Rapid,remote and responsive research during COVID-19

Abstract

In March 2020, the UK Research and Innovation announced an emergency call for research to inform policy and practice responses to the COVID-19 pandemic. This call implicitly and explicitly required researchers to work rapidly, remotely and responsively. In this article, we briefly review how rapid response methods developed in health research can be used in other social science fields. After outlining the literature in this area, we use the early stages of our applied research into criminal justice responses to domestic abuse during COVID-19 as a case study to illustrate some of the practical challenges we faced in responding to this rapid funding call. We review our use of and experience with remote research methods and describe how we used and adapted these methods in our research, from data gathering through to transcription and analysis. We reflect on our experiences to date of what it means to be responsive in fast-changing research situations. Finally, we make some practical recommendations for conducting applied research in a ‘nimble’ way to meet the demands of working rapidly, remotely, responsively and, most importantly, responsibly.

In March 2020, the UK Research and Innovation (UKRI) responded to the COVID-19 emergency by launching a rapid response fund which would support impactful research across the humanities, social and natural sciences. In this article, we describe and reflect on how we have used rapid response research methods to ‘work nimbly’ (Ledger and Sherlaw-Johnson, 2019), illustrated by examples from our project, which explores the criminal justice system’s response to domestic abuse (DA) during and immediately after the COVID-19 emergency period. We start by describing the context for our research, in terms of the restrictions put in place to limit social transmission of the disease, and the funding landscape which was designed to produce policy-relevant research – all of which required the academic community to work rapidly, remotely and responsively.

Rapid research methods

Manderson and Aaby (1992) trace the development of rapid methods in the health field during the 1980s, linking it to the growing involvement of social scientists in practical disease management. Murray (1999) similarly traces rapid appraisal methods to the 1980s and early 1990s in the movement towards involving communities in assessing and planning for local needs in developed and developing countries (see Annett and Rifkin, 1995; World Health Organization (WHO), 2002). Thereafter, the methodology spread across public health to the social sciences. Stimson et al. (1998) and Coomber (2015) used rapid methods for research on drug use; Young (2016) and Murphy et al. (2018) for research on mental health; and so on.

Rapid research can take many forms and, indeed, has assumed many descriptions over the past 20 years. The United Nations Office for Drug Control and Crime Prevention (1999) prefers the term ‘Rapid Situation Assessment’ to describe short, multidisciplinary, mixed-methods research with an emphasis on interventions, while Pink and Morgan (2013) might make a case for rapid anthropological methodologies or short-term ethnography; Ong et al. (1991) and Ong (1996) use the term rapid appraisal for their work on community health. Whatever term is used, the characteristics of rapid research usually involve short timeframes and team-based research, using a range of data sources (including secondary data) and data collection techniques, being done in a speedy way, with a strong link to timely interventions or recommendations (Vindrola-Padros, 2019). It can include collection of quantitative data (through surveys and reviews of existing data sets), qualitative data (through formal and informal interviews with key informants) and the mapping of affected populations and other phenomena (McNall and Foster-Fishman, 2007). Together, the respective work of McNall and Foster-Fishman (2007), Johnson and Vindrola-Padros (2017) and Vindrola-Padros and Johnson (2020) develops a set of guidelines concerning rapid research methods that we have found useful in navigating a relatively new and fast-developing area of research methodology.

They make a number of key points. First, they stress that data collection and analysis (the ‘reflexive interpretation of findings’) should happen simultaneously in order to speed up research (Vindrola-Padros and Vindrola-Padros, 2017). The second key recommendation is that data requests to respondents should not conflict with professionals carrying out their jobs or conflict with other requests for similar data (McNall and Foster-Fishman, 2007; Vindrola-Padros et al., 2020). The third recommendation was to designate areas of expertise and focus within a, preferably large and experienced, team. In this way, different areas can be explored simultaneously rather than sequentially (Vindrola-Padros et al., 2020). Finally, they suggest that recommendations should be made available to policymakers and practitioners as soon as possible. They observe,

