The adaptable law enforcement officer: Exploring adaptability in a covert police context

Adaptability refers to cognitive, behavioural and emotional adjustments that assist in effectively responding to novel and uncertain situations. It is acknowledged as a key attribute of the successful management of dynamic interpersonal interactions. Yet, adaptability remains largely unstudied in the field of psychology and law. Here, we take the first steps to fill this research gap. In Study 1 , university students ( n = 30) acted as ‘agents’ that had to complete three ‘undercover missions’ that required an adaptive response. Adaptability was measured through a self-report scale. In Study 2 , practitioners ( n = 22), experienced with covert policing, watched recordings of the undercover missions from Experiment 1. The practitioners rated the adaptive responses of the agents, as well as their ability to attain the mission objectives. The findings showed that our experimental set-up successfully elicited adaptive behaviour. Practitioners’ ratings of adaptability were strongly related to their ratings of trustworthiness, rapport and


I N TRODUC T ION
When faced with novel or uncertain situations, the ability to adjust behaviour appropriately -the ability to adapt -is an invaluable skill.Adaptability is a central part of naturalistic decision-making (Klein et al., 2014) and has been praised as a necessary condition of expertise (Ward et al., 2018).The importance of adaptability has been highlighted in an array of practical areas, including academic achievement (Martin et al., 2013), teaching (Collie & Martin, 2016), selling behaviour (Spiro & Weitz, 1990) and work performance across different military and federal workplaces (Pulakos et al., 2000).The present study extends previous practical work on adaptability to a law enforcement context, in particular, to covert policing.Specifically, we develop a novel experimental set-up that allows us to systematically elicit and observe adaptive responses.
Adaptability refers to cognitive, behavioural and emotional adjustments that assist in effectively responding to novel and uncertain situational demands, when goals cannot be changed or disengaged (Martin et al., 2013).Adaptability is closely related to the constructs of resilience and coping: all three concern one's ability to manage different situational demands.What primarily distinguishes these three constructs is the type of demand.Both resilience and coping concern adversities -negative or threatening situational demands.In contrast, adaptability refers specifically to one's ability to adjust to situational demands that are novel or uncertain, but not necessarily negative or threatening (Martin et al., 2013).
To examine adaptability in a law enforcement context, we developed an experimental set-up inspired by our observations of the undercover training at the Los Angeles Police Department (October 2016).In its most simple form, it plays with three key features: an objective, an expectation and a violation of that expectation.Specifically, participants take on the role of an undercover agent who has to complete mission objectives during a covert operation (e.g., collect a secret note hidden in the office of a professor).Before each mission, the agent receives a brief casefile providing some background information on the upcoming situation.However, during each mission the agent faces a social encounter that is inherently different from what has been described.To successfully complete their missions, agents must adjust their behaviour to meet the unexpected situation.That is, they must adapt (as defined by Martin et al., 2013).
Initial pilot testing (n = 119) shows that this set-up successfully elicits adaptive behaviour.Specifically, the agents were goal-oriented and highly motivated to accomplish their mission objectives.Self-report measures showed that participants did not expect and could not predict the situational changes.Moreover, to attain their mission objective, agents reported a need to adjust behaviour in response to the changed situational demand.
In addition to an experimental set-up to elicit adaptability, a suitable measure of the construct is also required.Several measures of adaptability have been developed (e.g., van Dam & Meulders, 2021;Griffin et al., 2010;Pulakos et al., 2000).Here, we use Martin et al.'s (2012) adaptability scale.The adaptability scale is a self-report questionnaire that assesses people's ability to constructively regulate their thinking, behaviour and feelings, when faced with novel and uncertain situational demands.Factor analyses suggest that adaptability, as measured by the scale, is composed of two sub-categories.The first, cognitive and behavioural adaptability, refers to one's ability to adapt one's thoughts or actions to situational demands.The second, affective adaptability, concerns one's ability to adapt one's feelings or emotions to situational demands.In the high-school context, where the scale was validated, adaptability was positively correlated with academic achievement and enjoyment of school (Martin et al., 2012), as well as with self-regulation and life satisfaction (Martin et al., 2013).
We opted for Martin et al.'s (2012) adaptability scale for several reasons.First, whereas some scales focus more on the prevalence of adaptive responses (e.g., Griffin et al., 2010), this scale focuses primarily on the adaptive behaviour itself.Second, although developed and validated in an academic setting, the scale offers quite a general measure of adaptability, making it suitable for many contexts.This can be contrasted with the Job Adaptability Inventory (Pulakos et al., 2000), which, for our needs, is overly focused on specific work-place behaviours.Finally, the scale is brief and easily administered, but with adequate psychometric properties.This includes a high Cronbach's α for the global 1-factor model of adaptability; a theoretically coherent lower-level 2-factor model, consisting of (a) behavioural and cognitive adaptability and (b) affective adaptability; and factor structure invariance across sex, age and English and non-English speakers (Martin et al., 2012).

