STUDY 1
Study 1 adopted a mixed methods approach comprising two parts. Study 1a provided a quantitative comparison of memory performance between PWS and a comparison group during a mock eyewitness interview. Based on previous research suggesting that PWS may experience higher cognitive load when recalling past events, it was hypothesized that PWS would report fewer correct details (i.e. reduced completeness), and make more errors (i.e. increased number of incorrect details) and therefore be less accurate (with accuracy operationalized as correct details/correct details + errors) than a comparison group with no stammer. Analyses also examined whether there were differences between groups within the free‐ and cued‐recall phases separately (without directional hypotheses). Study 1b provided a qualitative exploration of PWS's lived experiences of recalling the event under a mock witness interview in the study. Via open‐ended survey responses and follow‐up semi‐structured interviews, PWS were probed on their views on challenges, what worked well and what could be done differently to better support them during interviews.
Study 1: Method
Design
A mixed methods approach encompassed a quantitative comparison of the completeness and accuracy of participants' eyewitness recall between PWS and a comparison group, followed by a qualitative exploration of PWS witnesses' lived experiences of recalling the event under a mock witness interview.
For the quantitative eyewitness accuracy phase (Study 1a), PWS and comparison (no stammer) participants viewed a video of a mock crime before being interviewed on their memory for it. The independent variable was participant group (PWS vs comparison). The dependent variables were number of correct details (completeness), number of incorrect details (errors) and accuracy (number of correct details divided by the sum of correct and incorrect details).
After providing their eyewitness testimony, in Study 1b, the PWS group were asked to provide qualitative feedback on their experiences of giving evidence via open‐ended survey questions. Additional follow‐up semi‐structured interviews provided a further opportunity for PWS to elaborate on their experiences in a focused yet flexible manner (Trainor & Bundon, 2021). This part of the study was grounded in a critical realist framework, enabling a deep and nuanced exploration of the subjective lived experiences of PWS while considering the broader bio‐psycho‐social factors that influenced these experiences (Jordan et al., 2024).
The study design, hypotheses and planned analyses were pre‐registered on the Open Science Framework (https://osf.io/dbqr5/?view_only=06e499b49446401dae6c86a6a9f614a4).
Participants
All participants were required to be a UK resident, 18 years of age or older and proficient in English at a first‐language level. Additionally, PWS had to have a current stammer, irrespective of its duration. Recruitment occurred through purposive sampling (Nyimbili & Nyimbili, 2024), using advertisements on closed social media groups (Facebook, LinkedIn), the University of Bath's Psychology Community Research Panel and STAMMA's research platform.
Eighty‐one participants took part in the study; however, 14 were excluded for either not residing in the United Kingdom (n = 10) or failing to replicate their participant‐generated ID code (n = 4). This resulted in a final sample of 67 participants: 26 in the PWS group and 41 in the comparison group. Demographics and mean cognitive ability scores are shown in Table 1. An independent samples t‐test indicated the PWS group was significantly older than the comparison group, t(45.64) = −4.15, p < .001, d = 1.09, 95% CI [−1.61, −0.56], and a chi‐square test revealed a statistically significant difference in sex distribution between groups, χ 2(1) = 10.66, p = .001, Cramer's V = .40, OR = 5.69, 95% CI [1.93, 16.79], with relatively more females in the comparison group. However, there was no difference in cognitive ability score (as measured by the Self‐Administered Vocabulary IQ Test – see Materials section below) between groups, t(40.78) = −0.51, p = .616, d = 0.14, 95% CI [−0.64, 0.37].
Table. Demographics and cognitive ability score of the stammering and comparison group in Study 1. (See the original article for the full table.)
PWS participants received a £15 Amazon voucher upon completion, while the comparison group (whose participation was shorter as they did not complete the post‐testimony reflections survey/interview) received a £10 voucher. The study was approved by the University of Bath Research Ethics Committee (reference number: 5309‐6811).
Materials and procedure
All participants took part individually online. First, participants viewed a 2‐min crime video via screenshare. The film was originally created for use in a previous study (see Maras et al., 2020) and depicted a fight between two males, witnessed from the perspective of a bystander. After viewing the video (and before being interviewed on their memory for it), participants completed demographic and cognitive ability questions on QuestionPro, taking around 15 min.