Due to the immediacy of the situation, research in this context demands the sharing of findings in almost real time, requiring a type of data analysis that is not common in the social sciences. It also requires that ‘actionable’ findings are shared. This refers to straightforward recommendations that can be easily understood and translated into changes in policy and/or practice. (Vindrola-Padros et al., 2020: 2)

One would hope, of course, that all social science research would be able to contribute effectively and quickly to easily understood policy and practice recommendations. However, the difference may lie in prioritising the dissemination of findings for effective and speedy impact with policymakers and practitioners rather than the publication of academic literature. Vindrola-Padros et al. (2020) are aware of this as a potential tension in comparison with ‘standard’ academic research and comment that

Our study designs might also be interpreted as instrumental in the sense that all studies sought to produce findings that could be used to make changes to policy and practice, in the first instance, and considered the production of knowledge of interest to academic audiences as a secondary aim. (p. 10)

Vindrola-Padros et al. (2020) acknowledge that these kinds of rapid research approaches are unconventional in their research area (qualitative health research) and in the social sciences generally. While they are not commonplace in criminological or criminal justice research, we suggest that the context of COVID-19, in needing rapid solutions to extremely challenging problems, means that this approach is likely to grow in the next few years.

Rapid funding

The methods that would facilitate rapid research became a vital consideration for applicants to the rapid funding schemes that were put in place in March–September 2020 (https://www.ukri.org/funding/funding-opportunities/ukri-open-call-for-research-and-innovation-ideas-to-address-covid-19/). Dispensing with much of the bureaucracy normally accompanying applications to the UK Research Councils, applicants were asked to provide a short case for support and approximate budget following a tight deadline call for applications. As soon as the call was announced, a team from the University of Liverpool met to discuss how they could use their skills and experience to understand police responses to the increase in DA that was widely feared (and which seemed to have taken place according to support services) as a result of the lockdown restriction which came into force in late March 2020. The team also wanted to see the impact of any delays in dealing with DA cases in the courts resulting from the closure of most of the UK Crown and Magistrates Courts between April and July. The aim of the research was also to be able to identify good practice and make policy and practice recommendations to the criminal justice system.

Between 31 March and 1 September 2020, UKRI received 2500 Rapid Research bids. In June, the team received notice that Domestic Abuse: Responding to the Shadow Pandemic (ESRC ES/V00476X/1) was one of three DA-focused projects that had been recommended for funding. Ninety-two projects were funded across a wide range of areas. Just over half were within the remit of the Economic and Social Research Council (ESRC). The intention to make some impact on the crisis using robust empirically founded and theoretically driven research was clear and frequently reinforced by UKRI. By June 2020, nearly £23m had been committed by the ESRC, and all the commissioned projects were underway. The funders required a quick start-up, the immediate gathering of empirical evidence and the immediate dissemination of preliminary results to criminal justice agencies – the project team were asked to work rapidly.

Working rapidly

Working rapidly requires several practices and systems to come together in a coordinated manner. This is something which cannot be taken for granted. For example, the host universities were asked to allow researchers to begin the research before the official award letter was sent out – this was designed to considerably shorten the set-up period and allow researchers to begin their research activities immediately, and UKRI was prepared to reimburse reasonable research costs before an official award letter could be sent out. ¹ However, not all university accounting systems are able to facilitate this. University research functions, which are stratified and operate at University, Faculty, School and/or Departmental levels, can be complex and can work at different paces depending on the pressures put upon each part of the system. Some universities, which have attracted funding for consultancy work, may have introduced a more streamlined process. It may be that processes of facilitating consultancy work and research council–funded research should be unified. The disjuncture in systems can hinder the quick appointment of a post-doctoral research assistant (PDRA). We avoided some delays by naming the PDRA in the application, and therefore were able to appoint without going through a lengthy appointment process. Nevertheless, it still took time to go through the institutional human resources appointment procedures. This meant that we could not gain all the necessary accoutrements of modern academic research (website, email, assignment of a cost centre code, etc.) as quickly as we would have liked. Lack of an institutional email account made it difficult to send emails to official agencies, the lack of a library card made it slower to construct a literature review and so on. We were fortunate that the School in which we work has a Digital Communications Officer who was able to help create a website very promptly.