The present studies
The present article consists of two studies.In Study 1, we introduced an experimental set-up designed to elicit adaptive responses.In Study 2, we had police officers, experienced with covert policing, assess the adaptive responses elicited in Study 1.
In Study 1, university students took the role of an 'agent' that had to complete three 'undercover missions'.Adaptive behaviour was elicited by three features: a goal, an expectation, and a violation of that expectation.This violation creates the novel or unexpected situational demand that mock-agents must adapt to in order to attain their mission objective (i.e., their goal).We then examined how agents' self-reported adaptability relates to variables central to the quality of interpersonal interactions, such as rapport and trust -both of which are considered key to effective and humane relationship building (Brimbal et al., 2019).
To further examine adaptability in a law enforcement context, we had the mock-agents interact with other participants -'foils' -who were induced to unknowingly stand between the agent and the agent's mission objective.We also examined these foils' perception of the quality of the interaction (i.e., perceptions of trustworthiness and rapport), and related these perceptions with agents' self-rated adaptability and agents' success in accomplishing mission objectives.
In Study 2, we examined practitioners' perceptions of the mock-agents' adaptive responses.A sample of practitioners, experienced with covert policing, watched video recordings of mock-agents from Study 1.They rated mock-agents' adaptability, as well as other variables central to interpersonal interactions, such as rapport and trustworthiness.

Participants
Participants were recruited as either 'agents' or 'foils'.Agents took the role of an undercover operative who had to undertake three undercover missions.Foils allegedly took the role of a new employee at the University and were to perform simple work tasks.However, the simple work tasks were used as a cover for these participants' true purpose, which was to unknowingly stand between the agent and the agent's mission objective.The participants received payment in the form of vouchers (10 Euro for agents and 5 Euro for foils) or, alternatively, received course credit for their participation.Agents 30 University students (11 females, 19 males) with ages ranging between 19 and 41 (M = 22.0, SD = 4.01) took the role of an agent to perform three covert missions (collect a secret note; secure fingerprints; take a photograph). 1The majority of agents were German (53%) and Dutch (30%).Most agents had a bachelor's degree (57%) or a master's degree (10%).Twenty-eight agents performed all three missions whereas two agents performed the photograph mission only (as four foils did not show up for the other missions).

1
We had originally planned to collect data from at least 40 agents and 120 foils, but due to the corona pandemic the data collection was terminated early to ensure the safety and health of participants and research employees.

Foils
A total of 86 University students (38 females, 48 males) served as foils.Foils were told they were to take part in a study examining what it is like to be a new employee at the University and were randomly assigned to one of three different work tasks.These tasks matched onto the agent's undercover operation, allowing the foil to unknowingly stand between the agent and the agent's mission objective.Eleven females and 17 males (age range 18-25; M = 21.6;SD = 1.77) were assigned to assist a professor (i.e., foils for the secret note mission).Four females and 14 males (age range 18-34; M = 23.0;SD = 4.45) were assigned to act as a student guidance counsellor (i.e., foils for the fingerprints mission).Thirteen females and 17 males (age range 18-30; M = 23.0;SD = 2.69) were assigned to act as a manager of a research lab (i.e., foil for the photograph mission).Four foils did not show up for the experiment: two in the assisting a professor task and two in the guidance counsellor task.Most foils were German (51%) and Dutch (36%).Most foils had a bachelor's degree (64%) or a master's degree (30%).

Design
We used a repeated-measure design (Mission: secret note vs. fingerprints vs. photograph) that was semi-randomized by changing the mission order every day of data collection: Order A (the secret note, the fingerprints, the photograph), Order B (the fingerprints, the photograph, the secret note) or Order C (the photograph, the secret note, the fingerprints).Due to the complexity of the data collection, we opted for a semi-randomized procedure instead of full randomization, so as not to mix up instructions, questionnaires and tasks.One research assistant was tasked with hosting the agent and one research assistant was tasked with hosting the three foils throughout each study session.The study was approved by IRBs at the University of Twente and Lancaster University.Below we describe the covert missions and the corresponding tasks given to the foils (see Appendix for each specific casefile).

