Eliciting information from small cells of sources

Journal of Policing, Intelligence and Counter Terrorism ISSN: 1833-5330 (Print) 2159-5364 (Online) Journal homepage: http://www.tandfonline.com/loi/rpic20 Eliciting information from small cells of sources Pär Anders Granhag, Simon Oleszkiewicz & Steven Kleinman To cite this article: Pär Anders Granhag, Simon Oleszkiewicz & Steven Kleinman (2016) Eliciting information from small cells of sources, Journal of Policing, Intelligence and Counter Terrorism, 11:2, 143-162, DOI: 10.1080/18335330.2016.1215507 To link to this article: http://dx.doi.org/10.1080/18335330.2016.1215507 Published online: 16 Oct 2016. Submit your article to this journal Article views: 26 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rpic20 Download by: [T&F Internal Users], [Arlene Krimmel] Date: 25 October 2016, At: 12:27 JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM, 2016 VOL. 11, NO. 2, 143–162 http://dx.doi.org/10.1080/18335330.2016.1215507 Eliciting information from small cells of sources Pär Anders Granhaga,b,c, Simon Oleszkiewicza and Steven Kleinmand a Department of Psychology, University of Gothenburg, Gothenburg, Sweden; bNorwegian Police University College, Oslo, Norway; cDepartment of Psychology, University of Oslo, Oslo, Norway; dOperational Sciences International, Monterey, CA, USA ABSTRACT ARTICLE HISTORY In past research, the Scharff technique has consistently outperformed different comparison techniques with respect to the elicitation of human intelligence. This study extends previous work by examining the efficacy of the Scharff technique applied to small cells of sources. The sources worked in triads (N = 180), and were given information about a planned terrorist attack. They were then interviewed individually with either the Scharff technique (conceptualised as five tactics) or the Direct Approach (open and direct questions). The two techniques resulted in an equal amount of new information. As predicted, the sources in the Scharff condition underestimated, whereas the sources in the Direct Approach condition overestimated, their own contribution of new information. Furthermore, the Scharff tactics resulted in the sources overestimating the amount of information revealed by their fellow group members, whereas this was not the case in the Direct Approach. The paper advances the knowledge on elicitation techniques on several accounts: with respect to the context (focusing on small cells of sources), measures of efficacy (introducing a new dependent measure) and tactics (introducing novel context-dependent tactics). The outcome of the study marks the Scharff technique as a promising technique for eliciting information in intelligence settings. Received 14 October 2015 Accepted 9 March 2016 KEYWORDS The Scharff technique; Cells of sources; information elicitation; Human intelligence gathering Researchers have only recently started to map the efficacy of interview techniques aimed for collecting human intelligence (HUMINT) (e.g. Evans et al., 2013; Granhag, Cancino Montecinos, & Oleszkiewicz, 2015). This new line has thus far focused exclusively on how to interview individual sources. However, many criminal activities (including terrorist attacks) are often planned and executed by groups (Carrington, 2002; Gill, Jeongyoon, Rethemeyer, Horgan, & Asal, 2014). Hence, the aim of the current study was to examine the comparative effectiveness of intelligence gathering techniques applied to small groups of sources. HUMINT gathering is about the collection of information through the interaction between two or more individuals (Justice, Bhatt, Brandon, & Kleinman, 2010). One specific form of information gathering is information elicitation, where the goal is to collect information in such a manner that the source (a) underestimates his or her own contribution of CONTACT Simon Oleszkiewicz [email protected] burg, P.O. Box 500, Gothenburg 405 30, Sweden © 2016 Department of Security Studies and Criminology Department of Psychology, University of Gothen- 144 P. A. GRANHAG ET AL. new information and (b) remains unaware of the interviewer’s information objectives (Oleszkiewicz, Granhag, & Kleinman, 2014). The Scharff technique Hanns Joachim Scharff (1907–1992) was a very successful interrogator working for the German Luftwaffe during the Second World War (Toliver, 1997). By observing the interrogations conducted at the Luftwaffe’s Intelligence and Evaluation Centre, Auswertestelle West, he discovered that the prisoners (American and British fighter pilots) adopted various strategies to resist the conventional interrogation techniques (Scharff, 1950; Toliver, 1997). Scharff identified three specific counter-interrogation strategies frequently employed by the prisoners; (i) ‘I will not say very much’; (ii) ‘I will try to figure out what they are after, and not provide this information’ and (iii) ‘I will only provide already known information’ (for recent empirical support of the frequent use of these counter-interrogation strategies in modern times, see Alison et al., 2014). In response, Scharff carefully tailored his own elicitation tactics to circumvent these counter-interrogations strategies. The Scharff technique is strongly tied to the psychological concept perspective taking, defined as the cognitive capacity to consider the world from another’s viewpoint, this facilitates the anticipation of other people’s behaviour and reactions (e.g. Galinsky, Maddux, Gilin, & White, 2008). The importance of perspective taking can be understood by introducing a set of interrelated principles: (a) a source typically forms a hypothesis on what and how much information the interviewer already holds, (b) the source’s perception (of the information already held by the interviewer) will affect his or her counterinterrogation strategies and (c) the counter-interrogation strategies employed will affect what and how much information the source will reveal (for empirical support of this model, see Granhag & Hartwig, 2015; Tekin et al., 2015). We argue that this causal model is at the very core of the Scharff technique. The first conceptualisation of the Scharff technique was presented by Granhag (2010), who identified five different tactics: (1) A friendly approach, creating an atmosphere in which the source feels relaxed and comfortable. In essence, the Scharff technique promotes egalitarian conversations in contrast to emphasising authority over the source. (2) Not to press for information, instead of being asked explicit questions, the source is offered opportunities to add information and to respond to claims. (3) The illusion of knowing-it-all, the interaction is opened by the interviewer presenting the already known information to the source. (4) Confirmations/disconfirmations, the interviewer presents claims that s/he wants to have confirmed or disconfirmed. For example, if there exists information pointing in two different directions (e.g. an attack will happen either in Sydney or in Melbourne) the interviewer can present one alternative as a claim for the source to confirm/disconfirm. (5) Ignore new information, when provided with previously unknown information, the interviewer downplays it as unimportant or already known, hiding the fact that the information is of interest. The direct approach In an effort to try to end the use of so-called enhanced interrogations techniques in the US military and intelligence community, the US Army Field Manual (FM 2–22.3, 2006) was JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 145 formally established as the standard for conducting intelligence interviews (Executive Order 13491, January 2009). This manual prescribes that the source should be approached with open-ended and explicit questions posed in a business-like manner (‘Direct Approach’). Only if the source refuses to answer, or if the answers are perceived as untruthful, the interviewer is advised to adopt an alternative approach. The Direct Approach is one of the most commonly used techniques by interrogators in the U.S.A (Redlich, Kelly, & Miller, 2011). Furthermore, without providing any empirical support, the US Army Field Manual (2006) states that 90% or more of the sources cooperate in response to the Direct Approach. For the current study, the Direct Approach is compared against the Scharff technique. Previous studies on the efficacy of HUMINT gathering techniques Until very recently, legal-psychology research was silent with respect