Even if approval to apply is given by Heads of Department/School, the award of funding following a short application period means that adjustments must be made to workload and teaching allocations for project staff, which can be contentious and difficult (especially when universities face challenging financial conditions due to COVID-19). Without proper allocation of time to do the research, rapid research can have personal and health costs for those carrying it out. We would recommend that universities and UKRI synchronise their responses to rapid funding mechanisms both during the application procedure and, critically, when funding has been secured. There are many complex behind-the-scenes processes, human resources, compliance, financial and so on, which are necessary for good governance and to comply with internal and external regulations. However, where possible, universities should strive to endure that these do not inhibit the fast take-up of research opportunities and consider developing a fast-track system for projects funded through rapid response schemes.

Ethical practices and philosophies must be at the forefront of research projects, and all aspects of the research and its methodology must be carefully thought through. As with many universities, non-invasive research, as in our proposed project, is normally dealt with expeditiously. This still involves rigorous peer-review and is still time-consuming. For a ‘normal’ grant with a longer lead-in time, this is relatively unproblematic. For a rapid response project, time is valuable, and even a week or two’s delay (which may be necessary and inevitable) was still problematic, not least because funding is usually conditional on ethical approval having been gained. Accordingly, within the framework of rapid response methodology, we adapted our research methods in a way which prioritised effective and speedy research.

Thus, the first research activity carried out while awaiting ethical approval had not been part of the original plan proposed to UKRI. Available ‘online’ data (what Salmons, 2016, refers to as extant online data) became the first focus of our research. Each of the 43 police force websites was analysed for the presence and ease of access to information on DA and information on policing DA during COVID-19. This led to a further exploration of police Facebook pages and the subsequent online publication of a set of recommendations in June 2020 (https://www.liverpool.ac.uk/media/livacuk/law-and-social-justice/3research/Policing,Domestic,Abuse,-,Covid,Website,Reviews.pdf). It may be that we should have started to apply for ethical approval at application stage. Although this seems wasteful of academic time (both in submitting and in reviewing potential proposals which may not be funded), there is a great time-saving in having ethics processes completed at (or near to) the point when funding is confirmed. As soon as we received ethical approval on 9 June 2020 (ref. 7858), questionnaires about police response to DA during the lockdown period were sent out to DA Leads in each of the 43 police forces in England and Wales via the National Police Chiefs Council (https://www.npcc.police.uk) – which enabled rapid and targeted coverage. Respondents gave informed consent to participate, and from the 25 forces who returned a completed questionnaire (58% return rate), 21 (84%) agreed to be interviewed.

This high response rate was probably achieved because the questionnaire was sent out via senior police gatekeepers. We used our gatekeepers to facilitate access to police officers who held institutional memory and knowledge of the local organisational cultures and practices so that we could understand how processes and practices unfolded during the COVID-19 period. Without the endorsement of our senior-level gatekeepers, processes of contacting and arranging access may have become incrementally less rapid, especially when the organisations we were engaged with were under severe resource and time pressures at the time.

It is likely that the high degree of trust between the gatekeepers and the research team was based on relationships developed through past contact and the applicants’ academic track-record. With a project that unfolds over a longer period, it is possible to build a trust-relationship between participants and researchers, and between the academic team and the gatekeepers. In rapid response research, there is less time to gradually develop relationships – they must be pre-existing, or they have to be forged very quickly based on the track-record of individual academics. Had our team not contained experienced and established researchers, we may well have struggled to engage the criminal justice community (Fitch et al., 2004) or to have started the research so promptly. However, having access and the endorsement of high-level gatekeepers are not a panacea. Different units within multi-level complex organisations, such as a large police force, may not simply or quickly carry out the will of their organisation’s senior managers. We found that, although senior officers had requested data to be sent to us, force data and statistical units were not able to do this because they adhered to their own local rules and regulations, or to national data protection measures, which prevented them from forwarding relevant data to us. In some cases, we were advised by police statistical units to submit Freedom of Information requests – which we did and which did elicit data quickly. The people who staffed statistical units were in no way unhelpful; they merely wanted to abide by their own, or governmental regulations, and while it did increase some of those incremental delays that we had wished to avoid, the access and advice provided by both police gatekeepers and statistical departments still resulted in data being provided much more quickly than would normally have been the case.