The agent's missions and the foil's tasks
The three mock covert missions were designed to elicit an adaptive response on behalf of the agent by controlling a mission objective (i.e., a specific goal the agent had to accomplish), an expectation (i.e., some indication for what might happen during the mission) and a violation of that expectation (i.e., the description of the event was inherently misguiding).The agents were encouraged to commit themselves to achieving the mission objective and were informed that all mission objectives would be attainable, even if it would not seem so at first.
All foils were told that they would participate in a study on being a new employee at the University.Each foil was assigned to one task only and received instructions on their respective work tasks.

The secret note
The agent was instructed to collect a secret note hidden in a book inside a professor's office (objective).To collect the note the agent had to borrow the book from the professor who was unaware of the existence of the note.The agent was informed that the professor was known to be friendly towards researchinterested students and should therefore be willing to lend them the book (expectation).However, when the agent entered the professor's office they would find that the professor was away on a business trip and that his assistant (i.e., the foil) was in the office in his stead (expectancy violation).
The foil was told that the University was considering employing assistants tasked with managing the administration of professors, to allow the professors more time for research.The foil was told their professor was away on a two-week leave and would be back in the office the following week.The foil's tasks included organizing the professor's research papers, cataloguing his books and scheduling his appointments.The foil was told that the professor was very keen on having everything in order upon his return.

The fingerprints
The agent was instructed to collect the fingerprints of a student advisor at the University (objective), as the advisor was suspected of committing fraud.To collect the fingerprints, the agent had to make the advisor hold a paper with the agent's grades during a scheduled consultation meeting (expectation).However, when the agent would request the consultant to hold any items or objects the advisor would put on plastic gloves before touching them (expectancy violation).
The foil was told that the University was considering a new program called 'students consulting students'.The foil was tasked to meet another student (i.e., the agent) and help the student work out a study plan for the next semester.Additionally, the foil was informed about a new health policy to reduce the high number of sick leave among University staff.This initiative ordered all personnel to use plastic gloves when receiving items or objects from students to reduce the spread of viruses and bacteria.The foil could easily access a box with plastic gloves in the case any items would be offered by the student. 2

The photograph
The agent was instructed to take a picture of the face of a research assistant named 'Lucas' (objective) who was suspected of stealing and selling sensitive personal data from research participants.The agent was told that Lucas could be found in the Social Psychology Lab (expectation).However, when the agent arrives, he or she would encounter the research lab manager who informs them that Lukas is sick.But, there are pictures of three research assistants (including Lucas) on the wall, so it would still be possible to complete the mission objective.Moreover, there is a strict no-camera policy in the research lab, meaning that phones are not allowed outside of the pocket (expectancy violation).
The foil was told that the University was considering employing research lab managers tasked with monitoring the ethical handling of research data.Their most pressing task was to reschedule the appointments for Lucas' wine tasting experiment, as Lucas was sick and his appointments had to be distributed to other research assistants.Moreover, the foil was told that the lab office had a zero-use policy on cell phones.This was rationalized by claiming that many students would attempt to take pictures of the schedule for the experiments, and that due to the personal and sensitive information kept in the office it was forbidden to take pictures inside the office.

Procedure
Participants, taking the role of agents, were scheduled to meet a research assistant outside the research lab.The assistant would then escort the agent to the instructions room in which the agent would read the general instructions of the study.After having signed the informed consent forms the agent was provided with brief instructions for the first mission.The agent was then escorted to the office in which the interaction would take place.The agent was told that the study would start when they knocked on the door (they were allowed as much time as they wanted before knocking on the door).As soon as the agent entered the office the research assistant would start timing 2 The majority of data for this study was collected before the coronavirus pandemic.We ran six agents after the corona virus was starting to become a concern in Europe.Importantly, we did not identify a single case in which the agent expected the advisor to put on gloves.
the interaction and monitor the video recording of it on their cell phone outside the office.If the interaction would take longer than 5 min, the research assistant would politely enter the room to request the participants to come to an end within the next minute.When the agent left the office the timer was stopped and the mission concluded, and the agent was escorted back to the instructions room.The agent would return to their laptop and fill in the between-mission questionnaire.When the questionnaire was completed the agent received instructions for the second mission.The same procedure was repeated for the second and third missions (see Figure 1).After the agent had filled in the third between-mission questionnaire they would read an initial debriefing statement before filling in the post operation questionnaire.The agents were then fully debriefed on the study and asked to confirm their consent for using their data.The agents then read another informed consent for including the videos of them in Experiment 2. Finally, the agents were asked not to share the details of the study until the data collection had been concluded, and received their compensation, before being thanked for their participation.
A set of three foils were arranged for each agent.The foils were scheduled to arrive 30 min apart outside their respective office (carefully arranged so they would not encounter the agent).The foils would be seated in front of a computer and read the general instructions of the study.After having signed the informed consent forms the foils were provided with instructions for their specific work tasks.The foils were told that the study would begin when they started working on their tasks and were then left alone in the office.This was timed to be about 5-10 minutes before the agent would knock on the door to the office.After the interaction with the agent the foils were informed that the study was over, before being seated in front of the computer to fill in their respective questionnaire.The questionnaire started with a statement clarifying the deceptive nature of the study and that all details would be explained to the foil after the questionnaire had been filled in.After having filled in the questionnaire the foils were fully debriefed on the true purpose of the study (i.e., that the study was on the adaptability of the agent and that their role was to unknowingly stand between the agent and the agent's objective).No foil voiced objections to this deception.They were then provided with an updated informed consent as well as the informed consent for including their videos in Experiment 2. The foils were asked not to share the details of the study with anyone until the data collection had been concluded.They then received their compensation before being thanked for their participation.
F I G U R E 1 Flow chart of the procedure for the agents

