SCAN NING FOR TRUTH . S CHOLARS ’ AND PRACTITIONERS ’ PERCEPTIONS ON THE USE ( FULNESS ) OF S CIENTIFIC C ONTENT A NALYSIS IN DETECTING DECEPTION DURING POLICE INTERVIEWS

SCAN (Scientific Content Analysis) is an analytic method that claims to detect deception in written statements. Although the validity of SCAN is contested in literature, various (law enforcement) agencies across the globe are trained in using the technique. To date it remains unknown how the technique is perceived and to what extent it is used in practice. Based on a scoping review and an open- and closed-ended survey, we identified practitioners’ and scholars’ prevailing perceptions on the use(fulness) of SCAN. Data were collected from 48 participants (35 practitioners and 13 scholars). Key findings illuminate a discrepancy between practitioners and academics. While practitioners position themselves positively towards SCAN, academics urge for a complete disappearance of the technique. Practitioners apply an incomplete, personalized version of SCAN. Since SCAN is not applied in its originally designed form and existing research demonstrates the technique has important shortcomings, we advise practitioners to abandon SCAN altogether.


INTRODUCTION
Investigative interviewing is an essential part of police work. Investigative interviewers spend up to 80% of their time in the interview room during a judicial investigation (De Greef & De Fruyt, 2006). Obtaining truthful information about the offence and involvement of suspects is an important goal of investigative interviews (Gudjonsson & Pearse, 2011). Based on the assumption that verbal structures of true and false statements significantly differ from each other, verbal credibility assessment (VCA) tools can assist investigative interviewers in detecting lies and/or deception (Meijer et al., 2008;Nicklaus & Stein, 2020;Vrij, 2008). In particular, VCA methods test verbal indicators of statements based on the tool's specific criteria. The presence or absence of certain criteria is considered as an indication of either truthful or misleading/deceitful information (Bogaard et Smith & Willis, 2001). In this article, we focus on SCAN. 1 SCAN was developed by former polygraph examiner Avinoam Sapir (LSI, n.d.) and claims to detect misleading and hidden information, as well as involvement in a crime by analysing a written statement (Vanderhallen et al., 2016). SCAN thus represents itself as a technique to detect deception Smith & Willis, 2001;Vanderhallen et al., 2016), not as a lie detection tool. When applying SCAN, practitioners follow six steps . First, the 'initial phase' serves as an introduction between the interviewer and interviewee in order to minimize resistance of the interviewee . Second, in order to obtain a 'pure version statement', the interviewee is asked to write everything down about a certain event without any guidance or influence (Bogaard,   . Third, this written statement is analysed using SCAN criteria Meijer et al., 2008; in order to detect 'highlights' or 'hotspots' that indicate deception. Table 1 represents SCAN's criteria following Belgian police superintendent Bockstaele (2019). Fourth, these hotspots are discussed more thoroughly in subsequent interviews Smith & Willis, 2001). Fifth, as part of the detailed interview, the interviewer asks open questions and aims to build rapport, for example by using the word 'we' and smiling at the interviewee. Sixth, debriefing after the detailed interview allows the interviewer to probe for the interviewee's experienced emotions during the interview (Bockstaele, 2019).
[Insert Table 1 here] Extant research consistently dissuades the further use of SCAN for several reasons. Most importantly, SCAN criteria differ from study to study (Meijer et al., 2008;Oberlader et al., 2020). Besides the unclear content of these criteria, the number of criteria to be used ranges from ten to sixteen in different studies (see Bockstaele Vanderhallen et al., 2016). What is more, SCAN does not contain a criterion that allows to distinguish between true and false statements  . In other words, there seems to be a major discrepancy between academics and practice on the prevailing use and perceived usefulness of SCAN. In order to confirm or refute this discrepancy, it is essential to survey both practitioners and academics on their perceptions of SCAN. Furthermore, given the lack of academic substantiation, it is pivotal to gain insight into how and why SCAN practitioners continue to use this technique. Extant researchof which the majority experimental designs using students as participantshas not taken into account the use and perceptions of SCAN practitioners and academics. In this study, we fill this gap by assessing the use of SCAN among practitioners who followed a SCAN training in Belgium, as well as practitioners' and academics' attitudes towards the usefulness and validity of SCAN. The following research questions are answered by an online questionnaire with open-and closed-ended questions: (1) To what extent do practitioners trained in using SCAN in Belgium use the technique in practice?; and (2) What are the prevailing attitudes towards SCAN (and its future use) among practitioners trained in SCAN and academic experts specialized in SCAN, investigative interviewing, or lie/deception detection?
This article is structured as follows. The method section describes participant selection of both practitioners and academics, as well as the content of both the practice-and academics-oriented online questionnaire. The results dig deeper into the use of SCAN by practitioners, as well as the perceived usefulness and future of the technique by academics and practitioners. The article concludes with a general discussion, including limitations of the study and recommendations for future practice and research.

