Probing for information in online interactions: Cultural differences in perception of rapport, errors and recovery

Background

The past decades have seen a major shift from face-to-face interactions to the use of digital platforms for communication about all aspects of 
human life, from personal, social, political, and professional activities through to crime-related communication. This transformation has had an impact on the activities of law enforcement and security sectors, particularly with respect to intelligence and information gathering. As a result, there has been increased reliance on computer-mediated communication technologies, such as text-based chat systems, online use of forums and other digital spaces to support both overt and covert information-gathering activities (Stanier & Nunan, 2021).

However, just as in face-to-face interactions, communication errors can occur in online interactions. To date, little is known about the impact of such errors on rapport maintenance and future cooperation. In this novel exploratory research, we examined the effects of communication errors and recovery strategies in the context of an online information elicitation attempt on perceptions of rapport and likelihood of future engagement. Additionally, given that many online interactions occur between people from different cultural backgrounds with potentially different communication norms, we also compared responses between two different cultural groups.

Culture, Communication and Information Gathering

As elsewhere, interactions between individuals from diverse cultural backgrounds are becoming increasingly common in law enforcement and human intelligence contexts. While empirical research on the effects of culture on investigative interviews is still limited, different features of cultural expression affect cross-cultural interviews (for a review, see Hope et al., 2022). A key aspect of culture that affects cross-cultural interviewing contexts is communication style.

Through socialisation, individuals learn norms, rules and values of their culture that shape their communication preferences. Hall (1976) proposed a differentiation between low- and high-context communication styles. Low-context communication is characterised by explicit and direct messages, where meaning is conveyed within the content of the communication itself. This type of communication is typically more transparent and relies on clear, unambiguous language. In contrast, high-context communication involves implicit and indirect messages in which meaning is heavily dependent on the social or physical context of the interaction. In high-context communication, much of the meaning is understood through the relationship between the individuals, the setting, and shared cultural knowledge.

Previous research suggests that the fit between individuals’ preferred communication styles can affect interview outcomes (Beune et al., 2010). For instance, interviewers may need to adapt their rapport-building techniques to suit the cultural context, as certain strategies may be more effective depending on the cultural background of the interviewee (Goodman-Delahunty & Howes, 2016). Fieldwork and professional experience further highlight that establishing and maintaining rapport in cross-cultural interactions can be particularly challenging (Hope et al., 2025).

Online Information Gathering

Most of the research examining communication style, rapport and interviewing or intelligence gathering has focused entirely on face-to-face interactions. However, some research suggests that the medium of communication, whether face-to-face or online, has limited detrimental effects on the elicitation of accurate and detailed information in simulated police interviews (Goodman-Delahunty & Howes, 2016). For example, Nash et al. (2014) found that, while communication modality influences certain aspects of interaction, the quality and quantity of information provided by witnesses in investigative contexts remain relatively stable. However, this finding has largely been limited to synchronous communication formats, such as video calls (e.g., Zoom or Skype) and telephone interviews, which allow for real-time verbal exchanges and the occasional use of visual aids.

Less is known about information elicitation outcomes in text-based online formats, which are increasingly used on platforms such as Messenger, WhatsApp, or secure chat applications such as Signal and Telegram. These platforms present unique challenges, including delays in response, reduced nonverbal communication, and the potential for misinterpretation due to the absence of tone and visual context.

However, recent work by Hope et al. (2025) showed that the use of rapport-based strategies in online chat-based interviews can enhance disclosure of intelligence-relevant information. Interestingly, participants interviewed using a rapport-based approach rated higher levels of rapport with the interviewer and did not necessarily detect that they had provided more information than participants in the control interview.

Rapport and Information Gathering

Rapport-building has been empirically and anecdotally linked to positive disclosure outcomes in interrogation, investigative interviewing, and human intelligence debriefings.

“A growing body of scientific research demonstrates that fostering rapport helps create a non-coercive environment conducive to cooperation and information sharing.”

A growing body of scientific research demonstrates that fostering rapport helps create a non-coercive environment conducive to cooperation and information sharing (Abbe & Brandon, 2013; Alison et al., 2013, 2021). Data from laboratory studies, simulation exercises and real-world investigative interviews indicate that rapport-based interviewing encourages adaptive interpersonal behaviour in both suspects and victims, which leads to increased information yield (Alison et al., 2013; Brimbal et al., 2021; Kim et al., 2020). In a professional investigative context, using rapport-building techniques to facilitate a positive interaction between the interviewer and the interviewee, in turn, facilitates information elicitation (Gabbert et al., 2021).