Between mission questionnaire
After each mission, the agents filled in a questionnaire on their perception of the mission.The agents first filled in the domain-specific adaptability scale (Collie & Martin, 2016), which we modified to match the expectancy violation of each mission.The scale included three items measuring cognitive adaptability (e.g., I was able to think through a number of possible options to assist me when I realized the professor would not be present), three items measuring behavioural adaptability (e.g., I was able to seek out new information or useful resources to effectively deal with the assistant [rather than the professor]) and three items measuring affective adaptability (e.g., During the operation, I was able to reduce negative emotions [e.g., social anxiety, feeling awkward] to help me deal with the fact that the professor would not be present).These items were answered on 5-point Likert scales (1 = strongly disagree; 5 = strongly agree).The internal consistency of the scale in the current study was acceptable (Cronbach's raw α = 0.91).
The agents then rated how their expectation of the situation had been violated (e.g., Please indicate how well [x] describes your perception of the fact that the professor was not present).For each item, see Table 1.These items were answered on 5-point Likert scales (1 = strongly disagree; 5 = strongly agree).
In addition, the agents rated whether they found the mission challenging, on a 5-point scale (1 = strongly disagree; 5 = strongly agree).Lastly, the agents answered if they had accomplished the mission objective (e.g., did you collect the secret note?) with a Yes or No.

Post-operation questionnaire
The post-operation questionnaire started with control questions about the participants motivation (e.g., How easy/difficult was it for you to take your role seriously?;How motivated were you to complete your objective in operation 1?) and the need to adjust behaviour (e.g., Did you perceive that you had to adjust your behaviour to complete your objective in operation 1?; Did you adjust your behaviour in operation 1?).All these items were rated on 7-point rating scales (e.g., 1 = Not at all easy; 7 = very easy).The agents were then asked to rate their expectation before each mission (e.g., Before operation 1 started, I believed that I would meet a professor [instead of the assistant]) and rate the predictability of the violation within each mission (Before operation 1 started, I had predicted exactly that the professor would not be present).These items were answered on Likert scales (1 = strongly disagree; 5 = strongly agree).
The agents were then asked about their perception of rapport with the foil by rating six items on the communicative atmosphere and coordination (drawn from Bernieri, 1991), (see Table 2).The rapport items were rated on 5-point Likert scales (1 = strongly disagree; 5=strongly agree).The foil questionnaires The foils first rated two questions on their motivation (e.g., how motivated were you to do your job during your 'day at work'?) on 7-point rating scales (e.g., 1 = Not motivated at all; 7 = very motivated).
The foils then answered questions about the agent's trustworthiness (modified from Colquitt et al., 2007;Mayer & Davis, 1999).Specifically, the foils rated six items on benevolence (e.g., I believe that Kim [all agents used the name Kim] would go out of his/her way to help me), six items on ability (e.g., Kim was very skilled at getting me to agree to his/her request) and four items on integrity (e.g., I believe Kim tries hard to be fair in dealing with others).All trustworthiness items were rated on 5-point Likert scales (1 = strongly disagree; 5 = strongly agree).The internal consistency of the three trustworthiness facets were acceptable (Cronbach's raw α = 0.86 for Benevolence, 0.84 for Ability and 0.86 for Integrity).
The foil then rated their perception of rapport with the agent.The foil rated six items into the communicative relationship and coordination (same items given to the agents, see Table 2).The rapport items were rated on 5-point Likert scales (1 = strongly disagree, 5 = strongly agree).The internal consistency for the rapport measure was acceptable (Cronbach's raw α = 0.91).