Online self‐administered vocabulary IQ test (SA‐VIQT)
The SA‐VIQT (available at https://openpsychometrics.org/) consists of 45 cognitive ability questions whereby participants select two out of five words that share the same meaning. Participants receive one point for each correct answer, and one point is deducted for each incorrect answer (‘Don't know’ responses yield zero points). The test demonstrated moderate convergent validity, with Pearson's r ranging from .48 to .54 (Logos et al., 2021).
Eyewitness testimony interview
Next, participants were interviewed on their memory of the video by one of four research assistants (three females and one male, aged between 22 and 24 years). Interviews followed a standardized schedule designed to incorporate best practice for interviewing witnesses (Ministry of Justice, 2022). Initially, participants were asked to recall all details they remembered, but only those they were confident were accurate (‘free recall’). They were then asked a series of questions based on their free recall accounts using orienting ‘tell’, ‘explain’, ‘describe’ prompts (e.g. ‘You mentioned someone tried to intervene; please describe this person in as much detail as you can remember’).
Qualitative post‐testimony reflections survey
After the eyewitness interview, participants in the stammering group received a link to a survey asking about their experiences of providing testimony. The survey was developed based on existing empirical findings. In line with participatory research principles, which emphasize the value of co‐producing knowledge through systematic inquiry in direct collaboration with the populations being studied (e.g. Pérez Jolles et al., 2022), the survey (and wider set of studies more broadly) was co‐produced with an individual with lived experience of stammering and supporting other PWS. It comprised 20 open questions and seven closed questions, divided into six sections: memory, ability to articulate details, emotional impact of the interview, management of stammer during the interview, influence of interviewer reactions and suggestions for improvement (Supplementary Material A).
Qualitative post‐testimony reflections interview
Following a review after 3 weeks of data collection, it was determined that survey responses lacked sufficient depth. Consequently, an interview schedule was developed to further explore the topics most salient in survey responses. PWS participants who had completed the initial qualitative survey were contacted to participate in a follow‐up interview; subsequent participants were given the option to complete either the survey or follow‐up interview. Ten PWS participants completed this final optional phase. Twelve open‐ended questions covered key topics, but discussions were further guided by what the researcher interpreted to be meaningful to the participant (Trainor & Bundon, 2021). The interview schedule mirrored the survey's focus areas, including memory and articulation of event details, cognitive and emotional impact, interviewer reactions, coping strategies and suggestions for improvement (Supplementary Material B).
Coding
Eyewitness testimony
The recorded eyewitness interviews were transcribed and coded by the third author. Each new item of detail reported by participants was coded as correct (if it matched that seen in the video) or incorrect (if it did not match, was not present or was recalled as occurring in a different sequential order to what was shown in the video). The accounts of participants were coded at the finest level of detail available. For instance, ‘A female [correct] came over [correct]. She was wearing a blue [incorrect] jumper [correct]’ would receive three correct points and one incorrect point (her jumper was burgundy).
Items were coded only once the first time they were mentioned, but items were coded multiple times if they were necessary for understanding the context. For instance, the male friend was coded multiple times when participant reported ‘The male friend [correct] tried to intervene [correct] but one guy [correct] pushed [correct] the male friend [correct] away’. Items that were subjective (e.g. ‘He looked more aggressive’) were not coded. Twelve transcripts (17.91%) were randomly selected and independently coded by a second researcher. Excellent agreement was reached between the two raters, with intraclass correlation coefficients of >.99 (95% CI [0.996, 0.998]) for correct details and .99 (95% CI [0.98, 0.99]) for incorrect details.
Surveys and interviews with PWS
PWS participants' post‐testimony reflections were coded using an inductive reflexive thematic analysis by the second author. Adopting a critical realist perspective, the analysis recognized that participants' interpretations of their experiences were influenced by their broader ideologies, social contexts and prior stammering experiences (Braun & Clarke, 2022). The researcher's interpretations were also recognized as contributing to the construction of the analyses. The analysis was carried out at a semantic level, focusing on the explicit meaning of what participants said (Braun et al., 2023).