to interview techniques for gathering HUMINT (Evans et al., 2014; Justice et al., 2010). However, the development during the last few years has been positive, and the number of studies within this new strand of research is now approaching a critical mass (e.g. Granhag, Vrij, & Meissner, 2014). To examine the comparative effectiveness of information elicitation techniques, Granhag, Cancino Montecinos, et al. (2015) introduced an experimental scenario mirroring some key aspects of HUMINT interactions. The scenario focused on semi-cooperative sources possessing incomplete information about an impending terrorist attack. The sources were motivated to share some portions of the information held, but not willing to reveal it in full. Furthermore, mapping the efficacy of elicitation techniques required the development of a set of novel measures. These new measures captured both the objective outcomes of the actual interview, as well as the sources’ subjective perceptions of the interaction (see Oleszkiewicz, Granhag, & Cancino Montecinos, 2014). There is now a series of empirical studies for which the Scharff technique has been compared against the Direct Approach, and four consistent findings can be identified. First, the Scharff technique elicits more new information compared to the Direct Approach (e.g. Oleszkiewicz, Granhag, & Cancino Montecinos, 2014; Oleszkiewicz, Granhag, & Kleinman, 2014). Second, the Scharff technique is better at masking the interviewer’s information objectives (e.g. Granhag, Oleszkiewicz, Strömwall, & Kleinman, 2015; May, Granhag, & Oleszkiewicz, 2014; Oleszkiewicz et al., 2014). Third, sources who have faced the Scharff technique consistently underestimate how much new information they have revealed, whereas sources who have faced the Direct Approach overestimate how much new information they have revealed (e.g. May et al., 2014; Oleszkiewicz et al., 2014). Fourth, the sources faced with the Scharff technique (vs. the Direct Approach) believe the interviewer to have held more information prior to the interview (e.g. Oleszkiewicz, Granhag, & Cancino Montecinos, 2014; Oleszkiewicz, Granhag, & Kleinman, 2014). Importantly, these findings also hold for sources with different levels of cooperation and different levels of capability to provide information (Granhag, Oleszkiewicz, et al. 2015). The present study The present study advances previous work on the Scharff technique on several accounts. First, it examines the efficacy of the technique in a new context: when applied to small cells 146 P. A. GRANHAG ET AL. of sources. Second, to examine elicitation techniques for small groups, a new dependent measure of efficacy is introduced: the extent to which the individual source over/underestimates the amount of information revealed by the other members in the group. Third, interviewing individual sources who are all members of the same group leaves room for a novel tactic: to use the information elicited from an already interviewed source when interviewing the source next in line. Hence, the current study advances research on elicitation techniques in terms of (a) the context, (b) measures of efficacy and (c) the tactics. The efficacy of the Scharff technique draws on a source’s competing goals to be perceived as willing to cooperate, while simultaneously trying to minimise his or her contribution. If the interviewer presents already known information at the outset of the interview (i.e. with the aim of establishing the illusion of ‘knowing-it-all’), the source cannot simply repeat the information told by the interviewer. In order to be perceived as minimally cooperative, the source must go beyond the interviewer’s story. Hence, the source will provide information that is new to the interviewer. Research has shown that for modern insider spies, the incentive to talk is commonly tied to divided loyalties and less often to financial incentives (Herbig, 2008). Furthermore, it has been argued that by influencing sources to challenge their own beliefs, the interviewer can add to a source’s already existing internal conflict in (Alison, Giles, & McGuire, 2015). In line with this research, a further aim for the current study was to make all sources believe that their fellow cell members had already revealed information. A source who is convinced that his or her group members have revealed information might go in different directions. One direction is relaxing his or her counterinterrogation strategies, shifting from primarily withholding information to being more forthcoming (‘If they talk, I might as well’). Another direction is to stay even closer to an already adopted counter-interrogation strategy (e.g. ‘I will only tell already known information’). A third possibility is that the source is minimally (or not at all) affected by his or her belief about the others’ behaviour (‘They may talk, but I will not’). In brief, both the first and the second direction might aid the interviewer (i.e. resulting in the collection of new information). To inflate the individual source’s estimate of how much information the other group members had revealed, three tactics were used: (1) convincing the source that s/he was the last in line to be interviewed, (2) offering ‘proof’ that the others had indeed talked (by telling a detailed story in order to establish the illusion of ‘knowing it all’) and (3) continuously update the story used for painting the illusion of ‘knowing it all’. That is, the story presented to the source interviewed second in line consisted of (a) the information that the interviewer held from before, plus (b) the new information gathered from the source interviewed first in line. Consequently, the story presented to the source interviewed last in line consisted of (a) the information that the interviewer held from before, plus (b) the new information gathered from the two sources already interviewed. This continuous update was intended to reflect parts of the interplay that takes place between an interviewer and an analyst. We had five hypotheses: Hypothesis 1: The Scharff protocol was expected to result in more new information than the Direct Approach protocol. JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 147 Hypothesis 2: The sources faced with the Scharff protocol were expected to perceive the interviewer as having relatively more information prior to the interview. Hypothesis 3: The sources interviewed with the Scharff protocol were expected to have a relatively less clear understanding of what information the interviewer was after. Hypothesis 4: An interaction effect was predicted between the interview protocols and the objective and subjective amount of new information revealed. Specifically, we predicted the sources interviewed with the Scharff protocol would perceive they had revealed less new information than they objectively revealed, whereas the sources interviewed with the Direct Approach would perceive they had revealed more new information than they objectively revealed. Hypothesis 5: We predicted the sources in the Scharff condition would estimate their fellow cell members to have revealed more information compared to the estimates given by the sources in the Direct Approach condition (5a). Furthermore, we predicted that the sources interviewed with the Scharff protocol would perceive their fellow cell members to have revealed more information than they objectively revealed, whereas the sources interviewed with the Direct Approach would perceive their fellow cell members to have revealed less information than they objectively revealed (5b). The rationale behind this prediction was that, although the interviewer in both interview conditions aimed to convince each source that s/he was the last in the group to be interviewed, it was only for the Scharff condition that the sources received ‘proof’ that their fellow group members had revealed information (i.e. the interviewer presented a story attempting to paint the illusion of ‘knowing it all’). Method Participants One hundred and eighty adults (77 men and 103 women) between 18 and 70 years of age (M = 27.3, SD = 9.2) were randomly allocated to groups of three for the two interview conditions: 30 cells in the Scharff condition and 30 cells in the Direct Approach. The participants were recruited by posters placed at the University’s libraries and departments. The posters described the study as research on ‘interviewing methods’. Each participant was paid an amount equivalent to €16 (approx. $21). The experiment was granted approval by both the Ethical Review Board of the University of Gothenburg (Sweden) and the FBI Institutional Review Board (Washington DC). After the experiment, all participants were fully debriefed. Five participants misunderstood the instructions. They were excluded, along with their cell members, resulting in the exclusion of 15 participants in total (five cells). These 15 participants were not included in the total sample of N = 180. Procedure Part 1: background and planning All sources received identical instructions explaining that they had to take on the role as a source possessing incomplete information about an upcoming terrorist attack. They were presented with a sheet of paper containing both general and specific information about an extremist group that was planning a bombing at a shopping mall. The information was given in the form of a coherent story that was built up by a total of 39 separate pieces of information. Of those 39 pieces, 14 pieces were already known to the interviewer 148 P. A. GRANHAG ET AL. and the remaining 25 pieces were unknown to the interviewer. The sources received no information on how many or which pieces of information the interviewer already held. The sources were told that the study would take between one and two hours. Planning in groups: Each cell of sources was taken to a room and asked to sit down at a table facing each other. The sources received a randomly distributed name tag (John/ Johanna; Kris/Kristine; Fredrik/Frida). They were told that this would be their names throughout the experiment in order to assure their anonymity (the names were also used as codes for the interview order: J = first, K = second, F = third). In front of each source was a paper with identical instructions. The sources were instructed to individually read the paper carefully and to follow the instructions. The instructions told that each source strived to become a full member of a left-wing terrorist group. To achieve this goal, the sources were instructed to help each other memorise the information about the group’s upcoming attack. They were also told that they were (i) not allowed to discuss their own specific instructions with the other group members and (ii) not allowed to share any personal information during the study. Importantly, although all sources received identical instructions, they did not know the exact instructions given to the other group members. To aid in memorising the information a ‘questions box’ was placed in the room. This box held 15 questions about the upcoming attack written on single note cards (e.g. ‘At what exact time will the bomb be placed in the shopping mall?’). The sources were instructed to share the notes evenly amongst each other and repeatedly question one another on these notes until all members knew all answers. The group was instructed that all group members had to fluently know all 15 answers before they were allowed to enter the next phase of the study. The sources were given 20 minutes to read the instructions and memorise the material. Separation of the sources: After memorising the material, the sources were separated and placed alone in a small room with only a piece of paper and a lap-top in front of them. This paper asked the sources to imagine they were being held in custody and suspected of being involved in planning a terrorist activity. They were further informed that the interviewer would call them (via Skype) and question them on the planned terrorist act. We used phone interviews so that the interviewer could read the interview protocols word by word (more information on this in ‘the analysts’ paragraph below). Importantly, phone conversations and face-to-face interviews have not proven to produce different outcomes (Oleszkiewicz, 2016). All sources approved being audio recorded. They were asked to wait patiently since it could take a while before the interviewer would call them (no time frame was provided). In addition, each source was informed that s/he needed to consider a dilemma (successfully used in previous studies, e.g. Oleszkiewicz et al., 2014) which involved striking a balance between (a) not providing too little information in the upcoming interview (in order to be released from custody without being prosecuted) and (b) not providing too much information (as the source had a negative attitude towards the police in general and sympathised with the terrorist group). The sources were told that their compensation could end up somewhere between $17 and $34 depending on their success in striking this balance (all sources received $21). Furthermore, the sources were given the opportunity to lie during the interview. They were informed that a successful lie would contribute positively to their being perceived as willing sources. However, they were also warned that if they were caught lying, they would put JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 149 their compensation at risk. The sources had all background information present during the interview. Part 2: the interview All interviews (for both conditions) started with the interviewer greeting the source in a business-like manner and explaining that s/he was taping the conversation. The interviews lasted on average 6.25 minutes (SD = 2.36; range: 2.27–14.31 minutes). The Direct Approach interviews (M = 4.52, SD = 1.46) were significantly shorter than the Scharff interviews (M = 8.34, SD = 2.17), t(178) = −14.07, p < .001, d = −2.10, 95% CI [−2.46, −1.73]. The difference in length was mainly due to the time needed to build the illusion of knowing-itall. Interview conditions: After the short formal greeting phase, the interviewer addressed the source by his or her temporary name and expressed understanding for the source’s situation: First I would like to point out that I understand that you are in a difficult situation. At the same time, you probably understand that we cannot allow this attack to be carried out. I would also like to make it clear that I have already talked to ‘source A’ and ‘source B’, who I know that you have met. From there on, the two interview conditions differed. For the Scharff protocol, the interviewer explained that s/he would share the information held with the source by saying: ‘so, in order to make this conversation more efficient I will start by telling you some of the information we already have’. For the Direct Approach protocol, the interviewer explained that s/he could not share the information held with the sources by saying: ‘However, as you surely understand, I am not allowed to share the information that we already hold, but I have a few questions I would like you to answer’. The Scharff interview: The Scharff protocol went on with the interviewer trying to create the illusion of knowing-it-all while maintaining a friendly approach. The interviewer told a story containing all the information that s/he already knew about the extremist group and the upcoming attack. Next, the interviewer said, ‘Now, I am sure you understand that we already hold all information that is of direct value, but you are of course welcome to share what you know’. After the source had finished talking, the interviewer implemented the confirmation/disconfirmation tactic by stating two claims that contained correct information. Each claim was followed by a three-second window, leaving room for the participant to comment. Independent of the source’s reaction to a claim (or lack thereof), the interviewer immediately went on presenting the next claim (more on the confirmation/ disconfirmation tactic below). Finally, after the second claim had been presented, the interviewer said ‘Well, there are some scary things going on here, but as you understand we have a pretty good picture of the situation. Would you like to add something before we end this?’ After awaiting the source’s answer, the interview was closed. Importantly, besides using a friendly approach and the illusion of knowing-it-all tactic, the interviewer did not at any time press for information and did not display unfamiliarity with or surprise at any information collected. The story for painting the illusion of knowing-it-all: For the Scharff condition, the sources interviewed first in the cell were faced with a story that contained all information that was known to the interviewer prior to interviewing the first source in the cell (this information 150 P. A. GRANHAG ET AL. concerned mostly details about the terrorist group and their previous activities, see Oleszkiewicz et al., 2014). The second and third sources in the cell were faced with an ‘updated story’. Specifically, the first source was faced with a story consisting of the 14 pieces of already known information. The second source was faced with a story consisting of the 14 pieces of already known information plus the new information provided by the first source. The third source was faced with a story consisting of the 14 pieces of already known information plus the new information provided by the source interviewed first and the source interviewed second. The Direct Approach interview: After the initial introductory phase, the Direct Approach protocol continued with a general invitation, ‘Okay, so you can start by telling me what you know about the situation!’ For this protocol, the interviewer did to not reveal what information was held on the terrorist attack and maintained an atmosphere similar to that of a business meeting. Next, the interviewer asked two direct questions. If the source did not answer a direct question, or did not provide the full answer, the question was repeated once. Hence, a source could be faced with as many as four direct questions (each of the two questions asked two times). The interviewers were trained to ask the two direct questions in a business-like manner and, if the complete answer was not provided, repeat the question with a firmer tone of voice (more on this below). When the direct questions had been posed, the interviewer said, ‘Well, there are some scary things going on here, but we are trying to get a better picture of the situation. Would you like to add something before we end this?’ Finally, the interview was closed. The very same Direct Approach protocol was used for all sources within each cell. Confirmations and direct questions: The information that the interviewer aimed to collect by presenting claims (Scharff technique) or asking direct questions (Direct Approach) concerned eight different information themes: (1) the location of the attack; (2) the date of the attack; (3) what time the bomb would be triggered; (4) how the bomb would be delivered; (5) when the bomb would be delivered; (6) how the bomb would be triggered; (7) where the bomb would be planted and (8) the expertise in the group. The order of the claims/direct questions was the following: The first theme [the location] and second theme [the date] were always used for the source interviewed first in the cell. For the sources that were interviewed second and third, the information themes asked for were dependent on the information that had been provided in the prior interview(s). If the answer to a theme (e.g. the date of the attack) was not provided by the previously interviewed source, then that claim/direct question was also posed to the next source. Importantly, all sources were only presented with two claims/direct questions. Thus, if a claim/direct question was not answered by a previous source that claim/direct question would be one of the two questions presented to the source in the cell interviewed next. Furthermore, if the information on an additional theme was provided by the source (e.g. the source revealed how the bomb would be delivered [theme 4] during an open-ended question), then that theme was considered to be answered and would not be asked for in the next interview. The interviewers: Two men and four women were recruited and trained as interviewers. The major aim of the training was to ensure that (i) the protocols were read fluently and (ii) the sources would be given enough room to answer. A further goal was to make sure that the interviewers agreed not to improvise during the interviews. The interviewers were examined through repeated pre-study rehearsals to make sure that they were consistent in their intonations when reading the scripts. In sum, the interviewers were instructed to JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 151 strictly follow the interview protocols and were not allowed to go beyond what was stated in the protocols. An analysis of the interview transcripts showed that none did. The interviewers each conducted as many Scharff interviews as Direct Approach interviews. The analysts: In order to make proper use of the information provided within each cell, the interviewers worked together with an analyst (the interviewers were also trained as analysts). The analyst’s job was to listen to the live interviews and to use a checklist to tick off the information revealed by the source. That checklist was then coded in Qualtrics (Qualtrics software, Version 2013, http://www.qualtrics.com) which provided the analyst with an output. For the next interview, the analyst provided the interviewer with the two claims/direct questions to be posed for that specific interview. For the Scharff condition, the interviewer also received an updated story to paint the illusion of knowingit-all. This protocol consisted of the information the interviewer had access to prior to interviewing the first source plus the information revealed by the source(s) interviewed in the cell. Furthermore, this updated story also included fabricated information, if such information was provided by the source. Importantly, all interview protocols were standardised (e.g. if the first source revealed information unit A, then this piece of information was added to the interview protocol used for the second source). The analysts’ training constituted of supervised examinations by listening and coding interviews from prior studies examining the Scharff technique. Part 3: post-interview questionnaire After the interviews, the sources filled out four questionnaires on a computer (Qualtrics software, Version 2013). Before filling them out, it was made clear that the role-taking part of the study was over and that the questionnaires should be answered truthfully. The first questionnaire concerned the perceptions of the interview: rated on a 7-point rating scale. The first crucial question related to the perception of the interviewer’s information objectives (How easy/difficult was it for you to understand what information the interviewer was seeking to obtain?) used a scale that ranged from 1 (Very difficult to understand) to 7 (Very easy to understand). In addition, the sources were asked to rate how motivated they were during the interview (1 = not at all motivated, 7 = very motivated). The second questionnaire consisted of a checklist with all information that was available to the source (39 units). The sources were instructed to mark the specific pieces of information they believed they had revealed during the interview. The third questionnaire contained the same checklist (39 units), but now the sources were asked to mark the specific pieces of information that they believed was known to the interviewer prior to the interview. The final questionnaire contained the same checklist (39 units), but focused on the pieces of information the sources believed the other members of the same cell had revealed. Coding of the interviews Each interview was transcribed verbatim and each transcript was then coded in terms of how much information the source revealed (range: 0–39 units). Importantly, each distinct piece of ‘new’ information revealed by the source was identified. The relevance of the pieces of information (in terms of being ‘new’ or ‘old’) was determined by the interviewer’s knowledge. That is, a piece of information was scored as ‘new’ if it was not known to the interviewer prior to the interview. Thus, for the source interviewed first in the cell, the 152 P. A. GRANHAG ET AL. number of new pieces possible to reveal ranged from 0 to 25. The new information revealed by a source was then added to the knowledge of the interviewer. Thus, for the sources interviewed in second and third orders, the number of new pieces of information that was possible to reveal depended on how much new information the previously interviewed sources had revealed. Moreover, to arrive at a proper score with respect to how much new information each individual source believed to have revealed during the interview (i.e. information not previously known by the interviewer), we used as starting point all the information that the source told that s/he revealed during the interview (as marked in the post-interview questionnaire), and then subtracted (a) the information the source believed the interviewer to already have held and (b) the