A novel research method for workshops and co-production of knowledge: using a secret Facebook group

Background : Co-production is reliant on good communication and consensus between participants but attending in-person meetings and workshops is hard for time-constrained groups such as new mums, who may be geographically dispersed without reliable transport. Discussions with a lay advisory group resulted in the decision to hold a workshop over a secret Facebook group. The aim of this study was to test the feasibility a secret Facebook group for co-production activities. In the example presented here, the population was women with previous gestational diabetes; the topic was physical inactivity; and the purpose was to develop an acceptable intervention to increase physical activity. Methods : The researchers created a secret Facebook group with content similar to an in-perso n workshop that sequentially progressed to develop a programme theory for an intervention. The researcher posted 1-2 times per day for 14 days and members of the group were invited to comment and discuss the content. Feasibility and acceptability of the group analysed using Facebook analytics and a post-workshop survey. Results : Twenty-one participants took part. In total, 521 comments were provided in response to 18 posts of varying types (average = 28.9 comments per post). The total word count of participant comments was 21,142 words. The workshop was viewed positively, with 20 of 21 participants saying they liked the workshop ‘somewhat’ or ‘a great deal’, and felt the group was a safe and open environment to share opinions. When asked if they would take part in something like this again, 15 of 21 said “Yes”. Participants mentioned the format was convenient; it allowed them to reflect on their own experiences and they liked helping research progress. Those who say “yes” said it was difficult finding time and depended on what else was going on. Conclusion : Using a secret Facebook group as a method of co-production or as a workshop in the research process is a feasible and acceptable method. Social media holds significant potential for co-production and involvement in research for populations who are geographically dispersed or time-constrained; with an uncommon condition; or in other circumstances where in-person meetings are either not appropriate or not possible.


Introduction
Co-production of complex public health interventions is seen as an increasing necessity in intervention development for improving acceptability and sustainability of interventions (1). It involves key stakeholders working alongside researchers to co-develop an intervention that is feasible, acceptable and sustainable by taking account of contextual factors (e.g. need, culture, geography, preferences, resources). Co-production in intervention development typically involves inperson workshops or consultations, and face to face interviews or focus groups in which intervention materials are developed and tested with those who will receive, deliver, or resource the intervention, and feedback provided on various aspects.
In-person co-production is an efficient, convenient method of working together, which aids relationship building, but is not suitable for all demographics. In particular, creating a workshop or focus group at a time and location that suits everyone is challenging when working with populations that have an uncommon condition, are geographically dispersed, or have limited opportunities to gather. However, the internet has dramatically changed modern modes of communication and connected those who may otherwise never have an opportunity to interact.
Facebook is an on-line social networking site developed in 2004. Originally intended for university students, Facebook has now exploded in popularity around the world -with nearly 1.59 billion daily users (2). It allows users to get in touch with people with similar interests, backgrounds, and experiences instantaneously. The capabilities of Facebook now extend far past keeping in touch with friends and family. People use it as a main source of news, forming connections around similar hobbies, interests and illnesses, and for self-promotion (3,4).
Groups are a feature of Facebook that allows like-minded individuals together to connect and share experiences. Facebook groups have varying levels of privacy: an open group is viewable to anyone on Facebook and not restricted to members of the group. A closed group has content only viewable by the group members, who are able to join through an approval process. Anyone can see the existence of a closed Facebook group. However, a secret group is not viewable by anyone except those in the group -who are able to join only by direct invitation (5). It is private and hidden in searches; those who are not members of a secret Facebook group cannot see the existence or the content of the secret Facebook group (6). This type of privacy holds particular benefit for research, as confidentiality and anonymity are ethically important. The British Psychological Society published ethical guidelines for internet-mediated research in 2017 encompassing 4 main principles: 1) respect autonomy, privacy and dignity; 2) maintain scientific integrity; 3) social responsibility and 4) increase the benefits whilst reducing the harms (7).
Previous studies have conducted focus-group style research in private or secret Facebook groups (8), with the purpose of conducting a needs analysis or collecting data in a more focus-group style to understand a problem. Research groups at the University of Indiana and the University of Edinburgh have particular experience with Asynchronous Remote Communities, or focus groups occurring online in which participants are not online at the same time, and have published widely on the topicprimarily short reports with recommendations for future research and lessons learned (9,10).
However, none of the studies we have found to date use asynchronous remote communities, particularly in a secret Facebook group, to co-design an intervention for future testing. This paper describes a method of co-production and intervention development specifically designed to overcome barriers of in-person research methods: online, secret Facebook groups. In this paper, we will describe the development and use of the secret Facebook group in a dispersed group of participants with a relatively rare health condition and discuss the benefits and limitations for research. This secret Facebook group study was part of a larger intervention development research study, using the Six Steps in Quality Intervention Development (6SQuID) framework. The 6SQuID framework is based on six steps: 1. Identifying a public health problem and its causes; 2. Clarify the modifiable and non-modifiable causal factors; 3. Identifying the theory of change (what causal pathways to interrupt and how); 4. Identify the theory of action (how to deliver the intervention); 5.
Test and refine on a small scale; and 6. Collect sufficient evidence of effectiveness to justify more rigorous evaluation and implementation.
The population group for our study was women with a history of gestational diabetes, a disease that affects 4.4% of pregnancies worldwide and significantly increases risk of Type 2 diabetes later in life (11,12). Physical activity was the target behaviour of this intervention development study as it is a beneficial tool for Type 2 diabetes prevention for women with previous gestational diabetes as well. A US-based prospective cohort study of 4,554 women with a history of gestational diabetes found that for every 100 minutes increase of moderate-vigorous physical activity performed per week, there was a 9% reduced relative risk of T2DM onset [95% CI, 0.88-0.94], even after adjusting for diet and BMI (13).
A lay advisory group composed of three women with previous gestational diabetes was consulted about the ideal method of co-production; they favoured a secret Facebook group for several reasons including convenience, high existing use, and ease of use of the website and phone application.
Previous research has shown that 81% of mothers use Facebook, and 56% of these mothers check the platform several times a day (14). Additionally, online focus groups and in-person focus groups have been shown to be comparable in terms of quality of content collected (15).