Manipulation checks
The agents reported that they took their role seriously (M = 5.03, SD = 1.72) and their missions seriously (M = 5.10, SD = 1.63).Motivation to complete all three missions was high, with ratings above the midpoint of the scale (see Table 3).In addition, agents reported that they perceived a need to adjust their behaviours to attain their mission objectives and, indeed, reported that they did adjust their behaviours (see Table 3).The three missions were perceived as moderately challenging, with average scores near the midpoint of the scale.Finally, for all three missions, the expectancy violation was successful, as agents reported that they had not predicted the specific expectancy violations prior to the start of the mission (see Table 3).

Suitability of the experimental set-up
As noted earlier, adaptability refers to adjustments in response to novel and uncertain situations, rather than negative and threatening situations.For this reason, to elicit adaptability, instead of resilience or coping, the expectancy violation of each mission should be rated as a novelty, an uncertainty, or a change, rather than a threat, an adversity, or a confrontation.Mean ratings on these measures suggest T A B L E 2 Items used to measure rapport

Topic Questions
Atmosphere Foil X was friendly towards me I liked Foil X Foil X is a warm person Coordination My interaction with Foil X was positive My interaction with Foil X was cooperative My interaction with Foil X was focused that this was largely the case.Two of the three expectancy violations received higher ratings on novelty, uncertainty and change, rather than being a threat, an adversity or a confrontation (see Figure 2).The remaining mission-the photograph mission-was rated similarly, with the exception of higher scores on adversity.This suggests there is room for improving the specifics of this mission.
In sum, our experimental set-up largely fulfilled all our criteria for eliciting adaptive responses.First, the agents were goal oriented as each mission had a fixed objective that could not be changed or disengaged, and the agents were highly motivated to complete their missions.Second, the agents perceived that the three missions demanded adaptive behaviour rather than resilience or coping.That is, the expectancy violation of the missions was consistently assessed as a change, a novelty or an uncertainty, more so than being assessed as a threat or a confrontation.However, the expectancy violation of the photograph mission was also assessed as adverse, meaning that the agents found this expectancy violation as rather difficult and/or unpleasant in addition to involving change, novelty and uncertainty.Third, the agents reported that they had to adjust their behaviour to attain their objectives.This was supported by the fact that the agents: (i) could not predict the exact expectancy violations, (ii) perceived that making adjustments would be necessary to attain their objectives and (iii) reported that they did adjust behaviour during the missions.In sum, our experimental set-up successfully incorporated key conditions for eliciting adaptive behaviour.

Interaction measures
Agents rated their adaptability after each mission.Table 4 shows how agents' self-reported adaptability correlates with foils' ratings of the agents' benevolence, ability, integrity (i.e., trustworthiness), as well as the general quality of the interaction (i.e., rapport).Correlations between adaptability and ability and adaptability and rapport were near zero.However, there was a small to moderate negative correlation between adaptability and benevolence, as well as a small to moderate negative correlation between adaptability and integrity, though neither of these were statistically significant.Against what we would have expected, these results suggest that agents who self-report as more adaptable were perceived somewhat less favourably than those who self-report as less adaptable.

Success measures
The agent's self-rated adaptability measures showed only a weak relationship with mission success.For exploratory purposes, we also correlated the two sub-facets of adaptability with success.Stronger correlations with success were seen for the behavioural and cognitive component (r = 0.19, p = .096),Note: The Expectancy Violation measure and the Challenge measure were on a 5-point scale.The remaining measures were on a 7-point scale.
compared to the affective component of adaptability (r = 0.05, p = 0.65).Nonetheless, these correlations are still smaller than the correlations between ratings of agent ability and success, and ratings of rapport and success.
F I G U R E 2 Agents' mean ratings of the three missions

ST U DY 2
The low correlations between adaptability and trustworthiness, and adaptability and rapport, observed in Study 1, could have been a consequence of adaptability being self-rated (by the agent), while trustworthiness and rapport were other rated (by the foil).To address this issue, in Study 2 we had observers -police officers experienced in covert policing -rate agents on all three measures.

Participants
We reached out to personal contacts within covert police units in the Netherlands, the United Kingdom and the United States.We asked our contacts to recruit police officers experienced with covert policing and requested that they take the online survey.Twenty-two officers (4 Females, 17 Males, 1 unknown) with ages ranging between 33 and 58 (M = 46.05,SD = 6.71) participated in this study.However, one officer watched and assessed two of the four videos only.The officers were Dutch (55%), British (32%) and American (10%) and had between 4 to 40 years (M = 23.24,SD = 8.97) of police experience.
Fourteen of the officers volunteered information on their experience(s) with covert policing, which included working as source handler and/or controller (15), working with surveillance and/or intelligence gathering ( 13) and working as an undercover officer (6).The officers received no compensation for participating in this study.This experiment was approved by IRBs at the University of Twente and Lancaster University.All of the video clips were in English.We used an English version of the survey for the United Kingdom and the United States officers and a Dutch version for the Dutch officers.The survey was translated to Dutch by two investigative psychologists from the National Dutch Police who worked closely with the first author.