Participants and procedure
Practitioners completed a practice-oriented version of the questionnaire (see Appendix 2), whereas academics completed an academic version of the questionnaire (see Appendix 3). Practitioners were recruited by attendance lists from previous SCAN trainings organized by the Belgian Centre for Policing and Security (CPS). 2 CPS organizes a yearly three day basic SCAN training, followed by a two day more in-depth specialized training, taught by SCAN's founder Avinoam Sapir. All 103 Dutch-speaking participants from 2012 to 2018 of the basic and specialized SCAN trainings were invited to participate in the study. The questionnaire was filled out by 44 practitioners, of which 9 only answered the first couple questions. A sample of 35 practitioners filled out the entire questionnaire, implying a 34% response rate. We used a non-random homogeneous purposive sampling method (Sharma, 2017) to select academics, based on two inclusion criteria: (1) active in the field of criminology and/or legal psychology, and (2) minimum one published article on investigative interviewing, lie/deception detection, or specifically SCAN. In total, 35 academics were contacted to take part in the study. The questionnaire was filled out by 13 academics, implying a 37% response rate. The full sample thus included 48 academics and practitioners.
[Insert Table 2 here] Participants were contacted directly by a personal email and did not receive an email reminder, due to lack of time. To mitigate the risk of non-response, participants were addressed personally and an indication of time to complete the questionnaire was given (McPeake et al., 2014). As depicted in Table  2, slightly more than half of the sample was male. The large majority of practitioners worked at police forces. Most practitioners were very satisfied with the SCAN training they attended and the large majority followed the specialized SCAN training at CPS. A small proportion of academics followed training in SCAN.

Questionnaire
The practice-and academics-oriented questionnaire took each approximately 20 minutes to complete. Before distributing, a draft of the questionnaire was reviewed by police, private practitioners, and academics. Respondents received a personal introduction mail containing information about the study, sampling method, and guarantees of confidentiality and anonymity. Before starting the questionnaire, a box regarding informed consent had to be ticked off, stressing guarantees concerning voluntary participation, the right to withdraw, anonymous reporting of the results and the possibility to contact the researcher. The online questionnaire was designed using The results described below focus on the use, perception and future perspectives of SCAN.

Use
As depicted in

Perception and future perspectives
As seen in Table 4, the overall majority of academics (77%) states that SCAN does not offer an added value during investigations, whereas the majority of practitioners (75,9%) acknowledges the added value of SCAN. With regard to positive aspects of SCAN, practitioners refer to the practical applicability and ability to use SCAN without other techniques. Moreover, they state SCAN provides a different perspective on the statement and allows the interviewer to gather more insight into the written text. The majority of practitioners (72,4%) agrees with the proposition that SCAN contributes to keeping the investigative directions broad and reduces the risk of tunnel vision. One academic states SCAN can contribute to "ringing alarm bells", but should be always weighed up against other information available in a case. Looking into the added value of SCAN during police interviews, some SCAN users argue that the technique allows the interviewer to look at the case differently and dig deeper into certain topics. One respondent mentions the reopening of a case because of SCAN and the discovery of a false statement provided by a victim of sexual violence. On the contrary, one participant refers to a false confession obtained from a suspect with low intelligence as a consequence of using SCAN.
[Insert Table 4 here] A minority of practitioners (34,29%) indicates that they cannot identify negative aspects of SCAN. Others argue that SCAN is time consuming and complex, as well as that it is perceived as offensive by the defence and does not fit into the usual practice of police interviews. What is more, SCAN does not take into account language barriers or the incapacity of some suspects to write a statement. Furthermore, respondents continuously refer to the lack of academic underpinning. Only a small proportion of practitioners (24,1%) claims to be fully abreast of academic research involving SCAN. According to one academic, SCAN is just a "jumble" of items from other methods with no standardization, no interrater reliability, and an increased risk of confirmation bias. One academic illustrates that compared to SCAN, "tossing a coin seems to work just as well to detect lies and deception". The vast majority of academics urges not to use SCAN in practice "the sooner, the better" and to apply other techniques such as CBCA and RM.
Considering the future perspectives on SCAN, the majority of practitioners (86,2%) is rather optimistic. A large proportion of practitioners already using SCAN (82,6%) claims they intend to continue using this technique if possible, either or not complemented with other methods. One respondent specifically mentions other colleagues looking into the statement as well, to counterbalance the low interrater reliability and risk of tunnel vision. A few respondents furthermore suggest the developing of a verbal version, where a written statement does not have to be obtained. Other practitioners (13,8%) are not willing to apply SCAN in the future. Contrary to most practitioners, the majority of academics encourages to ban the use of SCAN, with one academic stating that SCAN "hopefully disappears as soon as possible". According to the majority of academics, an integration of SCAN with other VCA techniques is not desirable. One academic explicitly urges for a "cross-fertilization" between practitioners and academics as this would be in the best interest of both, whether or not for a thorough reform of SCAN.