It may also have been the case that our research had potential benefits to the participants – research which was more critical may have struggled to gain the same level of trust and access. Without some degree of trust and facilitation, it would not have been possible to carry out the research in the way that we did. However, this does not mean that research is not possible, only that the direction of the research is somewhat shaped by the availability, seniority, and depth of existing relationships between participants and gatekeepers, and perhaps more so in rapid research.

So, in our case, the website analysis, literature review and all preliminary analysis of the questionnaire data had been completed by the end of August 2020 (8 weeks from the formal start of the project). We had some statistical data sent to us electronically, and we were able to begin interviewing police DA leads from 1 September. The interviews had to be carried out remotely due to the continuing COVID-19 restrictions which were in place in the autumn of 2020.

Working remotely: interviewing

As soon as was possible, we arranged interviews with DA Leads (Inspectors, Chief Inspectors, Superintendents and Chief Superintendents). In this article, we use the term professional to refer to anyone we interviewed because of the job role they held. In-depth and semi-structured interviews are one of the most used data collection/generation methods in the social sciences, due to their effectiveness, versatility and cost-efficiency (Given, 2008). Interviews, particularly those with professional respondents, also give access to different forms of knowledge, with ‘technical’ and ‘process’ knowledges being most relevant to our project (Bogner et al., 2018; Petintseva et al., 2020). Interviewing which is oriented towards gaining a combination of these kinds of knowledges is referred to by Bogner et al. (2018) as a ‘systematising’ or ‘grounding’ interview. The focus is on the interviewee providing the researcher with systematic information about events, processes and situations. As such, this requires an approach rooted in critical realism, by which we mean the belief that there is a reality that the researcher can gain knowledge of (rather than have direct access to) (Hammersley, 1992), and in contrast to a more constructivist approach (in which knowledge is co-constructed between the interviewer and interviewee).

Until recently, researchers have tended to assume that face-to-face interviews are the best option by default, due to the ability to develop rapport, to see more of the context and status of the interviewee, and through the ability to use and interpret non-verbal cues (Morris, 2015). However, in the first decades of the 21st century, researchers were beginning to offer comment on some of the problems of face-to-face interviewing as a gold standard (see, for example, James and Busher, 2009).

In addition, some researchers were also beginning to question the necessity of conducting interviews face-to-face. Morris (2015), for example, cites studies in which respondents chose remote interviews and goes on to suggest that online interviews offer ‘almost all the advantages of FTF [face to face] interviews’, including being able to conduct interviews ‘in all parts of the world’. O’Connor et al. (2008) also emphasise the increasing legitimacy of online interviews in social science research, highlighting their advantages in being able to overcome problems of time and space, and in being able to contact hard-to-reach groups. In the context of interviewing professional respondents, Bogner et al. (2018) suggest online interviews to be better than telephone interviews, especially given that professional respondents are likely to be familiar with the technology, but they still come down on the side of face-to-face as preferable. Petintseva et al. (2020) address briefly the possibility of interviewing professional respondents remotely (the book was published in 2020 but written pre-COVID), but mainly in cautionary terms. Hensen et al. (2020) refer to the opportunities that remote data collection offers, in terms of efficiency, expense and time (for researchers and participants), and hence for its ability to inform a response to a crisis such as COVID-19.

So, most of the literature published before 2020 is written from the understandable (but what would now perhaps be considered luxurious) viewpoint of being able to choose between either online or face-to-face interviewing. However, although the door to new ways and formats of interviewing was already ajar by 2019, COVID-19 pushed it fully open. From March 2020, for a period of months, face-to-face interviews became impossible, and alternatives had to be found.

Lupton (2020a), in a presentation on qualitative data collection during COVID-19, makes a distinction between traditional methods that have had to evolve into online methods and ‘born digital’ methods, which had always been planned to be online (see also Lupton, 2020b). Our research is somewhere between these two positions: we had not planned to do face-to-face interviews but would not necessarily have chosen to use digital methods were we not ‘forced’ to do so (although our views on this have changed).

