judgments. In the United Kingdom, asking an eyewitness about their confidence is not necessarily part of police practice in part because if the eyewitness says anything other than 100% confident or absolutely certain, their statement may be perceived as suggesting reasonable doubt.

Using eyewitness identification confidence

We used to think that an eyewitness’ level of confidence could not provide much useful information about the accuracy of their lineup decision. However, more recently we have come to understand certain circumstances, confidence does indicate how accurate an eyewitness may be. This is especially true for high confidence suspect identifications (What are these circumstances? More here. Although a highly confident eyewitness can still sometimes be wrong.

But interpreting an eyewitness’ confidence is not always straightforward. For example, sometimes an eyewitness may use quite ambiguous language when asked for their confidence in their own words or when asked why they made their identification. For example, they might say something like “I am relatively sure about my decision” or that he/she “looks right”. This is important because as I mentioned before, research shows that eyewitness’ confidence will be used by the criminal justice system to evaluate how reliable (i.e., how accurate) they are. Commonly, an eyewitness’s level of confidence is not absolute, but rather somewhere in between (i.e., “medium” confidence). Mansour( 2020 ) showed that, at least in the lab, over 70% of people asked “why” they made their identification gave a response that was interpreted by others as being medium confident. What is more, people tend not to use extreme phrases (such as 0% confident/very unlikely or 100%/Very likely) when talking about their confidence in a decision. Yet, eyewitness identification evidence is treated as being one or the other, 0% confident that the suspect is guilty (no ID) or 100% confident that the suspect is guilty (ID). So how should we make sense of confidence?

Eyewitnesses use a variety of phrases when judging their confidence (Mansour, 2020; Pennekamp Batstone, & Mansour, 2019). Importantly, we (Mansour, 2020;) and others have found that other people’s interpretations of these phrases vary a lot.

For example, the phrase “pretty sure” may have one meaning for one person but an entirely different meaning for someone else. These differences in interpretation increase the possibility that the eyewitness will be misunderstood. Numbers may reduce the risk of misunderstanding. For example, a number like “30%”, seems to be less likely to be misunderstood than a verbal expression. But communicating with numbers is not instinctive. People prefer to communicate confidence (uncertainty) in their own words. This is important to consider because verbal statements of confidence are currently standard in practice in many places, like the United States (see for example).

What can we do to make eyewitness evidence more reliable?

Confidence can be predictive of accuracy (Wixted & Wells, 2017;) and can thus assist the criminal justice system in determining guilt. We need a reliable way to obtain and interpret eyewitness identification confidence. Fields like climate science and intelligence analysis use lexicons (i.e., dictionaries), which translate verbal confidence (e.g., likely) into a range of numeric values (e.g., 40%-75%). My research involves creating and testing a lexicon for confidence in lineup decisions.

More information

This project is part of my PhD, supervised by Dr. Jamal K. Mansour, at Queen Margaret University in Edinburgh. My PhD is being funded by a Queen Margaret University bursary (More info on bursaries at QMU here). Data collection is underway. If you would like to contribute or have questions, please get in touch by email ( or Twitter (@PennekampPia).

We are also always looking for individuals to join our lab (Mansour Lab, Dr. Jamal K. Mansour), so do get in touch if this topic is of interest to you. You may be interested to know that Queen Margaret University offers one-year Master of Research (MRes) programme as well as a PhD programme. If you are interested, you can also find out more here.

For updates and more blogposts related to memory in general, check out the Queen Margaret University’s Memory Research Group


Interested in eyewitness confidence? Read Dr. Mansour’s newest findings here (free to download until August 11).
After August 11, you can get a free copy from Dr. Mansour’s ResearchGate page.
You can also follow her work on Twitter (@EyewitnessIDup).

Back to blog