Aims And Objectives
The aim of the study was to test feasibility and acceptability of using a secret Facebook group for coproduction. This paper describes the approaches used and their benefits and limitations.

Participants and recruitment
The majority of participants (16/21) had previously had face-to-face or telephone interviews with the researchers in June to September 2018. New participants (5/21) were recruited from a Facebook group that is specifically for women with current or previous gestational diabetes.
All participants were living in the UK during the study but were geographically dispersed.

Content development
The Facebook group content was developed to address steps 2 through 4 of the 6SQuID framework: to clarify modifiable and causal factors of the problem; develop a theory of change and begin to develop a theory of action for a physical activity intervention (1). A theory of change is the process by which change comes about for individuals, groups and communities -it asks: how are we going to change this behaviour? To create a theory of change, we first performed a situation analysis (commencing in the previous in-person interview portion) and developed a fishbone diagram with participants to explain key barriers and facilitators to physical activity in their daily lives. Then we aimed to understand modifiable and important factors to understand which factors can be changed with the greatest scope for improvement in their physical activity. Finally, the researchers analysed this information to begin to develop a theory of how to change their behaviour: this was presented back to participants for views of acceptability and feasibility.

Running the workshop
The workshop took place over a 15-day period in May to June 2019, the length of which was based on previous research (16). To action the 6SQuID steps, the lead researcher created posts for the were informed of this in the consent process. A secret Facebook group is clearly a private spaceparticipants should feel able to write freely without concern that what they are saying is viewable to those who are not in the group. In contrast to confidentiality involving managing private information, anonymity involves obscuring identifiable information for participants and can be used to maintain confidentiality (17). The part of qualitative analysis that involves the highest risk to anonymity is in disseminating the results. For research using open and online internet forums, the risks to anonymity are higher, as participant quotes can be traced back to the original source and can potential identify the participant. However, this is not a risk in a secret Facebook group: A secret group is not searchable on Facebook or any search engine, and details given in the group are only viewable by group members. As such, anonymity is preserved barring any data breaches through participants sharing or Facebook security lapses.
Identity verification holds two primary considerations in social media research: 1. Are participants who they say they are? 2. Does the person that a participant portrays in a Facebook group accurately represent their real self? As "Facebook profile page amounts to a blank canvas on which each user has free reign to construct a public or semi-public image of him-or herself", it is possible that the participants the researcher has not met before may not fit into the eligibility criteria (p. 213) (18).
However, this research is not concerned with who the participant actually is, but rather with each participant's opinions and perceptions on intervention ideas. It is possible that people may say they approve or disprove of something that they do not actually approve or disprove of (in real life) -thus harming the data collected -but this is a risk in any research with people (5).
Informed consent was taken online in a Qualtrics survey, as studies have shown that an online consent form provides equal comprehension compared to a written consent form (19). Participants were emailed a link to a survey that laid out the same information as a written participant information sheet and consent form, but clicked 'agree' after each consent statement to indicate consent, and then provided identifying details to verify their identity including year of birth, year diagnosed with gestational diabetes, name, and Facebook profile name and link. The online study was approved by the University of Edinburgh Health in Social Science ethics committee.

Evaluation of the online workshop
An evaluation plan was developed to explore the feasibility and acceptability of this method to generate a theory of change and theory of action for the intervention.
The key components of a feasibility study include acceptability, practicality, demand, implementation, adaptation, integration, expansion and limited efficacy (20). Acceptability and practicality are most relevant for this study, and how they were assessed are described in detail in Table 1. Demand can be assessed using the Facebook analytics because it shows actual use. Implementation will be based on analytics and the researcher's own experiences of executing the plan. Practicality, integration and limited efficacy were assessed based on the overall results of the workshop, whereas adaptation and expansion should be explored in future research.