Using Salmons’ (2016, 2020) terminology, our interviews can be described as ‘elicited data collection’, using remote technology as the medium. We have chosen to use the term ‘video conferencing’, as our interviews are conducted through platforms enabling visual as well as oral communication (other terminology includes VOIP, or Voice Over Internet Protocol). Archibald et al. (2019), in a paper published pre-COVID-19, draw attention only to the ‘potential’ of VOIP platforms for interviewing in qualitative research and highlight the limited literature in this area and the potential issues with different operating systems, functionality with low bandwidth and security settings. Their advice on using video conferencing during social distancing is, unsurprisingly, to identify the platform that best fits the project.

We considered three potential platforms for conducting our interviews with policymakers and practitioners: Zoom, Microsoft Teams and Skype for Business. These seemed to be the video conferencing platforms already in use by some parts of the criminal justice agencies, and their use spread as working at home directives came into place. Archibald et al. (2019) concluded that researchers and participants were positive about Zoom interviews in terms of features including convenience, ease of use, security and personal connection. Marhefka et al. (2020), in a COVID-era paper, were similarly positive about Zoom for working with people and professionals in the HIV field during social distancing.

The majority of police forces in our study installed MS Teams in March 2020, so we used that platform in police interviews, and for them the security settings of this platform made it preferable. Her Majesty’s Courts and Tribunals Service (HMCTS) switched between different preferred VOIPs between March and August. Interviews with Magistrates Association representatives were conducted via Zoom (secured with meeting password). The Home Office and the Crown Prosecution Service used either Zoom or MS Teams depending on who we were interviewing. The suggestion that professional bodies were unfamiliar with video conferencing use became outdated after the first half of 2020, and the key factor in our choice of platform was that we used any platform preferred by our participants. There was some initial concern in 2020 about the security of video conferencing platforms, so we used all the security measures that were available on the platforms (e.g. passwords and waiting rooms) to mitigate any potential problem.

Our interviews were carried out with respondents predominantly in two environments – offices and homes. These were obviously the result of the personal preference of respondents and partly dictated by the ‘work from home’ guidance issued during the COVID-19 lockdown periods. One might expect that professional environments allow professional contexts (interviews that were more formal) to predominate, but we did not experience different levels of intimacy/distancing/professionalism within any of the environments people chose to be interviewed. A few people chose to obscure their backgrounds, so we did not know where they physically ‘sat’, but again there was no noticeable difference in the ‘feel’ of the interview. This was not the case for the two respondents who could not (for technological reasons) use their laptop camera. This did seem to engender a distance between the interviewer/interviewee. In essence, it changed a face-to-face (online) interview into a phone call, and although the transmission of information was unaffected (they seemed to be similar in character to other interviews), we felt more distanced from the interviewee. For the academic team, the settings for the interviews were a home-office and a kitchen. The home/work environments for interviewers and interviewees were, in fact, very similar. This lessened power-imbalances and gave opportunities for the early establishment of rapport. O’Connor et al. (2008) outline the ways in which researchers have attempted to replicate the process and experience of conducting face-to-face interviews online, including establishing rapport and sharing personal information – commenting on a pet curled up on a sofa behind the interviewer, how nice a desk light looked and so on. These are the usual, short but important, ways of building a friendly working context for interviews which are often taken for granted with face-to-face interviews – the offering of coffee or tea, the complaints about how difficult it was to park the car, and so on. We found that these things naturally occurred as part of our online experience, as did the ‘tech fails’ that we experienced thankfully only on one or two occasions. They were remedied quickly but allowed further bonding through common experience.

There is, of course, the issue that the interviews did not take place in either the interviewees’ or interviewers’ locations, but both, simultaneously. And, of course, we acknowledge that for those using ethnographic methods, the loss of a wider context would be problematic. However, we were working from a position of not having a choice and considering how to maximise the potential of what we could do. However, it was clear that we were all comfortable in our own locale and that, again, was an advantage of online interviewing. We were all on neutral but very personal ground. We would also emphasise here that the methods we discuss were used for interviewing professionals in their professional capacity – interviewing in a different context would require a different set of considerations.