Braun and Clarke's (2006) six‐stage approach was flexibly followed, comprising (1) familiarization with the data through reading and re‐reading responses; (2) generating initial codes reflecting potentially interesting data features regarding the research question; (3) generating initial themes by grouping similar codes to elucidate shared meaning; (4) reviewing themes to ensure they tell the story of the full dataset; (5) refining and defining themes; and (6) developing the analytic narrative through contextualizing the themes in existing literature. Throughout, the researcher remained reflexive, iteratively revisiting and refining codes and themes to ensure they remained true to the data and adequately reflected the complexities of participants' experiences. The coding was discussed with another member of the research team, and it was agreed that the codes accurately reflected participants' experiences.
Study 1: Results
Eyewitness testimony
Three independent samples t‐tests were conducted to examine group differences in completeness (number of correct details), errors (number of incorrect details) and accuracy (number of correct details divided by the sum of correct and incorrect details), respectively. Due to unequal group sizes, Welch's t‐tests were used. These tests were repeated in further analyses to test for group differences in completeness, errors and accuracy within the free recall and cued recall phases separately.
Overall recall performance
There were no significant differences between PWS and comparison groups across the interview overall, in terms of the number of correct details, Welch's t(58.65) = 1.54, p = .129, d = 0.37, 95% CI [−0.12, 0.87], errors, t(56.68) = 1.69, p = .097, d = 0.42, 95% CI [−0.83, 0.91] or overall accuracy, t(50.25) = −0.90, p = .372, d = 0.23, 95% CI [−0.72, 0.26] (Table 2).
Table. Mean number of correct details, errors and accuracy for PWS and comparison groups across the interview overall (standard deviations in parentheses). (See the original article for the full table.)
Free recall
In the free recall phase, comparison witnesses reported significantly more correct details than the stammering witnesses, Welch's t(59.86) = −2.28, p = .026, d = 0.55, 95% CI [0.05, −1.05]. There were no differences between groups for errors, t(57.86) = 1.36, p = .179, d = 0.33, 95% CI [−0.16, 0.82] or accuracy, t(44.47) = −0.06, p = .950, d = 0.02, 95% CI [−0.51, 0.48] (Table 2).
Cued recall
There were no significant differences between PWS and comparison witnesses in number of correct details, t(63.11) = −0.50, p = .617, d = −0.12, 95% CI [−0.61, 0.37], incorrect details, t(61.83) = −1.19, p = .240, d = 0.28, 95% CI [−0.21, 0.78] or accuracy, t(63.69) = 1.85, p = .069, d = 0.43, 95% CI [−0.93, 0.07] during the cued recall phase (Table 2).
PWS reflections on providing testimony
The inductive reflexive thematic analysis of PWS participants' reflections (in semi‐structured interviews and survey responses) of providing testimony generated three overarching themes: (1) ‘A vicious cycle of anxiety, pressure, and stammering’ (subtheme: ‘worries about perceived credibility’); (2) ‘Impression management’ (subtheme: ‘collateral damage’); (3) ‘Levelling the playing field: ensuring accessible testimony’ (subthemes: ‘individual and systemic facilitators’ and ‘the crucial role of the interviewer’). A thematic map is presented in Figure 1, while Supplementary Material C presents each theme and subtheme with their associated codes and example pseudonymized quotations.
Figure. Thematic map illustrating the three themes and four subthemes of participants' reported experiences. (See the original article for the figure.)
Theme one: A vicious cycle of anxiety, pressure, and stammering
Participants expressed that the process of providing testimony inherently disadvantages PWS compared to fluent speakers. While participants acknowledged that ‘the stress and pressure alone would affect anyone’ (Matthew, 29, male), they emphasized that having a stammer introduces an additional layer of difficulty. A core element of the anxiety experienced by PWS was their ‘fear of stammering’ (Oliver, 44, male). Participants reported that a desire to avoid stammering contributed to a cycle, whereby the high‐pressure environment heightened their anxiety about stammering, which in turn further increased the frequency and intensity of stammering. Moreover, many participants pointed to the added anxiety of speaking in front of multiple strangers, regarded by Ruby (32, female) as ‘shit scary’. Participants felt that the high‐stakes performance demands of providing testimony inhibited their ability to cope with their stammer.
