Empathy in the field: Towards a taxonomy of empathic communication in information gathering interviews with suspected sex offenders

Accordingly, the authors were supplied with the 36 interviews, all of which met the inclusion criteria: 18 were conducted by female police officers, 18 by male officers, all with adult males over the age of 18 years who had been arrested on suspicion of sexual offences against children. Additionally, all interviews had been conducted using the PEACE model (PEACE is a mnemonic for the stages of an investigative interview: Planning and preparation, Engage and explain, Account, Closure, and Evaluation (see Clarke, Milne, & Bull, 2011 ).

First, the audio data were transcribed verbatim by a professional transcriber (approved by both police forces involved). All transcripts were then passed to three coders who independently (and blind to the gender of the police interviewer) read each transcript one or more times. The focus then shifted to open coding , which involved coders working independently to identify empathic concepts within the text, to saturation ( Charmaz, 2006 ), and developing categories that represented their meaning in terms of properties and dimensions ( Corbin & Strauss, 2008 ). Throughout, extensive notes were written to summarise the researchers’ understanding, interpretations, and connections. Working together, using the research notes as a guide ( Corbin & Strauss, 2008 ), categories created in ‘open coding’ were then refined to provide more precise and selective codes for empathy types. Throughout, empathic incidents were compared for similarities, violations, and differences within and between interviews. Although there were some initial disagreements between the three coders about particular categorisations, agreement was reached through a process of critical and constructive debate between coders and the authors.

Empathy 2: Continuer comfort. Continuer comfort examples are displayed in Table 1 (including the preceding statement made by the interviewee). Continuer comfort concerned the same verbal offerings as above (comfort & refreshment breaks, time out, time with legal representative etc.). However, this type of empathy occurred only in response to empathic opportunities concerning the difficulties that the interviewee was experiencing. The interviewer's responses are referred to as empathic opportunity continuers (Oxburgh & Ost, 2011 ), and on this occasion the continuers concerned comfort (as above). An empathic opportunity precedes a continuer and is described as, “a statement or description from which a police interviewer might infer an underlying emotion that has not been fully expressed by the suspect”, and a continuer occurs where the interviewer “recognizes and reacts to the opportunity in a manner that facilitates the implied emotion or statement” ( Oxburgh & Ost, 2011, p. 184).

Empathy 3: Spontaneous understanding. Spontaneous understanding occurred when the interviewer spontaneously offered some understanding of the interviewee's situation. As with spontaneous comfort, this was done without any preceding statement or description from the interviewee. For example, acknowledging the difficulty of the interview situation for the interviewee and prefacing questions with empathetic statements (see Table 2).

Empathy 4: Continuer understanding. Interviewers also responded to understanding opportunities concerning the difficulties that the interviewee may be experiencing with empathic understanding continuers (see Table 2 ). The interviewer's responses concerned some understanding of the interviewee's situation, and the difficulties he may be experiencing (as above).

Freidman analysis for the occurrence of the four types of empathy (that is, the number of times the four behaviours were displayed) revealed that interviewers (irrespective of gender) used some empathy types significantly more often than others, χ 2(3, N = 36) = 50.11, p < .001 (see Table 3 for mean occurrence, and mean rank order

Post-hoc analysis with Wilcoxon signed-rank tests revealed that spontaneous comfort occurred significantly more often than spontaneous understanding, Z = -3.44, p = .001, continuer comfort, Z = -4.97, p < .001, and continuer understanding, Z = -4.84, p < .001. Spontaneous comfort occurred significantly more often than both continuer comfort, Z = -4.56, p < .001 and continuer understanding, Z = - 4.16, p < .001. There was no significant difference for the occurrence of the latter two behaviours, p = .978.

A series of one-way analysis of variance were used to investigate the occurrence of empathic behaviours in our sample, as follows. First, we analysed the overall occurrence of empathy per se as a function of interviewer gender. Our results revealed that female interviewers displayed more empathic behaviours ( M female empathy = 18.89, SD = 5.92), 95% CI [16.94, 21.83], than the male interviewers, ( M male empathy = 8.17, SD = 3.65), 95% CI [10.35, 9.98], F(1, 35) = 42.74, p < .001, η2 = .55. We then analysed the occurrence of the four types of empathy as a function of gender, again our results revealed significant differences for three of the four empathic behaviours. Female interviewers displayed considerably more spontaneous comfort ( M female spontaneous comfort = 9.33, SD = 3.85), 95% CI [1.33, 3.01], F(1, 35) = 30.28, p < .001, η2 = .47, spontaneous understanding, ( M female spontaneous understanding = 4.40, SD = 2.07), 95% CI [2.09, 5.29], F(1, 35) = 9.56, p = .004, η2 = .21, and continuer understanding ( M female continuer understanding = 3.28, SD = 1.18), 95% CI [1.69, 2.86], F (1, 35) = 14.04, p = .001, η2 = .29, than male interviewers, ( M male spontaneous comfort = 3.44, SD = 2.41), 95% CI [0.50, 1.61], ( M male spontaneous understanding = 2.61, SD = 2.17), 95% CI [1.53, 3.36], and ( M male continuer understanding = 1.08, SD = 0.72), 95% CI [0.69, 1.42]. No significant differences emerged (applying Bonferoni's correction) across gender for continuer comfort, F = 5.44, p = .029, (M female continuer comfort = 2.17, SD = 1.69, M male continuer comfort = 1.66, SD = 1.69).