Our interviews lasted between 30 and 70 minutes, with an average time of 55 minutes. We had allowed an hour when arranging the interviews but expected them to last less than this. This suggests to us that the interviews were at least equivalent to face-to-face interviews in terms of potential depth. We also think that this may be longer than a similar face-to-face meeting with a professional would have been.

There were occasional unforeseen interruptions which would not have taken place in face-to-face interviews (not as much anyway). Interruptions were revealing of details of participants’ (both interviewees’ and interviewers’) lives and living conditions, which both equalised and personalised the experience of the interview, and perhaps lessened any power-differential. Interruptions can be a risk to the presentation of self and the professionalisation of the interview. These risks lessen with second and repeat interviews as trust is extended and a bond is formed between the parties. However, the sudden appearance of a neighbour or delivery person at the window, a pet coming into view or some other incidental unanticipated activity must be either acknowledged or ignored. In our case, it often prompted a quick conversational aside while ‘normality’ was resumed. Although it could be seen as disruptive and could, depending on what the interruption is, be seen as unprofessional or disorganised, in the majority of cases it was a welcome glimpse of the sometimes-haphazard nature of home-working and did not disrupt the interview.

There were, however, significant sets of differences in our experience of online interviews when compared with face-to-face interviews. The first set are obvious and relate to logistics. In avoiding travelling to places around the United Kingdom, we could easily have timetabled four or five interviews in a day – journeying (virtually) from East Anglia to the South Coast to Wales in a single day – although we limited it to three per day (and usually two) so that we could leave time for post-interview project discussions and reflections. We could also react to timetable changes which arose through policing emergencies to quickly reschedule meetings, and of course we reduced the financial costs of the whole process considerably by not incurring unnecessary travel costs. Through use of remote platforms, we were able to complete interviews with respondents from a wide range of police forces much more rapidly than we would have done face-to-face.

Second, it became possible for all the project team to attend every interview. It would, of course, have been possible for the whole project team to attend each face-to-face interview, but this rarely happens. We took an early decision to have everyone attend because it seemed more likely to facilitate rapid research – removing the time to bring non-participating team members up to date about the contents of an interview, for one thing – and chose to retain this practice as the group approach seemed to produce better quality interviews, more rounded, with more thought and time taken to answer questions, and the opportunity for different members of the team to pick up on issues as the interview unfolded. Although there was still a question/answer format, the team approach led to different areas being explored within each interview. For two of our interviews, there was more than one police officer attending in addition to our group of three.

It must be stated here that we would not take this team approach to interviewing, nor the same type of remote approach, with participants who were not taking part in a professional capacity or who were vulnerable in any way. In that context, three interviewers would have been overwhelming and possibly created anxiety in interviewees. Given our focus on technical and process knowledges as opposed to experiential knowledge, the team approach suited the approach to talking with professionals and by the same token would not have been appropriate with other groups. Both face-to-face and online interview techniques have advantages and disadvantages. However, we suggest that remote interviews should be the default method for interviewing professional respondents because of some of the advantages they offer, as outlined throughout this article.

We noticed that it took our team some time to learn how to pick up online visual cues and non-verbal signage from each other and from interviewees. These are much easier to pick up in face-to-face interviews. It was also harder, at the beginning, to leave time for the natural gaps, ‘thinking time’ and pauses that are replete in interviews. It is tempting to rush in to fill gaps and silences, which can feel more exaggerated in online interviewing, in a way which did not seem to resonate with ‘normal’ face-to-face interviewing where gaps can feel more natural. We learned not to ‘rush in’ in time and to give time to think about what was being said or starting to analyse what was being said while we were in the interview itself. For this reason, transcribing the data was an important stage for us.