information that the source believed the other sources to have revealed. The information revealed as a result of presenting claims was scored and counted only if the participant clearly affirmed a confirmation (e.g. ‘yes’, ‘true’). Ambiguous responses were not scored (e.g. ‘hmm’). If the source – after listening to a claim – reacted with silence, this was not scored. In addition, we scored the number of lies (explicit fabrications) and ‘incorrect units’ told by the source. Inter-rater reliability One assistant coded all the transcribed interviews, while another assistant coded 30% of the transcripts (selected from the two interview conditions). Based on these, 30% interrater reliability was calculated for the total amount of information revealed (Cohen’s κ = 0.86). The disagreements were settled in a discussion between the coders after that kappa had been calculated. For the remaining 70% of the material, the primary coder’s scores were used (this procedure was repeated for discriminating explicit fabrications from memory errors, Cohen’s κ = .99). Results All dependent variables were initially analysed by the use of Hierarchical Analysis of Variance. As no group effects were detected, we decided to report individual-based inferential output. With respect to the sources’ motivation we obtained a mean score well above the midpoint of the scale (M = 5.97, SD = 1.24), and the sources found the instructions easy to understand (M = 6.04, SD = 1.23). We found no differences between the interview conditions with respect to the sources’ motivation (p = .281), or with respect to understanding of the instructions (p = .628). The lies uttered were too few to conduct any meaningful inferential tests, and the deceptive pieces of information were evenly distributed over the two interview conditions: Of all new information reported in the Direct Approach, only 5% was deceptive; the corresponding number for the Scharff condition was 4%. Manipulation checks For the Scharff condition, 91% of the sources believed that they were the last in the group to be interviewed, and the corresponding number for the Direct Approach was 90%. Notably, no source believed that s/he was interviewed first in the group. Perceived amount of information held by the interviewer as a function of interview order: The sources’ perception of the interviewer’s knowledge in the Scharff condition varied as a JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 153 function of when in line (first, second or third) the source was interviewed, F(1, 87) = 13.35, p < .001, η2 = 0.24. Pairwise Bonferroni tests showed that the sources interviewed second (p = .003) and third (p < .001) in line perceived the interviewer to have been more knowledgeable compared to the sources interviewed first in line. No difference was found between being interviewed second and third in line (p = .277). For the Direct Approach, no difference was found in terms of the perception of the interviewer’s knowledge as a function of when in line the source was interviewed (p = .108). Perceived amount of information revealed by the other cell members as a function of interview order: For the Scharff condition, the perceived amount of information revealed by the other sources varied as a function of when in line (first, second or third) the source was interviewed, F(1, 87) = 5.82, p = .004, η2 = 0.12. Pairwise Bonferroni tests showed that the sources interviewed second (p = .051) and third (p = .004) in line perceived the others to have revealed more information compared to the sources interviewed first. No difference was found between being interviewed second and third in line (p = 1.00). For the Direct Approach, no difference was found with respect to the perceived amount of information revealed by the fellow group members as a function of when in line the source was interviewed, F(1, 87) = 0.80, p = .453, η2 = 0.18. New information revealed Hypothesis 1 stated that the Scharff protocol would result in relatively more new information. An independent-sample t-test showed that the sources in the Scharff condition (M = 3.22, SD = 2.57) revealed more new information than the sources in the Direct Approach (M = 2.74, SD = 2.51), but not significantly so t(178) = 1.26, p = .209, d = 0.19, 95% CI [−0.10, 0.48]. Hence, Hypothesis 1 was not supported. For the Scharff technique, each cell revealed on average 9.67 (SD = 3.55) pieces of new information; the corresponding number for the Direct Approach was 8.20 (SD = 3.68). The sources’ perceptions Interviewer’s prior knowledge: For Hypothesis 2, we predicted that sources interviewed with the Scharff protocol would perceive the interviewer to have been relatively more knowledgeable prior to the interview. An independent t-test confirmed this hypothesis: The sources in the Scharff condition (M = 13.10, SD = 4.71) perceived the interviewer to have held significantly more information prior to the interview than the sources in the Direct Approach (M = 3.06, SD = 4.30), t(178) = 14.93, p < .001, d = 2.23, 95% CI [1.85, 2.60]. The interviewer’s information objectives: Hypothesis 3 stated that the sources interviewed with the Scharff protocol would have a relatively less clear understanding of the interviewer’s information objectives. An independent t-test showed that the sources in the Direct Approach (M = 4.28, SD = 2.09) found it easier to understand what information the interviewer was seeking to obtain than the sources in the Scharff condition (M = 3.70, SD = 1.91). This difference bordered on significance, t(178) = 1.94, p = .054, d = 0.29, 95% CI [−0.00, 0.58]. New information revealed: We predicted an interaction effect between the interview protocols tested and the objective and subjective amount of new information revealed. A mixed ANOVA with the two interview conditions as the between-subjects factor and 154 P. A. GRANHAG ET AL. the amount of new information revealed (Objective, Subjective) as the within-subjects factor was conducted. The Interview × Revealed information interaction was significant, F(1, 178) = 11.86, p = .001, h2p = 0.06. This interaction was analysed further by the use of simple effects tests for each interview condition. The sources in the Scharff condition perceived that they had revealed less new information (M = 2.78, SD = 2.33) than they, in fact, objectively had revealed (M = 3.22, SD = 2.57), F(1, 88) = 2.92, p = .089, η2 = 0.02. This tendency was in the predicted direction, but did not reach significance. In contrast, the sources in the Direct Approach condition perceived that they had revealed significantly more new information (M = 3.57, SD = 3.16) than they objectively had revealed (M = 2.74, SD = 2.51), F(1, 88) = 10.00, p = .002, η2 = 0.05 (see Figure 1). For the Scharff condition, the majority of the sources (54%) underestimated how much new information they had revealed during the interaction, whereas for the sources in the Direct Approach condition, the same held true for only about one-fourth (27%). Hence, we found partial support for Hypothesis 4. Knowledge revealed by the other cell members: We predicted that the sources in the Scharff condition would estimate their fellow cell members to have revealed more information compared to the estimates given by the sources in the Direct Approach condition. An independent t-test confirmed hypothesis 5a, t(178) = 5.09, p < .001, d = 0.76, 95% CI [0.46, 1.06]. Furthermore, we predicted an interaction effect between the interview protocols relating (a) the sources’ perception of the amount of information revealed by the other two members of the group and (b) the objective amount of information revealed by the two members. A mixed ANOVA with the two interview conditions as the between-subjects factor and the other cell members revealed information scores (Objective, Subjective) as the within-subjects factor was conducted. The Interview × Revealed information interaction was significant, F(1, 178) = 13.84, p = .001, h2p = 0.07. The interaction was analysed further by simple effects tests for each interview condition. The sources in the Scharff condition perceived the other two cell members to have revealed significantly more information (M = 9.38, SD = 6.65) than they objectively had revealed (M = 6.44, SD = 3.27), F(1, 88) = 14.80, Figure 1. Illustrating the interaction effect for the subjective and objective scores of information revealed for the sources in the cells within and between the two interview