Design
The unit of analysis was the video.Twenty-one officers watched a total of four videos covering two missions on a Qualtrics online survey.One officer watched two videos.In total, this results in 86 data points.
For each officer, Qualtrics randomly displayed a set of two videos of the same mission stimuli (e.g., two cases of the secret note mission) and randomly displayed a successful attempt and a failed attempt of that stimuli.Qualtrics then randomly displayed a set of two videos for one of the remaining two mission stimuli (i.e., either the fingerprints or the photograph mission) and randomly displayed a successful and a failed attempt for that stimuli (see Figure 3).
We made a pool of 18 videos across the three mission stimuli.The mission stimuli were limited to three successful attempts and three failed attempts (i.e., nine successful and nine failed attempts in total).This decision was based on the fact that only three agents attained their objective in the fingerprints mission, and we opted to standardize the number of successes/fails for all mission stimuli.For the missions that had more than three successful/failed attempts we randomly selected three stimuli to use in the pool.

Procedure
The officers received a password-protected link to the Qualtrics Survey from the first author or from our police contacts.The officers were informed that their task was to watch four short videos and assess the adaptability of University students engaging in mock covert missions.Importantly, the officers would first watch an example video to get a sense of what the videos would look like (the example video was always on a mission stimulus that would not be included in their survey).After having watched the example video officers were asked to confirm that the video and audio were working.The officers were then informed about key methodological procedures of Experiment 1.That is, we explained that the missions were designed to elicit adaptive responses (i.e., via a goal, an expectation and a violation of the expectation) rather than mirroring all complexities of real undercover work.We also explained that the foils had been unknowingly influenced to stand between the agent and the agents' objective (i.e., making it clear that the foils were not using role-playing scripts).Furthermore, we explained that the ending of the videos would be cropped so that the officers' assessments of the agents would not be influenced by the outcome of the mission (the ending was not cropped for the example videos).
The officers were then informed that their main task was to observe and assess the agents' adaptive behaviour.Specifically, the officers read the following text: By adaptability in this context, we mean the agent's ability to adjust, change or modify their behaviour in order to reach their goal when faced with an unexpected event.That is, we ask you to consider how skilled the agent was at adapting his or her behaviour when pursuing their mission objective.The officers had to confirm that they understood this specific task before continuing.
The officers were next presented with the background information of the first video (see Experiment 1).We summarized the specific casefile and mission objective of the agent as well as the specific work task of the foil.We also provided links to the agents' and foils' original instructions for the specific mission (see supplemental material).The officers then watched the video, rated the quality of the video and filled out the video-specific questionnaire.Before watching the second video of the first set, the officer would again be presented with the same background information and instructions.The officer then rated the quality of the video and filled out the second videospecific questionnaire.The same procedure was repeated for the second set of videos.After the officer had filled in the fourth mission-specific questionnaire they filled out the final questionnaire (see Figure 3).

Materials
For each mission, the officers rated (i) the agent's adaptability, (ii) the degree of rapport the agent had with the foil and (iii) the agent's trustworthiness.
The officers first filled in the domain-specific adaptability scale (Collie & Martin, 2016) which we had modified to fit the expectancy violation within each mission.The adaptability scale included the same nine items provided to the agents (see Experiment 1).These items were answered on 5-point Likert scales (1 = strongly disagree; 5 = strongly agree).The internal consistency of the scale in the current study was acceptable (Cronbach's raw α = 0.93).
The rapport scale consisted of three items assessing how positive, cooperative and comfortable the atmosphere between the agent and foil was.The rapport items were rated on 5-point Likert scales (1 = strongly disagree; 5 = strongly agree).The internal consistency for the rapport measure was acceptable (Cronbach's raw α = 0.80).
In addition, the officers rated to what extent they believed the agent succeed in their mission.This was rated on a 7-point scale (1 = the agent will definitely NOT be successful; 7 = the agent will definitely be successful).The officers also rated whether they considered adaptability to be more of a characteristic that people do or do not possess, or a skill that can be learned (1 = a characteristic; 7 = a skill).