Spontaneous empathy (here spontaneous comfort and spontaneous understanding) could be described as ‘standalone’ verbal behaviours, which emanate from the police interviewer in the absence of any preceding ‘trigger’ behaviours (e.g., verbal statements/behaviours and/or physical indicators). On the other hand, continuer empathy is vicarious, that is it is occurs following the actions of another, here the interviewee. Hence, it is sensible to not only measure the occurrence and frequency of interviewer continuer behaviour (as above), but also to consider the presence of continuer opportunities in the sample, which will allow us to understand whether opportunities are being missed, that is whether they have not been continued but rather ‘terminated’ by the interviewer, or whether they simply do not arise. It is entirely possible that the apparent disparity in the occurrence of spontaneous versus continuer empathy may have resulted from reduced empathic opportunities.

In line with previous research, an empathic opportunity is defined as when an interviewee “provides some kind of information, consciously or otherwise, in the hope that the interviewer will respond” ( Oxburgh & Ost, 2011 , p. 184). For the purposes of this research, we confine ourselves to verbal information, only. Overall, the mean number of empathic opportunities per interview (which we do not categorise as being either comfort or understanding) was 7.91, ranging from three to eleven ( SD = 2.15). A one-way analysis of variance for the number of opportunities as a function of interviewer gender (that is the number of opportunities that occurred in the interviews being conducted by female interviewers compared to male interviewers) revealed a significant difference, F(1, 35) = 6.67, p = .014, η2 = .16 (see Table 4 for exemplars of opportunities, and the interviewer's response). The mean number of opportunities for female interviewers was higher ( M = 8.28, SD = 2.02), than for male interviewers ( M = 6.55, SD = 1.97).

This research reports an investigation of empathic communication in the field with suspected sex offenders. We considered types of empathy, whether male and female police interviewers differed in the type and occurrence of empathic behaviours displayed, and missed empathic opportunities. First, we used grounded theory to identify types of empathy – four distinct types emerged, which we have labelled as spontaneous (stand-alone, interviewer initiated verbal utterances) comfort and understanding, and continuer (utterances immediately following an empathic opportunity offered by the suspect) comfort and understanding. Overall, the spontaneous empathy behaviours occurred significantly more often (approximately 6 times during each interview) than the continuer empathies (which on average occurred less than twice in each interview). The former were used consonantly by all of the officers (irrespective of gender) in the sample despite being employed by two different police forces.

The spontaneous empathies appeared rather systematic in nature because typically they simply prefaced verbal information given to the interviewee during the ‘engage and explain’ phase of the interview where processes/procedures, and the topics/events to be discussed were explained to the interviewee. Here, interviewees did not interact verbally with the interviewer, but instead were passive receivers. Hence, it could be argued that this type of non-cognitive empathy is not pro-social, which may serve to reduce, rather than scaffold cooperation. Uninvited, automatic non-experienced empathy might be negatively interpreted and poorly received by the interviewee (e.g., Cuff et al., 2014; Maijala, Ã…stedt-Kurki, Paavilainen, & Väisänen, 2003; Tansey & Burke, 2013 ), and so may not be followed by the same behavioural responses as empathy emanating from perceived empathic opportunities (see Gerdes & Segal, 2009; Geer, Estupinan, & Manguno-Mire, 2000; Hodges & Biswas-Diener, 2007 ).

On the other hand, continuer empathy was less automatic, and more cognitive in nature: it was interviewee driven, and relied on the interviewer recognizing an empathic opportunity and continuing it. Effortful, more complex cognition, such as that displayed by some of the interviewers when recognizing and responding to an empathic opportunity has been found to amplify behavioural responses ( Cuff et al., 2014; also see de De Vignemont & Singer, 2006; Eisenberg & Miller, 1987 ). While the experience that is understood by the interviewer remains that of the interviewee, this type of state empathy is associated with prosocial behaviour (e.g., Hein, Salini, Preuschoff, Basoton, & Singer, 2010; Williams, O’Driscoll, & Moore, 2014 ), and in clinical settings is linked to improved outcomes and levels of satisfaction (see Gleichgerrcht & Decety, 2013).

Despite an average of eight continuer empathy opportunities provided by suspects in each interview, continuer empathy occurred very infrequently. However, that officers did not continue the majority of empathic opportunities raises two possibilities. First, a lack of knowledge/training concerning the use of empathy may have meant that officers simply did not recognize this type of empathic opportunity. Second, that officers ignored, rather than missed these opportunities. However, despite a lack of empathic training, we found that some officers in our sample did exhibit continuer behaviour, which suggests that some police interviewers may be intrinsically more empathic than others.