Working remotely: transcription

McNall and Foster-Fishman (2007) recommend that data collection and analysis happen simultaneously, particularly when dealing with qualitative data. To this end, they suggest not transcribing verbatim but simply recording digitally. Vindrola-Padros and Johnson (2020a) review different techniques for dealing with qualitative data in rapid research, focusing on either eliminating transcripts completely or finding ways to produce transcripts faster. Given the rapid nature of our research, we had not planned to carry out time-consuming full verbatim transcription. However, the availability of auto-transcription meant that we were able to produce comprehensive transcripts of every interview. A major advantage of Zoom and MS teams is the ability to record and auto-transcribe the interview (with permission) within the platform itself. In conducting the interviews (and later in storing the transcripts), we used the highest level of security and complied with General Data Protection Regulations (GDPR). All that is required post-interview is the checking and correcting of the automatic transcripts which takes approximately 1 hour per 1-hour interview. This represents a large time-saving over manual transcription, which would have taken 5–6 hours for each hour of a face-to-face interview. In carrying out our preliminary set of interviews, we collected approximately 25 hours (or 200,000 words) of data.

Transcription can be mind-numbingly boring and is always extremely time-consuming, but it has long played an important part in immersing the researcher in their interview data. There is therefore always a trade-off between the use of academic time and the benefits of personal engagement with data. This will be a decision that should be made consciously by the research team following consideration of online transcription facilities. Although the automatic transcription needed post-recording correction to make it perfect, we were able to quickly move on to analysis and found that we were as immersed in the data as we would have been involved in transcribing manually.

Once the transcription process was complete, we used an adapted form of Framework Analysis (Ritchie and Spencer, 1994) to identify themes that had emerged in the interviews. We had arrived at the themes by each researcher familiarising themselves with the body of transcribed data, discussing the data together as a group and identifying themes and sub-themes, which were then arranged in a framework (in Excel). The usual next step would be to apply this coding framework to each transcript – but instead we moved to populating the framework with summaries and quotations from each transcript. This meant that the data from each force were identifiable and comparable with other forces, and that an overview of the whole data set for each theme could be easily obtained.

Of course, there are many other approaches to analysis that could be taken. We chose Framework for its flexibility; Framework was originally designed to be used in applied policy research and lends itself well to team research conducted in a short timeframe (Ritchie and Spencer, 1994). Rapid applied research invariably means making choices about how to use a limited amount of time. Using the Framework approach meant that we focused very much on content (Bogner et al.’s, 2018, systematising or grounding interview we refer to earlier), to the exclusion of, for example, a more detailed discourse analysis.

Working responsively

Rapid research is likely to be commissioned at short notice and requires rapid dissemination of results (often with sets of recommendations for practitioners and policymakers). The production of reports and recommendations after a few months of a project’s start is unusual and there are trade-offs to be negotiated. Well-respected research is often a product of considering all the relevant published literature, collection and analysis of empirical data collection, and writing up results in a considered manner (usually garnering opinion from an academic or research-user audience at conferences and workshops along the way). Rapid research turns this on its head. Researchers are ‘aimed’ at problems – in our case, the response of criminal justice bodies during and after the pandemic – in order to impact on a changing social/political/environmental context.

In the first 6 months of funding, we have produced two academic publications, four working papers, two submissions to government inquiries, a presentation to regional senior police leads and two sets of policy recommendations. Research users – police officers, court managers, victims of DA and the charities that support them – require the delivery of robust recommendations so that practice can be changed and improved as quickly as possible. Implicitly, research commissioners (UKRI, ESRC and other Research Councils) demand the same. The Research Councils will have to demonstrate to government that they have allocated funding wisely and for public benefit. The fast delivery of research results and the transformation of analysis into recommendations involve concentrated work for the research team over shorter periods of time than that taken in normal academic research. Reactive timetabling and the development of good flexible working relationships across the research team are critical. It may be that the opening phases – or indeed every phase – can be time-intensive. In gaining rapid research funding and trying to make a difference with the production of high-quality recommendations, diaries may need to be cleared. However, working responsively also means working responsibly, producing recommendations which are going to improve the situation early in the process in order that the work is effective, but at the same time recognising that good research is ready only when it is ready. The need to produce speedy policy should never be used as an excuse for undermining the quality of the research, and while we recognise, as others have (Fitch et al., 2004), that adequacy rather than scientific perfection may be required in some disciplines or for some situations, we believe that the balance between the need to produce results quickly and to produce high-quality recommendations should always fall decisively in favour of quality. Lancaster et al. (2020) use the term ‘evidence-enough’, when describing the huge changes that COVID-19 has brought about in terms of how evidence is generated and used to support decision-making:

