A Mega-Analysis of the Effects of Feedback on the Quality of Simulated Child Sexual Abuse Interviews with Avatars

Pompedda, Francesco; Zhang, Yikang; Haginoya, Shumpei; Santtila, Pekka

doi:10.1007/s11896-022-09509-7

Cited by 14 publications

(14 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experiments conducted so far (Pompedda et al, 2015(Pompedda et al, , 2020Krause et al, 2017;Haginoya et al, 2020Haginoya et al, , 2021 have shown that this training increases use of open questions. A recent mega-analysis (Pompedda et al, 2022) including a total of 2,208 interviews containing 39,950 recommended and 36,622 non-recommended questions from 394 participants including European and Japanese students, psychologists, and police officers showed that feedback robustly increased recommended questions and decreased non-recommended questions resulting in more correct details being elicited from the avatar, and more correct conclusions being reached about what had "happened. "…”

Section: Introductionmentioning

confidence: 99%

AI avatar tells you what happened: The first test of using AI-operated children in simulated interviews to train investigative interviewers

Haginoya

Ibe²,

Yamamoto

et al. 2023

Front. Psychol.

Self Cite

View full text Add to dashboard Cite

Previous research has shown that simulated child sexual abuse (CSA) interview training using avatars paired with feedback and modeling improves interview quality. However, to make this approach scalable, the classification of interviewer questions needs to be automated. We tested an automated question classification system for these avatar interviews while also providing automated interventions (feedback and modeling) to improve interview quality. Forty-two professionals conducted two simulated CSA interviews online and were randomly provided with no intervention, feedback, or modeling after the first interview. Feedback consisted of the outcome of the alleged case and comments on the quality of the interviewer’s questions. Modeling consisted of learning points and videos illustrating good and bad questioning methods. The total percentage of agreement in question coding between human operators and the automated classification was 72% for the main categories (recommended vs. not recommended) and 52% when 11 subcategories were considered. The intervention groups improved from first to second interview while this was not the case in the no intervention group (intervention x time: p = 0.007, ηp2 = 0.28). Automated question classification worked well for classifying the interviewers’ questions allowing interventions to improve interview quality.

show abstract

Section: Introductionmentioning

confidence: 99%

AI avatar tells you what happened: The first test of using AI-operated children in simulated interviews to train investigative interviewers

Haginoya

Ibe²,

Yamamoto

et al. 2023

Front. Psychol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…The fact that only two interviews were conducted with feedback provided only once may explain the difference between the feedback intervention, which was unlikely to have reached its maximum impact, and the modeling intervention, for which a large improvement effect could be expected with a single intervention. On the other hand, given that a decrease in not recommended questions was shown by the mega-analysis already after it having been provided once (Pompedda et al, 2022), additional factors may be involved in the lack of a feedback effect for this variable in the present study. A possible explanation is the observer's influence (also known as the Hawthorne effect) on the participants' behavior during the interview simulation.…”

Section: Improvements In the Quality Of Investigative Interviews Usin...mentioning

confidence: 63%

“…Importantly, the simulated interviews with AI Avatars could be used to improve interviewing skills and quality of the information elicited from the avatars when combined with feedback and modeling interventions. The increase in the number of recommended questions shown in the feedback group has also been demonstrated robustly in a mega-analysis (Pompedda et al, 2022) that combined nine studies and examined the effects of feedback in Avatar Training.…”

Section: Improvements In the Quality Of Investigative Interviews Usin...mentioning

confidence: 75%

“…The experiments conducted so far (Haginoya et al 2020(Haginoya et al , 2021Krause et al 2017;Pompedda et al 2015Pompedda et al , 2020 have shown that this training increases use of open questions. A recent mega-analysis (Pompedda et al, 2022) including a total of 2,208 interviews containing 39,950 recommended and 36,622 non-recommended questions from 394 participants including European and Japanese students, psychologists, and police officers showed that feedback robustly increased recommended questions and decreased non-recommended questions resulting in more correct details being elicited from the avatar, and more correct conclusions being reached about what had 'happened'.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AI Avatar Tells You What Happened: The First Test of Using AI-Operated Children in Simulated Interviews to Train Investigative Interviewers

Haginoya

Ibe²,

Yamamoto³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Background: Previous research has shown that simulated child sexual abuse interview training using avatars paired with feedback and modeling improves interview quality. However, in order to make this approach scalable, the classification of interviewer questions needs to be automated.Objective: We tested an automated question classification system for simulated investigative interviews with children while also providing interventions (feedback and modeling) aimed at improving interview quality. Participants and Setting: Forty-two professionals were randomly assigned to the no intervention, feedback, or modeling group. Methods: The participants conducted two simulated child sexual abuse interviews online while receiving no intervention, feedback or modeling after the first interview. Feedback consisted of the outcome of the alleged case and comments on the quality of the questions asked by the interviewer. Modeling consisted of learning points and videos illustrating good and bad questioning methods.Results: The total percentage of agreement in question coding between human operators and the automated classification was 72% for the main categories (recommended vs. not recommended) and 52% when the eleven subcategories were considered. The intervention groups improved from first to second interview while this was not the case in the no intervention group (intervention x time: p = .007, ηp2 = .28).Conclusions: Automated question classification worked well for classifying the interviewers’ questions allowing interventions to improve the quality of the interviews.

show abstract

“…All investigative interviewers, but especially novices, may therefore benefit from tools that help them monitor their questions' quality during the interview. The design and testing of such tools may be a valuable and innovative avenue for future research such as using serious games (for example, see Pompedda et al 2022).…”

Section: Implications For Research and Practicementioning

confidence: 99%

A vignette study of novices' interviewing skills of asylum seekers

Kask

Groznov

Veldhuizen³

2022

Journal Invest Psychology

View full text Add to dashboard Cite

The aim of the current research was to replicate Van Veldhuizen, Horselenberg, Landstrom, et al.'s (2017) vignette study among novice Estonian police cadets to map their interviewing skills. Sixty-one police cadets from the Estonian Academy of Security Sciences were asked to respond to one of four vignettes that contained fictitious asylum narratives. Two of the vignettes contained no evidence for the origin of the asylum seeker, and the other two contained no evidence for the claim of persecution. The cadets were asked to formulate five questions that would help them to assess the credibility of the applicant's claim. We coded the style, type, and content of the questions. Our analyses showed that, in line with best practice, the cadets mainly formulated open questions in an information-gathering style. A thematic analysis revealed that when a claim about origin was assessed, cadets typically formulated questions about life in the country of origin, identity documents, and the flight to Europe. When assessing a persecution claim, in contrast, they mostly formulated case-specific questions.

show abstract

A Mega-Analysis of the Effects of Feedback on the Quality of Simulated Child Sexual Abuse Interviews with Avatars

Cited by 14 publications

References 46 publications

AI avatar tells you what happened: The first test of using AI-operated children in simulated interviews to train investigative interviewers

AI avatar tells you what happened: The first test of using AI-operated children in simulated interviews to train investigative interviewers

AI Avatar Tells You What Happened: The First Test of Using AI-Operated Children in Simulated Interviews to Train Investigative Interviewers

A vignette study of novices' interviewing skills of asylum seekers

Contact Info

Product

Resources

About