Why might officers ignore or terminate empathic opportunities? Social constructionist accounts suggest that empathy is not an automatic or inevitable social behaviour (see Lock & Strong, 2010 ). Rather, it is an effortful type of sociality that supports the development of dispositions and actions amid a set of social conditions. The negative underpinnings of such effortful non-automatic empathy imply ‘choosing a side… favouring one who is more closely like one's self’ ( Brown, 2012 ), which may provide a framework for understanding why police officers might ignore and/or terminate empathic opportunities. Continuer empathy may be inherently uncomfortable for police interviewers who have consistently reported finding interview encounters with suspected sex offenders extremely demanding ( Soukara, Bull, & Vrij, 2002 ). As a consequence officers may choose, consciously or otherwise, not to ‘use’ this type of empathy during suspect interviews, which are unchosen relationships within which they may believe there is little room for shared feeling or emotional attunement (also see Oxburgh et al., 2015).

All interviews with suspected offenders are audio recorded in England and Wales, and so police officers are aware that their verbal behaviour may be scrutinised by, among others, their peers. Empathising, other than in an automatic, systematic manner, may be perceived as loosing and/or being seen to lose a sense of one's self, behaviour that officers may not want to display, or seen to be displaying. Future research should seek to investigate police officer's perceptions of investigative empathy, comparing more automatic spontaneous and non-automatic continuer empathy with reference to social constructionist accounts.

Turning to gender, we found that female interviewers displayed significantly more empathic behaviour per se by a considerable margin (more than double the amount). Irrespective of the unique and complex nature of suspect interviews, this finding fits well with literature pertaining to cross gender empathic behaviour ( Baron-Cohen & Wheelwright, 2004 ). However, when considering types of empathy our findings may be less positive than they first appear because this type of non-cognitive, uninvited empathy maybe less pro-social than interview driven empathy (see Brown, 2012; Hein et al., 2010 ). Female interviewers actually displayed significantly more spontaneous empathy than men. Given that this type of empathy may at best be ineffective and at worst counterproductive, this finding has negative proclivities. That said, females were also found to display one of the two types of continuer empathy more often than men (continuer understanding), which might be important for scaffolding cooperation. However, it should be noted that females were also ‘offered’ more empathic opportunities than male interviewers.

Future research should consider investigating whether the types of empathic behaviours that have emerged from this study do affect information revelation. That said, using transcripts and recordings from real life interviews with suspects would be challenging on several counts. For example, deciding what constitutes investigation relevant information is subjective, particularly in the absence of ground truth. However, experimental laboratory investigations of the efficacy of empathic behaviour for information gain, including manipulating communication skill sets, and which draw on the paradigms employed by persuasion researchers, might prove fruitful, but could lack ecological validity.

If it transpires that continuer empathy is important for increasing information gain in suspect interviews, it may be that officers might be psychometrically assessed for empathy prior to being selected for advanced interview training (e.g., see Lietz et al., 2011 ). High trait empathy individuals have been found to display greater experienced empathy than low trait individuals under cognitive load, indicating that empathy may be more automatic for some ( Rameson, Morelli, & Lieberman, 2012 ). Future research should also seek to investigate gender and empathy in suspect interviews, paying particular attention to whether/how gender might affect cooperation and information gain, in both the presence and absence of the two different types of empathy highlighted by this research. To date, this aspect of interviewing/interrogative techniques has received very little attention, particularly in Europe. However, researchers in Japan have recently reported that a relationship-focused interviewing style resulted in increased information gain and more confessions compared with an evidence-focused approach to interviewing ( Wachi et al., 2014).

The limitations of this research must also be borne in mind when interpreting and considering our findings and suggestions. First, because our data was ‘real’ we were unable to control all of the variables that might have had some bearing on our results. For example, we are unaware of whether the suspects being interviewed were guilty or innocent of the offences about which they were questioned. We have not considered information gain with reference to the emergent empathic behaviours for reasons already discussed (subjectivity, no ground truth, no information regarding findings of guilt/innocence). Additionally, here we were unable to contact (due to ethical constraints) either the interviewers or the suspects to establish additional information.

To conclude, this research has gone someway toward developing a taxonomy of investigative empathy and provides a strong indication that blanket definitions of empathy are likely to have masked previous attempts to understand this type of pro-social behaviour during a suspect interview/interrogation. Additional research needs to be undertaken in this domain because there is much to suggest that interviewing might be more effective if interviewers were assisted to recognise and act upon empathic opportunities. This research topic is particularly timely, not only for the purposes of crime investigation, but also for developing effective intelligence-gathering techniques (see Dando & Tranter, 2015 ). Given that many countries worldwide are seeking to develop non-coercive, dynamic, intelligence/information-gathering methods, investigative empathy may support interviewers to develop an operational accord, or special ‘working’ or ‘professional’ relationship with an interviewee, a relationship that is characterised by a willingness to supply accurate information in response to an interviewer's questions.

The authors of this article declare no conflict of interest.