R ESU LTS A N D DISCUS SION
The officers were motivated to take the survey (M = 5.81, SD = 1.17), took the study seriously (M = 4.81, SD = 1.54) and found the scenarios to be well-designed for assessing adaptability (M = 4.86, SD = 1.34).When asked whether adaptability was more of a trait or a skill, the officers rated adaptability as falling somewhere between a trait and a skill (M = 3.71, SD = 2.22).

Correlates of observer rated adaptability
For our primary analyses we correlated the adaptability ratings with the ratings of trustworthiness, rapport (between the agent and the foil), as well as ratings of whether participants thought the agent would be successful (Success Rating) and whether they were in fact successful (Success Objective).See Table 5 for all correlations.
Adaptability ratings showed large and statistically significant correlations with trustworthiness and rapport.That is, agents perceived as more adaptable were also perceived as more trustworthy and as having had a more positive interaction with the foil.Insofar as trustworthiness and rapport are essential to an effective interpersonal communication, their high correlations with adaptability provide indirect evidence for the importance of adaptability in covert interactions.
Furthermore, all three of these measures were positively correlated with the success rating.Of note is that of these three measures, adaptability showed the strongest correlation with the success rating.In other words, perceptions of whether the agent would succeed were strongly related to how adaptable the agent was perceived as being.This again provides indirect evidence for the importance of adaptability in covert interactions.
Complicating these findings, correlations of adaptability, trustworthiness and rapport with actual success were close to zero.In fact, adaptability showed a trend in the opposite direction.That none of these variables were related to actual success raises questions regarding the experimental set-up, which will be discussed in the general discussion.

GEN ER A L DISCUS SION
The present study introduced an experimental set-up to assess and measure adaptability in a covert lawenforcement context.The experimental set-up successfully elicited adaptive behaviour, as the agents were goal-oriented, perceived the missions to demand adaptive responses and reported a need to adjust their behaviour to achieve their objectives.In addition to our aim of introducing an experimental setup, we also examined correlates of adaptability ratings.The explorative nature of these analyses mean we focus primarily on trends and strength of relationships (effect sizes), rather than formal significance testing.Any conclusions drawn should therefore be seen as preliminary, requiring hypothesis driven confirmatory studies.

Main findings
In Study 1, agents rated their own adaptability when completing different missions.Their behaviour was subsequently rated by the foil they interacted with during these missions.Agents' self-rated adaptability was associated with both positive and negative outcomes.Exploratory analyses showed that cognitive and behavioural adaptability had a small to moderate correlation with the successful completion of missions, albeit non-significant.This finding suggests that self-rated adaptability might be important when attempting to complete specific objectives during novel and uncertain events.
However, this potential positive finding must be weighed against other results.First, affective adaptability showed virtually no relationship with mission success.Second, and perhaps more worryingly, agents who rated themselves as adaptive were perceived more negatively by the foils.Specifically, higher self-ratings of adaptability were associated with lower ratings of integrity and benevolence, though again, these correlations were non-significant.Furthermore, the foils' ratings of the quality of the interaction, rapport, were the strongest predictor of mission success.This suggests that a stronger focus on relationship building would be more beneficial than adaptability in the types of scenarios we examined.In general, however, the foils' ratings showed low correlations with agents' self-ratings of adaptability.A possible explanation for this was the fact that adaptability was self-rated while the other measures were other rated.
We addressed this issue in Study 2 by having the same individuals -police officers with experience in covert operations -rate the agents' adaptability, trust, rapport and predictions of mission success.Ratings of adaptability showed the strongest relationship with the police officers' predictions of success.This indicates that police officers place considerable value in the construct adaptability, as measured by Martin et al.'s (2012) adaptability scale, when judging performance on the scenarios we examined.Furthermore, ratings of adaptability were also strongly correlated with ratings of trustworthiness and rapport.Insofar as trustworthiness and rapport are valued in police operations (Brimbal et al., 2019), this lends further support for the importance of adaptability in these situations.With that said, the high intercorrelations between ratings of adaptability, trustworthiness and rapport suggest that considerable work remains to disentangle these concepts.
Complicating these results, none of the police officers' ratings correlated with actual mission success.In fact, adaptability showed a small to moderate negative correlation, albeit non-significant.This suggests that higher ratings of adaptability were associated with lower rates of mission success.Relatedly, the police officers' predictions of mission success showed virtually no relationship with actual success.This disconnect suggests that further refinement of the experimental set-up may be required.