Subtheme: Worries about perceived credibility
For many participants, anxiety over speaking in front of others went beyond typical fears of public speaking; it was deeply rooted in concern over one's articulation abilities and the impact of these on listeners' perceptions. Participants reported acute awareness that ‘[…] for most people in the world stammering isn't something they are familiar with; they don't understand it’ (Oliver, 44, male), which contributed to concerns about how their testimonies might be misinterpreted in the CJS. A common fear amongst participants was that the societal bias linking fluency with credibility would undermine listener's perceptions of their honesty, and they expressed frustration with the lack of understanding about how stammering manifests. For some, the fear of appearing dishonest went beyond a mere point of anxiety to profoundly affect their belief in their reliability as witnesses, driven not by one's ability to recall events but by anticipation of listeners misjudging them as dishonest. For example, Sophia (31, female) remarked that she ‘would feel useless in a real‐life crime investigation’, indicating a dissonance between internal confidence in the accuracy of testimony and the anticipation of negative external perceptions.
Theme two: Impression management
As a result of their anxiety over listener misperceptions, many participants experienced a constant internal conflict between trying to speak fluently and accurately articulating their testimony. Participants felt preoccupied with anxiety over how they were being perceived by listeners, ‘Like if we do stammer, what impact that will have on listeners and also us, in terms of like … you know … our confidence for example, or like how we perceive the listener to be viewing us’ (Ben, 24, male). Participants indicated that providing testimony as a PWS is entangled with fears of social stigma, leading them to be hypervigilant for listeners' reactions and focusing less on testimony content and more on navigating ‘[…] what can go wrong with how they're perceived’ (Louie, 49, male).
Several participants attempted to manage listener misperceptions by employing pre‐emptive avoidance strategies to maintain fluency, including avoiding certain words or sounds that they knew would induce their stammer. For example, Ruby (32 female) noted ‘I also try to use alternate words most of the time when I am unable to speak a particular word. But I have also realised that the essence of the story is lost most of the times when using alternate words’.
Many participants also reported limiting their responses to avoid stammering or managing their stammer by breaking eye contact to regain control of their speech: ‘I mean it's better to not speak than to sort of get myself into more problems’ (Louie, 49, male).
Subtheme: Collateral damage
The pre‐emptive employment of avoidance strategies fed into a trade‐off between participants' prioritization of fluency over testimony content. Several participants explicitly expressed fears over the impact of this trade‐off on the clarity of their testimony. Most felt that although they were able to convey their intended message, the use of avoidance strategies made their testimony less precise and clear and thus compromised its effectiveness. For many participants, the effort it took ‘to go around the houses’ (William, 45, male) was ‘exhaustive’ (Ruby, 32, female) and ‘draining’ (William, 45, male). This cognitive burden comprised a complex interplay between fears of listeners' perceptions, employment of avoidance strategies and testimony articulation.
Contrastingly, a few participants prioritized the accuracy of their testimony over appearing fluent to listeners. For these participants, testimony content closely aligned with what they intended to say because, as Megan (40, female) explained, ‘Even if I knew I would stammer it was important to say what I wanted to say’.
Despite variability in stammer management, participants consistently asserted that their memory of events remained unaffected, with Talia (47, female) noting that she could ‘recall the whole incident vividly’. Instead, several participants acknowledged that the difficulty lay in effectively communicating these memories; ‘the way I talk is the only thing that is affected here’ (Jacob, 30, male).
Theme three: Levelling the playing field – Ensuring accessible testimony
Participants emphasized the need for accommodations to ensure that the testimony process is accessible and equitable for PWS. Drawing parallels with physical disabilities, many argued that the same level of consideration must be given to PWS, as conveyed by Kaya's (32, female) remark that ‘[…] if you had a courtroom and you had somebody in a wheelchair, you wouldn't ask them to walk up a staircase’.
Participants felt that the right accommodations would enable PWS to provide effective testimony, with Matthew (29, male) stating, ‘So, make the allowances and I don't see any reason why we couldn't’. However, participants stressed that these accommodations must consider the unique needs and abilities of each individual because ‘it's not a one size fits all’ (Matthew, 29, male).
Subtheme: Individual and systemic facilitators
Participants identified a range of strategies and adjustments to facilitate the testimony process and overcome the challenges outlined previously. These facilitators consisted of both individual coping strategies (e.g. slowing speech, taking time and rehearsing beforehand) and systemic accommodations (e.g. familiarization with processes). For example, Logan (59, male) notes: ‘I would want to meet the judge in person and have a look round the venue and familiarise myself with the court and the judge’.