conditions. JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 155 Figure 2. Illustrating the interaction effect for the subjective and objective scores of information revealed by the other two sources in the cell, within and between the two interview conditions. p < .001, η2 = 0.08. In contrast, the sources in the Direct Approach perceived the other two cell members to have revealed less information (M = 4.38, SD = 6.53) than they objectively had revealed (M = 5.46, SD = 3.29) but not significantly so, F(1, 88) = 2.00, p = .159, η2 = 0.01 (See Figure 2). Hence, we found partial support for Hypothesis 5b. A correlational analysis showed a low correspondence between (a) how much information the individual source believed the other two group members to have revealed and (b) the amount of new information revealed by that individual source (Scharff: r(89) = −.06, p = .59, Direct Approach: r(89) = .12, p = .25). A similarly low correspondence was found between (a) how much information the individual source believed the other two group members to have revealed and (b) the degree to which that individual source over/underestimated his or her own contribution of new knowledge (Scharff: r(89) = −.06, p = .57, Direct Approach: r(89) = −.16, p = .14). Discussion Past work on intelligence-gathering techniques has focused solely on eliciting information from single sources. This is a serious limitation considering the abundance of criminal activities planned and executed by groups (e.g. Carrington, 2002) and a critical shortfall, given the increasingly complex operational methods employed by terrorist groups (with attacks primarily carried out by cells rather than individuals). The current study remedies this gap and advances the knowledge on elicitation techniques on several accounts by (a) studying small cells of sources, (b) introducing new measures of efficacy and (c) introducing novel context-specific tactics. New information revealed The Scharff technique did not result in more new information than the Direct Approach, although the difference was in the predicted direction. We believe that this can be 156 P. A. GRANHAG ET AL. explained by considering the information management dilemma the groups, and ultimately each individual source, had to navigate. In previous research, the individual sources’ willingness (motivation) to reveal information has stemmed from their need to receive assistance from the police (e.g. Granhag, Cancino Montecinos, et al., 2015). In contrast, each group in the current study had to carefully strike a balance between (a) providing enough information to avoid prosecution for having engaged in the planning of a terrorist attack and (b) maintaining their commitment (loyalty) to the organisation behind the attack. First, this dilemma was recognised at the group level, which might have resulted in the perception that it could be balanced by a joint effort (in contrast to each source working for individually exclusive causes). Second, almost all sources believed that they were last in line to be interviewed. Contrary to our a priori speculations, this might have resulted in a lowered motivation to reveal new information (‘My fellow group members have already told the interviewer what can and should be told’). Furthermore, this lowered motivation might have been particularly pronounced in the Scharff condition for which the sources received ‘proof’ that their fellow group members had talked. These two factors pull in the same direction and might explain why the sources were relatively restrictive in terms of providing new information. In brief, the sources’ decision to be restrictive in terms of revealing new information left very little room for detecting differences in terms of new information elicited. The sources’ perceptions of the interview Perception of the interviewer’s prior knowledge: In line with our prediction, the sources in the Scharff condition perceived the interviewer to have held significantly more information prior to the interview compared to the sources in the Direct Approach. This finding is not very surprising; reasonably, it is a direct effect of the tactics used in the Scharff condition (first and foremost presenting a story in order to build the illusion of ‘knowing it all’). Perception of the interviewers’ information objectives: In line with our prediction and previous work (for a summary, see Granhag, Kleinman, & Oleszkiewicz, 2016), the sources interviewed with the Scharff technique found it relatively more difficult to understand what information the interviewer sought to collect (this result bordered on significance). We attribute this to the interaction of the Scharff tactics where (i) no explicit questions were asked and (ii) no feedback was provided when new information was revealed. It is likely that these two interview tactics work in tandem to mask the interviewer’s information objectives. In contrast, an interviewer asking explicit questions, which is the case for the Direct Approach, sends clear signals with respect to his or her information objectives. Perception of own knowledge revealed: An effective information elicitation technique should leave the source with the impression that s/he revealed less new information than was the case. For the present study, and in line with the outcome of prior research (e.g. Oleszkiewicz et al., 2014), the pattern for the Scharff technique differed from the pattern for the Direct Approach. That is, the sources in the Scharff condition underestimated their own contribution of new information (although not significantly so), whereas the sources in the Direct Approach condition significantly overestimated their own contribution. Differently put, more than 50% of the sources who had faced the Scharff technique underestimated the contribution of new information, whereas the JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 157 corresponding number for the Direct Approach was 27%. This finding highlights a vitally important contribution of the Scharff technique to the practice of eliciting information from relatively uncooperative sources, especially those who recognise that providing some information to an interviewer is in their interests yet also wish to carefully minimise the value of that information. The Scharff tactics once again demonstrated their value in fostering a perception whereby the sources believe that they had been sufficiently cooperative to advance their interests while, at the same time, leaving most of them with the impression that they had shared less information than they actually had. We attribute this to the Scharff tactics, and in particular to the ‘confirmation’ tactic (i.e. a source who is ‘only confirming a claim’ might not realise that he or she is, in fact, revealing new information). The finding that the sources faced with the Direct Approach significantly overestimated the amount of new information revealed was expected. In short, answering direct questions can easily, and often accurately, be construed by the source that almost all information revealed is new to the interviewer. Perception of information revealed by the other group members: For the Scharff condition, we successfully inflated the sources’ perception of how much information the other cell members had revealed. In fact, the sources in the Scharff condition estimated that their fellow group members had revealed (a) 45% more information than they actually had revealed and (b) two times as much information as compared to the corresponding estimate made by the sources in the Direct Approach condition. We argue that this outcome was caused by three tactical initiatives, all pulling in the same direction. First, as many as 90% of the sources believed that they were last in line to be interviewed. Second, the interviewer opened the interaction by presenting a story which offered ‘proof’ that the others had indeed talked. Our analysis showed that this proof, whether real or misperceived, had a strong impact on the sources’ perception. In brief, only telling a source that his or her fellow members have already been interviewed may change little (if anything); what matters instead is ‘proof’ that their fellow group members have started to talk. Third, the story used to paint the illusion of ‘knowing it all’ was continuously updated, and we did this by using the new information collected from the first source for interviewing the second and, in turn, the third source. Our analysis showed that this ‘update’ contributed