Instrumental and relational adaptability
It was quite surprising that agents who rated themselves as adaptive were less successful in establishing rapport and displaying benevolence.However, this might be explained by the fact that agents were given very specific mission objectives to accomplish.That is, since the objectives were instrumental (e.g., collecting a secret note), the agents may have responded instrumentally to the expectancy violation, which could have reduced their attention to making appropriate relational adjustments.Put differently, if the agents had instead been given a relational objective -if they had been tasked with creating a positive relationship with individuals who might be connected with their instrumental objective (rather than attempting to obtain that objective) -we predict self-reported adaptability would correlate positively with rapport and trustworthiness.Although speculative, indirect support for this explanation comes from the result showing that self-reported cognitive and behavioural adaptability had higher correlations with actual success (r = .185),than affective adaptability (r = .052).In sum, we propose that adaptability is specific to the nature of the goal at hand, and that adaptability can be tailored to instrumental tasks or relational tasks (cf.Taylor, 2002).Future research should examine this proposal.
In contrast to self-reported adaptability, observer-rated adaptability showed strong positive correlations with observer rated rapport and trustworthiness.We believe this supports the idea that experienced practitioners might consider operational success to be different to, or more complex than, simply attaining instrumental objectives in the situations modelled in our experimental set-up.That is, when assessing these types of situations practitioners might be guided by the premise; if a positive relationship can be established, then instrumental transactions will follow.This would indicate that instrumental adaptability might not be functional without a positive relationship, at least in situations requiring interactions with others.Importantly, this would not mean that instrumental adaptability is irrelevant in operational settings.Rather, we speculate that it means that relational demands come first and instrumental demands come second.

Eliciting adaptive behaviour
We believe the primary contribution of this article is the development of an experimental set-up to examine adaptability in a law enforcement context.This basic set-up includes three key features: (i) a mission objective that cannot be changed or disengaged, (ii) an expectation of the upcoming situation and (iii) a violation of that expectation.Importantly, as the mission objectives are fixed, it is the expectancy violation that controls the situational change.Hence, with this set-up it is the agent who has to adjust their behaviour to the situation (i.e., to reach their objective) rather than modifying the objective.

The goal-orientation
Generally speaking, the agents' goal attainment was quite low in this study.A potential reason for this is that participants became fixated on goal achievement at the expense of relationship building.Importantly, as this study focused on how agents would make autonomous decisions for how to adapt to novel situational changes, we found it important to avoid providing any suggestions for how they should deal with the upcoming situation.Future research could examine an increased focus and instruction on how to nurture relationships in the modelled scenarios and whether such a focus might improve goal attainment and adaptability.

The expectancy violation
Martin makes the argument that adaptability is different to related constructs, such as resilience, because responses to change, novelty and uncertainty, differ to responses to adversity, threat and confrontation (Martin, 2017;Martin et al., 2013).In this study, we controlled for this by having the agents rate the expectancy violation via these six items.However, by doing this we open up for criticism regarding the expectancy violations of the missions in this study.For instance, the photograph mission was deemed as somewhat too adverse for eliciting distinctly adaptive responses.We nevertheless find it important that these control measures are included in future research so that we can better understand what type of expectancy violation is being modelled and what type of response is being elicited and examined.Furthermore, by altering the type of expectancy violation, researchers should be able to examine an array of situations relevant for law enforcement contexts.

Measuring success
For the present study, foils were blind to the purpose of the study and were free to either grant or deny agents' requests.The benefit of this approach is that it allows for an unbiased measure of agent success.However, this greatly increases the cost and logistical difficulty involved in carrying out the study, and reduces experimental control.Future research could use confederates as foils, so that responses are standardized.This would allow for more controlled assessments of the agents' behaviours.Through consultation with practitioners, it may even be possible to develop a gold standard list of plausible adaptable behaviours that agents' actual behaviours can be compared to (for a similar approach in investigative decision making, see Fahsing & Ask, 2018).This would allow for a measure of adaptive success while still maintaining experimental rigor.

CONC LUSION
Adaptability has long been acknowledged as a key attribute to the successful management of dynamic interpersonal interactions (Martin et al., 2013;Spiro & Weitz, 1990).Its absence from the legal psychology literature is therefore puzzling.In this article, we take the first steps towards a systematic examination of adaptability in a law enforcement context, by introducing an experimental set-up that allows for controlled observations of adaptive responses.Our results support the value of adaptability in accomplishing specific objectives in novel and uncertain situations, but highlight the importance of establishing positive relationships to achieve success.
original or unusual situation A threat A situation likely to cause damage or danger An uncertainty An unsure or unknown situation A confrontation A hostile or argumentative situation

F
Flow chart of the procedure for the officers | 13 THE ADAPTABLE LAW ENFORCEMENT OFFICER Agents' motivation and perception of the expectancy violation T A B L E 3