Aspects and design
Aim Responses Table 2 provides a summary of the posts and responses. In total, 521 comments were provided in response to 18 posts of polls, video, text or photos for an average of 28.9 comments per post. The total word count of all comments (excluding the researcher's responses and comments) was 21,142 words. Table 2 Summary of Facebook posts and interaction with participants.

Survey Question
Main results (%) Specific comments Overall, how much did you like participating in the "If it helps someone else then it's always worth doing" (Shannon) "It was a nice way to do the studywas able to participate at the time that suited me and my schedule each day." (Poppy) "Was great to be able to dip in and contribute when had time (normally once the kids were in bed) was good to hea[r] other viewpoints and experiences" (Rebecca) "Made me reflect and was nice to see others in same position reflect. Nice to think it was helpful for research too." (Elizabeth) "I'm always happy to take part in studies to help research into little known conditions such as gd" (Lorena) "Yes, it's good to know people want genuine experiences to help others in future and it's nice to meet likeminded mums" (Irene) "It's hard to get time when you have two kids under 3 as they want all your attention. Then there's the dog and husband as well it's hard even to get time to do a survey." (Yvonne) "The study wasn't as detailed as I expected it to be" (Emily) "It will depends on what else I have on during the week. I was away for part of the first week and I had to play catch up for some of the posts" (Allie) The The results from the end-of-workshop survey are detailed in Table 3. In short, overall the workshop was viewed positively. It was enjoyed by participants, with 20 of 21 participants (95%) saying they liked the workshop 'somewhat' or 'a great deal'. The same percentage said they felt the group was a safe and open environment to share opinions, with the dissenting view mentioning her dislike of Facebook more generally. Perhaps the most important indicator of acceptability was asking participants if they would take part in something similar again, in which case over 70% (N = 15) said 'Yes'. Participants who said they would take part again mentioned how the format was very convenient to fit into their day and when they had time; it allowed them to reflect on their own experiences; and they liked helping research progress for gestational diabetes. The six participants who said they may take part again said it was still difficult finding time to do it, it depended on what else was ongoing in their lives, and one participant suggested it was not as detailed as she anticipated it to be. Overall, the workshop was acceptable to participants.

Discussion
The results from this research suggest that co-producing an intervention over a secret There also were no travel or transcribing costs for the Facebook group, as compared to a typical focus group or interviews.
Another key strength of the research was that location was not an obstacle to participation.
Participants in this study were spread throughout Scotland -geographic spread and rurality of participants is typically a significant barrier to co-production. Holding an intervention development workshop online vastly increases the input from groups who may not be able to attend face-to-face, due to time, location or circumstantial constraints. This opens up the possibility of doing research with hard to reach groups.
A further strength of this method was that, although potentially novel in its use, it remained evidencebased in the asynchronous remote community (ARC) literature. Previous research has suggested optimal methods of recruitment, content, sample size and analysis method, which this study followed closely (5,9,10). Additionally, given that this method allowed for participants to take as much time as needed to consider and respond to a prompt -in contrast to the immediacy of in-person interviews and focus groups -another strength could be increased thoughtfulness of answers. The researchers could also take time to consider responses and ask follow-up questions in a more considered manner than what may occur during in-person qualitative research.
There were a few limitations of this study mainly related to the use of Facebook

Conclusion
This methodology paper suggests that using a secret Facebook group for running an interventiondevelopment workshop is acceptable to participants, feasible for the researcher to conduct and generates high quality, nuanced data. This method holds significant promise in similar future work with geographically remote communities, those who have difficulty travelling or limited time, or those with relatively uncommon diseases or risk factors.

Abbreviations
Six Steps in Quality Intervention Development: 6SQuID Declarations Ethics approval and consent to participate Ethical approval was obtained prior to the commencement of the study by the School of Health in Social Science's ethical committee at the University of Edinburgh. Participants provided informed online consent prior to the workshop beginning.

Consent for publication
Participant's consented for anonymised quotes and aggregate data regarding age, time since last gestational diabetes diagnosis to be described in reports and publications.

Availability of data and materials
Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Competing interests
The authors declare that they have no competing interests.

Funding
This methodology research was undertaken as part of a Diabetes UK PhD studentship. The funding body had no role in the design of the study nor data collection, analysis, interpretation or in writing the manuscript.

Authors' contribution
AB designed the study and completed data collection, analysis and interpretation under the supervision and advice of RJ and AK. AB drafted the initial manuscript with input from RJ and AK. computing systems; 2013: ACM. Figure 1 Example image of family-related facilitators to physical activity (identified during previous interview stage). Participants were asked to comment on the image and list 1-3 factors within this theme that were most important to them Figure 2 The ranking of the top 10 factors women said made physical activity easier. Women then comments on this post with their approval and any additional comments they wanted to share.