The thing we call ‘evidence’ is being enacted in a different mode, authorised, validated, and made legitimate through non-traditional platforms and practices, with judgements about its relevance for policy entangled within the evolving situation and urgent matters of concern. (p. 481)

Working responsively in the interests of producing quality also means working in environments and with people who are also subject to change. We recognised that the bodies we were working with were experiencing significant and rapidly shifting changes (in policy and personnel) during the research period. Our first round of interviews was conducted with police DA leads. Our follow-up interviews will also be with people in that role, but then may be staffed by different people. By then, some of the officers we originally interviewed will have changed their role in the organisation, some will be on maternity leave and some will have retired. Where possible, we will carry out follow-up interviews with all the original respondents as well as the people new-in-post (we hope to capture the views of experienced and new officers in this way). We have, however, taken care to form a strong relationship not just with the police forces (so that the change of personnel does not damage our access to data) but also with the DA Leads we originally interviewed so that we can carry out follow-up interviews with them even when they have moved to different roles – we may capture further insights in this way. We have also ensured that both the results of the interview analysis and the policy recommendations are of continuing relevance to our research users. We will monitor whether our recommendations are being followed by research users and partners. Since conducting a review of police websites for their accessibility and utility for victims of DA in June, several websites have been redesigned. Other sets of recommendations are more complex and will take longer to implement (not least they will take more discussion with criminal justice bodies before they can be translated into practice).

These may or may not be embedded over a much longer period of time. This does not obviate the need for rapid production of recommendations; it merely recognises that some rapidly produced recommendations can be quickly implemented, others will roll-out over a much longer period of time. In both cases, it is vital to start this process as soon as possible. The early dissemination of results and recommendations is essential to speed up the roll-out process wherever possible. ‘Rapid’ is relative and takes its meaning only with regard to prevailing institutional and socio-political contexts. Overall, the key thing to remember is that carrying out data collection, subsequent analysis and the production of reports/recommendations is exhaustive of resources – yours and the bodies you are working with. To effectively respond to important and rapidly changing conditions, it is important to work nimbly.

Conclusion: working nimbly

Interestingly, there is no recommendation for the timescale over which rapid research ‘should’ take place. This fits with our contention that the rapidity is in the responsiveness, not necessarily the duration of the research. It is not possible to know how long a health (or any kind of) emergency is going to last and therefore perhaps not helpful to define the time period in that sense. However, research funding is given for a fixed amount of time – in our case, 18 months – so is there a place for longitudinal rapid-response research?

Fitch et al. (2004) acknowledge that the period for rapid research to take place is difficult to quantify, not least because of knowing exactly when the start and end points are. That was certainly the case for the Shadow Pandemic. Originally, the research had a clear focus period – the start of the lockdown brought about by COVID-19 regulations (3 months) to the ‘post-lockdown’ period (the next 3 months). The project timetable extended beyond then, in order that we could write up results, make recommendations and so on. However, the pandemic did not follow a linear route, and neither did the social policy which accompanied it. We have witnessed, and may see much more, see-sawing between different levels or tiers of lockdown regulation, in different parts of the country. Rapid research is conducted during the period under examination, rather than after it. Therefore, researchers may finish one piece of rapid research, only to follow it with another period of rapid research, as the incident/episode/situation continues to develop. For example, longitudinal rapid response research might involve returning again to the participants for additional interviews to capture a changing situation; having the opportunity to explore some areas in more depth, through case studies; making recommendations and being able to see whether they are implemented; and looking to establish the efficacy of the implementation process. The extent to which such longitudinal rapid response research would look different from or the same as other forms of action research is perhaps moot. However, the demands of working rapidly, remotely, responsively and nimbly carry with them greater risks than other forms of action research since rapid response work is often conducted within an environment also characterised by rapid change. Thus, rapid response research in demanding nimble working (responding quickly to a changing environment and what that implies for the project planning, resources etc.) also demands a research team willing to take risks in constructing nimble responses to the research process itself. As Lancaster et al. (2020) state, it is also important to take these lessons that ‘illuminate the possibilities and affordances of an emergent and adaptive evidence-making’ forward beyond a crisis (p. 486).