However, the maintenance and recovery of rapport during information gathering interactions has not received as much attention in the existing literature. Insights drawn from research in crisis communication, particularly in hostage negotiations, suggest that communication errors are inevitable during such interactions, and the key to maintaining effective communication lies in the response to these errors (Oostinga et al., 2018a, 2018b). How communication errors are managed can substantially impact the success of attempts at restoring rapport and ultimately influence interview outcomes.

Communication Errors

Communication errors can be categorised into three main types: i) factual errors, where the message includes incorrect factual information; ii) judgement errors, which refer to behaviours that are inappropriate, not aligned with or ‘out of tune’ with the thoughts and feelings of the other person; and iii) contextual errors, which arise when there is a failure to adhere to established protocols or procedures (Oostinga et al., 2018a). Among these, judgement errors – particularly those that occur early in the interview – can be especially damaging to the interviewer-interviewee relationship. In mock-interviews, judgement errors have been shown to lower perceptions of trust and rapport (Oostinga et al., 2018b).

Factual errors
Where the message includes incorrect factual information

Judgement errors
Behaviours that are inappropriate, not aligned with or ‘out of tune’ with the thoughts and feelings of the other person

Contextual errors
Which arise when there is a failure to adhere to established protocols or procedures

Error Recovery Strategies

In response to communication errors, various recovery strategies can be deployed to restore rapport and rectify the communication breakdown. These strategies include: i) accepting the error, which involves accepting an error was made and reassuring the other party that it will not be repeated; ii) apologising, where the error is acknowledged, and the interviewer takes responsibility and apologises for the error; and iii) contradicting, where the error is denied or disputed (Oostinga et al., 2018b). Other strategies include attributing blame to some other source or simply ignoring or moving on without acknowledging the error.

In the wider literature, there is some limited evidence that recovery strategies are perceived differently across various cultural contexts, with individuals from different cultural value orientations evaluating these strategies in distinct ways. For example, individuals from collectivist cultures may view certain recovery responses initiated by the offending party as more favourable than individuals from individualist cultures (Patterson et al., 2006). However, to date, research has not examined cultural differences in the perceived impact of communication errors on rapport and other assessments of an interaction.

Current Research

This research examined the perceived effect of communication errors on rapport in online interactions in the context of a brief interaction between apparent strangers, one of whom has the covert purpose of obtaining personal information.

To explore possible cultural differences, we compared the perceived effects of communication errors in two different cultural groups (low vs high context communication; UK and China, respectively). We also examined recovery strategy preferences following communication error.

What we did

Transcripts presenting an online chat between two people were generated. The core elements of these fictional transcripts were informed by practitioner input and modelled on training materials. Additional elements were then added to the transcripts to reflect the recovery strategies described in Oostinga et al. (2018b). Pilot testing facilitated the further refinement of the stimuli. The general context for the interaction is that one person (X) has struck up an ‘out of the blue’ online chat conversation with another person (Y) and is attempting to obtain some personal information about them (such as location, travel plans).

Each transcript comprised three interaction ‘blocks’ which were part of the same interaction instance between the two parties. In the first interaction block, a judgement error was either made or not made by one party in the interaction. This error reflected an 'over-stepping' by Person X in pressing Person Y for particular details (e.g., their location, nearest airport). 
In the second interaction block, a judgement error was presented to all participants. Here the error reflected a more general 'over-reach' by Person X in asking for personal information (e.g., how much they earn; where they’ve travelled).

In the final interaction block, participants are presented with one of four potential recovery strategies. In the Acceptance interaction, Person X acknowledges they made an error [“I get it, many people don’t like discussing financials and we don’t have to talk about that topic”]. In the Apology interaction, Person X makes an explicit apology for the error [“I’m sorry, I know many people don’t like discussing financials so I shouldn’t have asked you. I didn’t mean to talk about this if you’re not comfortable with the topic.”]. In the Contradiction interaction, Person X takes a more combative approach in their response [“I never understood why people don’t like discussing financials. Everyone should be more comfortable talking about this topic”]. Finally, in the No Recovery condition, Person X makes no remarks relating to the error and just moves on.