to the sources’ inflated perception. A source who (correctly or incorrectly) believes that his or her fellow group members have started to reveal information, and who decides to act on this belief, could either decide to open up (‘If the others have talked, I will as well’), or to close up (‘The others have told what can be told’). It is a challenge for the interviewer to implement ethically sound tactics to ensure that the source acts on his or her belief in ways which promote, and not hinder, information elicitation. We showed that it is possible to inflate a source’s perception of the other group members’ contribution, but failed to send a more precise message with respect to the implications of this finding. For example, the degree to which the sources’ overestimated the others’ contribution did not correlate with the sources’ own contribution of new information. It is for future research to test specific interview tactics which may realise the positive effects that such inflated perception might carry. Relatedly, the potentially positive effects might be difficult to find if limiting the analysis to the interview where the inflated perception was induced. It might be that the effects will show only after the source has had time to reflect on the initial interaction (a reflection which might result in a shift of 158 P. A. GRANHAG ET AL. counter-interrogation strategy). Differently put, to properly map the effects of the sources’ inflated perception of their fellow group members’ contribution, a repeated interview schedule may be necessary. The potential impact of inflating an individual’s perception of the level of cooperation – and the amount of information – offered by fellow cell members must not be overlooked. An interview of a source in a criminal investigative or intelligence setting is, to be sure, a risk management proposition for both the interviewer and the source. As noted previously, there are conflicting objectives at play. The interviewer seeks to gain the maximum amount of truthful information pertaining to very specific interests while, at the same time, masking his/her information objectives. Similarly, the source may seek to provide the minimal amount of information required to present the image that he or she is being cooperative, but not enough (or not sufficiently precise) to enable the interviewer to make a case for the source’s guilt or complicity. As demonstrated in previous studies (see Granhag et al., 2016), the interviewer’s employment of the Scharff tactics both furthers his/her agenda (i.e. maximising information gain while masking objectives) and counters that of the source (i.e. maximising appearance of cooperation while minimising the revelation of new information). However, the employment of the Scharff tactics when interviewing members of a cell offers additional value. While interviewing a single source presents complexities and opportunities, interviewing members of a cell – comprising individuals with shared experiences, objectives, and information – presents both greater complexities and greater opportunities (Granhag, Strömwall, & Jonsson, 2003; Mac Giolla & Granhag, 2014; Vrij et al., 2012). While further research is required to better understand how the bonds among cell members might make the interviewer’s task more difficult, several decades of social psychological research into tactics of influence offer two compelling explanations for the effectiveness of the Scharff tactics in such a setting. First, social proof suggests that individuals facing novel circumstances (such as a source’s experience of being interviewed by authorities) will look for a reference group for cues as to how to act and react (Cialdini, 2001). The literature also adds that when looking for such a reference group, individuals are most influenced by people like themselves (e.g. nationality, ethnicity and organisation affiliation). In the scenario explored within the current study, other members of the cell would thus serve as the most influential reference group. Employing the Scharff tactics, the interviewer can create the circumstances whereby the source believes that (a) he or she is not the first member of the cell to be interviewed and (b) information presented by the interviewer is the product of previous questioning of other cell members. Together, this tactic would effectively leverage the principle of social proof by offering evidence that the other members of the cell are cooperating with authorities and thereby establishing a norm for behaviour – cooperation – where one did not previously exist. Another principle of social influence that might explain the effectiveness of the Scharff tactics in general, and when interviewing cell members in specific, is that of authority. This principle suggests that a person viewed as possessing relevant expertise and/or status can materially influence an individual’s choices. The interviewer employing the Scharff tactics are likely to be viewed by the source as possessing expertise by virtue of the illusion of knowing-it-all. Moreover, the interviewer’s status is concurrently enhanced by the fact that the source believes he/she has already successfully questioned other members of the cell. While principles of social influence have been well established when introduced JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 159 in isolation, they can generate even greater leverage when introduced in combination (Cialdini, 2001). Limitations and future research Both the current study and the Scharff technique as such come with a number of limitations. A first limitation is that the present study is based on a student sample, a sample very different from a typical criminal or terrorist source. We, however, believe that it is reasonable to argue that the Scharff technique might be equally effective in real-life settings. Our argument is based on the assumption that many real-life sources will be more motivated than students to evoke counter-interrogation strategies. As the Scharff technique is tailored to circumvent these counter-interrogation strategies, the technique might be equally effective in real-life settings. In other words, just as the risks of inadvertent disclosure of information faced by a source in a real-world setting would be substantially greater than that experienced by a subject in this study, the influences introduced by the Scharff tactics (e.g. how the illusion of ‘knowing-it-all’ decreases a source’s interest in protecting information that appears to be already known to the interviewer) are also proportionately enhanced. It should, however, be noted that for some situations, it would be a tactical mistake for the interviewer to reveal how much and what intelligence that is held on a certain topic. For example, if the source is not in custody, s/he may inform the cells of individuals to which the intelligence pertains, and/or networks planning future illegal actions could deploy false sources in order to tap how much intelligence is held on them. Second, although the Scharff technique outperformed the Direct Approach on a number of critical measures, our findings also show that there is room for further improvements. As already mentioned, the two techniques examined elicited only a relatively small portion of all new information that was possible to obtain, and a stimulating challenge for future research is how to find ways to increase the collection of new information, without making it obvious to the source that s/he is, in fact, revealing new information. The present paper is the first to address how to most effectively elicit information from small groups of sources. We believe that our paper points to a potentially fruitful strand of research, raising a number of urgent research questions. For example, for the present study, all members of the group shared the same knowledge. Even though it is an easy task to find real-life situations where this is indeed the case, there is an array of situations where the information distribution within the group is uneven (e.g. sources within the same cell may hold different amount of information on different topics). In addition, and as touched upon above, future research might profit from conducting repeated interviews with each member of the cell in order to better map the effects that follow from sources overestimating the amount of information revealed by their fellow group members. Implications and conclusions In the light of the terrorist