DISCUSSION
SCAN is an analytic method used by various practitioners across the globe that aims to detect deception in written statements. A full SCAN analysis consists of six phases and presumably sixteen criteria to apply to the written text. As the first qualitative research on SCAN, we surveyed both practitioners and academics on prevailing (future) perceptions of the use(fulness) of the technique. The study gains a unique and thus far only insight into how SCAN is used in practice and perceived by academics and practitioners. The study demonstrates that SCANinsofar that it is usedis applied in a limited and slimmed-down form on suspects, witnesses, and victims in different types of interviews. In particular, the vast majority utilizes a personal selection of SCAN criteria and no SCAN user follows all six phases, mainly because of the complexity, lack of consensus, and difficulty to obtain a written statement. SCAN is thus not used as originally intended and designed, what complements with the lack of clarity on the number and interpretation of SCAN criteria as mentioned in the scoping review. This applied 'personalized version' on the one hand further weakens the low interrater reliability of SCAN. On the other hand, it indicates a possible flexibility of SCAN, that allows interviewers to customize the technique to their or the interviewee's needs. Such downsizing and customization is however not desirable, since extant research already states that SCAN in its full format does not work. Regarding prevailing perceptions towards SCAN, the results clearly reflect a discrepancy between practitioners and academics as assumed in the scoping review. The overall majority of academics consistently dissuades the (further) use of SCAN, while the majority of SCAN users is rather optimistic about the (further) use of the technique. Academics urge for a shift toward other VCA techniques, while the majority of practitioners is not aware of these methods.
Considering both the results and prevailing literature on SCAN, we advise practitioners to abandon SCAN altogether. Extant academic research clearly states that SCAN falters in terms of reliability, validity, standardization and purpose. Based on the results of this study, it is clear that SCAN's applicability in practice also staggers. Besides this new insight, this study furthermore contributes to the state-of-the-art by the claim that the current body of knowledge on SCAN does not reach practice sufficiently. Academic conclusions pointing in the direction of abandonment of SCAN do not seem to be implemented in law enforcement practice. Eager for practice-oriented tools to assess truth or deception in interviews, practitioners have turned to a commercial tool as SCAN instead of academia (Nahari et al., 2019) and apply a dangerous "cherry picking" of criteria. The current unequivocal academic conclusions on SCAN not reaching practice is thus a reason for concern.
When interpreting the results of this study, two limitations are important to note. First, the predominant qualitative nature and small sample size reduce the external validity of this study. Second, since only Belgian practitioners were surveyed, this study was not able to take into account practitioners' perceptions from other countries. Despite this small sample size and predominant Belgian focus, the study's main results seem to reflect an international trend and can be mirrored to other countries as well.
First, the majority of practitioners' motivations why SCAN is not or partially used (e.g. complexity, lack of academic support, lack of clarity on how to use SCAN, time consuming nature, ignoring language barriers, unavailability of written statements), is not country-specific or bound to a specific legal system and may thus also be at stake in law enforcement practice outside of Belgium. Second, the above mentioned discrepancy between academics' negative and practitioners' positive perceptions on the usefulness of SCAN reflects an international trend. Amongst others, Vrij (2018) (1) International Management Review, 15(1), 45-55.