Participants were randomly allocated to Initial Error condition (Initial Error vs No Initial Error). All participants were told they would read an online conversation between two people (Person X and Person Y) and that afterwards they would be asked for their views on how the conversation is going. Figure 1 shows the design for this study.

Fig. 1 Schematic showing the design of the study

At the end of the first two blocks, all participants were asked for their opinion on ‘how is the communication between Person X and Person Y going?’ and asked to provide an open response and a rating (7-point Likert scale, 1 = Conversation is not going well at all, 7 = Conversation is going very well). Participants were also asked to describe what aspects of the communication between Person X and Person Y were working and which were not.

In addition, at the end of the second block, participants were asked, in an open response format, to elaborate on which actions Person X could take to recover the communication and reconcile with Person Y. Participants were then presented with four different interview segments, presented in a random order, in which Person X addresses the error using three different error management strategies: accept, contradict or apologise (Oostinga et al., 2018b), or no error management strategy. Participants rated the likely effectiveness of the recovery strategies and were asked to choose which strategy they believed would be most effective to recover the interaction after the error. Finally, participants rated the likelihood that Person Y would accept an invitation for another interaction with Person X.

What we found out

1. Participants detected the initial communication error in the online interaction and described it in remarkably consistent ways

The majority of participants exposed to the initial interaction containing the error noted the (relatively nuanced) judgement error and described it consistently. Across the study, qualitative comments indicated that observers were very sensitive to norms for online behaviour, including cues to what information people are willing to share (or not).

2. Perceptions of rapport were significantly more negative for the interaction in which a judgement error (vs. no judgement error) was committed

The occurrence of a judgement error resulted in more negative perceptions of rapport. Perceived rapport dropped further following the main judgement error — although there did not appear to be a cumulative drop across conditions. In other words, participants who saw an initial error did not necessarily perceive rapport as even worse following a second error than those who just observed the main error.

3. Apology emerged as the recommended recovery strategy most likely to secure subsequent engagement

An apology strategy, which included an explicit apology for the error, was the preferred recovery strategy selected by participants to recover the relationship between the interactants and secure a later engagement. Interestingly, Chinese participants rated the likelihood of the apology resulting in a positive subsequent engagement as significantly higher than UK participants. Research has documented cultural differences in the use, function, expectation and appreciation of apologies between individualist and collectivist cultures (Lee et al., 2012; Lee & Park, 2011; Merolla et al., 2013; Shafa et al., 2017). In light of the current findings, this difference may well extend to online contexts.

4. Limited cultural differences in perceptions of the communication error or rapport were observed

Despite some interaction patterns, there were no differences between the cultural groups with respect to main effects for perceived rapport and other ratings. Qualitative data showed that participants in both cultural groups interpreted and described the communication error in similar ways. The most interesting cultural difference observed pertained to the greater endorsement of an apology strategy to recover the relationship and secure acceptance of an invitation to a later engagement.

Additionally, the results suggest that Chinese participants were more tolerant of subsequent errors than British participants. As such, an early error may be less damaging in high context communication contexts with greater tolerance of some ambiguity in communication - in contrast to low context communication contexts where explicit and direct communication is expected.

Conclusions

The main aim of this research was to examine the perceived effect of communication errors in the context of a brief online interaction between apparent strangers, one of whom has the covert purpose of obtaining personal information. Our findings clearly signal that observers are more than capable of detecting and consistently describing judgement errors and attribute reduced rapport to an interaction when such an error has taken place. The results pertaining to recovery strategies were also relatively straightforward; apologies were perceived as most likely to secure sustained engagement while pretending nothing happened (no recovery strategy) was clearly perceived as the worst possible strategy.

In short, these findings highlight the importance of relationship building in online information elicitation attempts.

The effects of culture were more nuanced and possibly signaled greater tolerance for early error in high context communication contexts. Further research is needed to examine the impact of error timing and recovery in new or developing interactions. It is also worth noting that the interactions were linear online chat-based interactions taking place between apparent strangers with little additional contextual or cultural features. It is also the case that internet-mediated global English might be considered a ‘cyberlingua franca’ and that cultural effects may be attenuated in ‘translanguaging space’ (Im et al., 2022; Wei, 2018) such as that available in computer-mediated communication platforms, including chat-based interactions.

Clearly a limitation of this research is that we were relying on third-party observations and ratings rather than feedback from actual people who were interacting. However, this initial research should allow us to formulate more accurate predictions for future research involving genuine interactions and information elicitation approaches.