attacks such as those that unfolded in New York, Madrid, London, Boston and Paris, and the overall increase in the threat of terror worldwide, the need for effective HUMINT techniques is more critical than ever (Brandon, 2011; Loftus, 160 P. A. GRANHAG ET AL. 2011). Moreover, media reports and public statements from a host of governments make it clear that the vast majority of interrogations of high-value detainees involve individuals who had operated within the cellular structure of major terrorist organisations. While the so-called ‘lone wolf’ terrorist – an individual operating independent of a group – has garnered considerable publicity in the last several years, this form of terrorism remains the exception. Despite this operational reality, the preponderance of research has curiously focused exclusively on the interaction between the interviewer and the source as an individual without consideration of the potential mitigating and aggravating effects of the group interaction or identity. We argue that the outcome of the current study is of practical relevance as it can directly inform the strategies and methods employed when interviewing small cells of sources and introduces and examines tactics specifically tailored for these situations. In conclusion, the Scharff technique and the Direct Approach elicited an equal amount of new information. However, by using the Scharff technique, we were able to (a) deflate the sources’ estimate of how much new information they believed to have revealed and (b) inflate the sources’ estimate of how much information they believed their fellow group members to have revealed. The combined evidence speaks to the advantage of the Scharff technique. Disclosure statement No potential conflict of interest was reported by the authors. Funding This work was funded by High-Value Detainee Interrogation Group contract [J-FBI-12–187] awarded to subcontract Pär Anders Granhag, University of Gothenburg, Department of Psychology. Statements of fact, opinion and analysis in the paper are those of the authors and do not reflect the official policy or position of the FBI or the US government. References Alison, L., Alison, E., Noone, G., Elntib, S., Waring, S., & Christiansen, P. (2014). Whatever you say, say nothing: Individual differences in counter interrogation tactics amongst a field sample of right wing, AQ inspired and paramilitary terrorists. Personality and Individual Differences, 68, 170–175. doi:10.1016/j.paid.2014.04.031 Alison, L., Giles, S., & McGuire, G. (2015). Blood from a stone: Why rapport works and torture doesn’t in ‘enhanced’ interrogations. Investigative Interviewing: Research and practice, Special Issue, 7, 5–23. Brandon, S. E. (2011). Impacts of psychological science on national security agencies post-9/11. American Psychologist, 66, 495–506. doi:10.1037/a0024818 Carrington, P. J. (2002). Group crime in Canada. Canadian Journal of Criminology, 44, 277–315. Cialdini, R. B. (2001). Influence: Science and practice. Needham Heights, MA: Pearson. Evans, J. R., Houston, K. A., Meissner, C. A., Ross, A. B., LaBianca, J. R., Woestehoff, S. A., … Kleinman, S. M. (2014). An empirical evaluation of intelligence-gathering interrogation techniques from the United States army field manual. Applied Cognitive Psychology, 28, 867–875. doi:10.1002/acp.3065 Evans, J. R., Meissner, C. A., Ross, A. B., Houston, K. A., Russano, M. B., & Horgan, A. J. (2013). Obtaining guilty knowledge in human intelligence interrogations: Comparing accusatorial and informationgathering approaches with a novel experimental paradigm. Journal of Applied Research in Memory and Cognition, 2, 83–88. doi:10.1016/j.jarmac.2013.03.2002 JOURNAL OF POLICING, INTELLIGENCE AND COUNTER TERRORISM 161 Executive Order No. 13491, Ensuring lawful interrogations, 3 C.F.R. page 4894 (2009). https://www. whitehouse.gov/the-press-office/ensuring-lawful-interrogations FM 2–22.3 (FM 34–52). (2006, September 6). Field manual. Headquarters. No. 2-22.3. Department of the Army. Washington, DC. Galinsky, A. D., Maddux, W. W., Gilin, D., & White, J. B. (2008). Why it pays to get inside the head of your opponent: The differential effects of perspective taking and empathy in negotiations. Psychological Science, 19, 378–384. doi:10.1111/j.1467-9280.2008.02096.x Gill, P., Jeongyoon, L., Rethemeyer, K. R., Horgan, J., & Asal, V. (2014). Lethal connections: The determinants of network connections in the provisional Irish republican army. International Interactions, 40, 52–78. doi:10.1080/03050629.2013.863190 Granhag, P. A. (2010). The Scharff-technique: Background and first scientific testing. Washington, DC: Professional Development Seminar, High-value Detainee Interrogation Group (HIG, FBI). Granhag, P. A., Cancino Montecinos, S., & Oleszkiewicz, S. (2015). Eliciting intelligence from sources: The first scientific test of the Scharff technique. Legal and Criminological Psychology, 20, 96–113. doi:10.1111/lcrp. Granhag, P. A., & Hartwig, M. (2015). The strategic use of evidence (SUE) technique: A conceptual overview. In P. A. Granhag, A. Vrij, & B. Verschuere (Eds.), Deception detection: Current challenges and cognitive approaches (pp. 231–251). Chichester: Wiley-Blackwell. Granhag, P. A., Kleinman, S. M., & Oleszkiewicz, S. (2016). The Scharff technique: On how to effectively elicit intelligence from human sources. International Journal of Intelligence and Counterintelligence, 29, 132–150. doi:10.1080/08850607.2015.1083341 Granhag, P. A., Oleszkiewicz, S., Strömwall, L. A., & Kleinman, S. M. (2015). Eliciting intelligence with the Scharff technique: Interviewing more and less cooperative and capable sources. Psychology, Public Policy, and Law, 21, 100–110. doi:10.1037/law0000030 Granhag, P. A., Strömwall, L. A., & Jonsson, A. C. (2003). Partners in crime: How liars in collusion betray themselves. Journal of Applied Social Psychology, 33, 848–868. doi:10.1111/j.1559-1816.2003. tb01928.x Granhag, P. A., Vrij, A., & Meissner, C. A. (2014). Information gathering in law enforcement and intelligence settings: Advancing theory and practice. Applied Cognitive Psychology, 28, 815–816. doi:10. 1002/acp.3093 Herbig, K. L. (2008). Changes in espionage by Americans: 1947–2007. Monterey, CA: Defence Personnel Security Research Center. Justice, B. P., Bhatt, S., Brandon, S. E., & Kleinman, S. M. (2010). Army field manual 2-22.3 interrogation methods: A science-based review. Unpublished manuscript. Loftus, E. F. (2011). Intelligence gathering post-9/11. American Psychologist, 66, 532–541. doi:10.1037/ a0024614 Mac Giolla, E., & Granhag, P. A. (2014). Detecting false intent amongst small cells of suspects: Single versus repeated interviews. Journal of Investigative Psychology and Offender Profiling. doi:10.1002/ jip.1419 May, L., Granhag, P. A., & Oleszkiewicz, S. (2014). Eliciting intelligence using the Scharff-technique: Closing in on the confirmation/disconfirmation-tactic. Journal of Investigative Psychology and Offender Profiling, 11, 136–150. doi:10.1002/ji.1412 Oleszkiewicz, S. (2016). Eliciting human intelligence: A conceptualization and empirical testing of the Scharff technique. Doctoral dissertation, Department of Psychology, Gothenburg, Sweden. Oleszkiewicz, S., Granhag, P. A., & Cancino Montecinos, S. (2014). The Scharff-technique: Eliciting intelligence from human sources. Law and Human Behavior, 38, 478–489. doi:10.1037/lhb0000085 Oleszkiewicz, S., Granhag, P. A., & Kleinman, S. M. (2014). On eliciting intelligence from human sources: Contextualizing the Scharff-technique. Applied Cognitive Psychology. doi:10.1002/acp. 3073 Redlich, A. D., Kelly, C., & Miller, J. (2011). Systematic survey of the interview and intelligence community. Washington, DC: Final report submitted to the Federal Bureau of Investigation/High-Value Detainee Interrogation Group. Scharff, H. J. (1950). Without torture. Argosy, 39, 87–91. 162 P. A. GRANHAG ET AL. Tekin, S., Granhag, P. A., Strömwall, L. A., Mac Giolla, E., Vrij, A., & Hartwig, M. (2015). Interviewing strategically to elicit admissions from guilty suspects. Law and Human Behavior. doi:10.1037/lhb0000131 Toliver, R. F. (1997). The interrogator; the story of Hanns Joachim Scharff master interrogator of the Luftwaffe. Atglen: Schiffer Publishing. Vrij, A., Jundi, S., Hope, L., Hillman, J., Gahr, E., Leal, S., … Granhag, P. A. (2012). Collective interviewing of suspects. Journal of Applied Research in Memory and Cognition, 1, 41–44. doi:10.1016/j.jarmac. 2011.12.002