Of particular importance to participants was the need to disclose their stammer to set listener expectations and relieve the pressure to remain fluent: ‘If they're waiting … if they're waiting for pauses and things on difficult questions … and they misconstrue the pause as an overthinking or trying to think of a way out of the answer rather than a block … yeah … then perhaps they should be aware that there's a stammer there’ (Louie, 49, male).
Moreover, participants highlighted the need for education to mitigate the impact of prejudices on listener perceptions. Listeners' awareness of ‘their own potential prejudices and stereotypes’ (Logan, 59, male) was also deemed crucial to fostering an environment where PWS feel confident that their testimony will be judged on its content rather than their speech. Further, several participants suggested that offering PWS the option to provide their testimony through alternative mediums, such as in written format, via video link or in a one‐to‐one exchange would further relieve the pressure to appear fluent (while acknowledging this would need to be the individual's choice). As Oliver (44, male) suggests, ‘[…] or possibly have it pre‐recorded so you're not as under so much pressure within a courtroom or a police station or something like that’.
Subtheme: The crucial role of the interviewer
Participants expressed the importance of the interviewer‐interviewee dynamic in facilitating – or inhibiting –testimony quality. Benedict (40, male) notes: ‘Because the interviewer was patient, conscientious, and understanding of my speech, it put me at ease, so I was able to talk more fluently’. Participants stressed the importance of establishing familiarity and rapport with the interviewer to reduce their anxiety and facilitate their testimony articulation: ‘I had the opportunity to engage with yourself a bit before, I sort of, you know, had trust in yourself’ (Matthew, 29, male). Equally, participants pointed to the detrimental effect that interviewer impatience, lack of understanding and negative demeanour would have on exacerbating their stammering, and in turn, inhibiting their ability to provide testimony. As Benedict (40, male) notes: ‘If I sense any negativity towards my speech, it makes it worse’.
Study 1: Discussion
This is the first study to explore eyewitness testimony by PWS. Contrary to hypotheses, PWS and comparison groups did not significantly differ in the completeness, errors or accuracy of their overall accounts, indicating that PWS were able to effectively allocate their cognitive resources to recall events, despite experiencing other demands such as speech monitoring and impression management. However, analyses within the free‐ and cued‐recall interview phases separately indicated that the PWS group reported fewer correct details than the comparison witnesses during free (but not cued) recall. In free recall, witnesses were asked to recall as many details as possible about the crime scene without any explicit prompts on what details should be retrieved. This lack of retrieval support may place excess burden on an already heavy cognitive load for the PWS group, resulting in less detailed eyewitness accounts relative to those with no stammer. These differences were diminished, however, with structured cued recall, which places less demand on cognitive resources (see Glenberg, 1997; Tulving, 1985).
In line with these quantitative findings regarding recall quality, in follow‐up surveys and interviews, the PWS group reported experiencing a cyclical interplay between communicative pressure, anxiety over listener misperceptions and stammer severity, which they navigated by prioritizing either fluency or testimony content. A primary source of reported anxiety was concern over being incorrectly perceived as dishonest or lacking credibility. For some participants, anxiety about external judgements diminished their confidence in their ability to provide testimony. Participants responded to fears over listener misperceptions in two primary ways. Some PWS attempted to avoid stammering through concentrated efforts, aligning with existing literature indicating that PWS often conceal their overt stammer to manage listener reactions (Constantino et al., 2022; Grisham et al., 2015; Jackson et al., 2015). In line with previous research, participants reported using coping strategies such as word‐switching, limiting speech output and avoiding eye contact (Connery et al., 2020; Constantino et al., 2022). However, participants expressed concerns that this compromised the clarity of their testimony. Conversely, other participants prioritized testimony content and chose to speak through their stammer. This variability in experiences highlights the diverse ways that PWS navigate the testimony process and manage their speech during testimony (Garcia‐Barrera & Davidow, 2015; Tichenor et al., 2022). These findings underscore the need to examine how PWS are perceived by others when providing evidence.
CC BY 4.0. "Eyewitness testimony by individuals who stammer: Evidence, experience and perceived credibility" by Maras K, Park S, Grafton P et al., British journal of psychology (London, England : 1953) (2026). https://doi.org/10.1111/bjop.70014. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC12783885). Headings and layout adapted for reading; figures and tables are in the original.