Using Facebook in Human Factors
Research?
A first exploration for
opportunities to gain future user
input.
Agnieszka van de Leur
12013234
A thesis presented for the degree of
Master of Science
Master Medical Informatics
University of Amsterdam
The Netherlands
19-02-2021
Using Facebook in Human Factors
Research?
A first exploration for opportunities to
gain future user input.
Agnieszka van de Leur
12013234
Student
Agnieszka van de Leur
Student number: 12013234
Email: a.chycko@amsterdamumc.nl
Mentor
Dr. L.W. (Linda) Dusseljee - Peute
Email: l.w.peute@amsterdamumc.nl
Amsterdam UMC, location AMC, Department of Medical Informatics
Tutor
Liesbeth van den Berg
Email: liesbeth.vandenberg@amsterdamumc.nl
Amsterdam UMC, location AMC, Department of Medical Informatics
Location of Scientific Research Project
Amsterdam UMC, location AMC, Department of Medical Informatics
Practice teaching period
Preface
This master thesis is the final assignment for the Master Medical Informatics at the University of Amsterdam and I am not going to lie, it was a long journey! But what a satisfying one!
It all started in September 2018, when I already knew I was pregnant with my first child. I went to Linda Dusseljee-Peute and asked if it would be possible to write a thesis under her supervision. She agreed and the first ideas came into our minds. However, the real start was postponed to September 2019, after all elective courses, last months of pregnancy and then maternity leave, which was longer than regular, but really needed and full of joy, as I could watch Lucas experiencing his small world with such a delight. Part-time work started and chapter two was ready by the time of the outbreak of the COVID-19 pandemic, which affected us all in different ways. For me, it meant another break. Work on chapter three started again in August 2020, with some extra motivation to finish my study as fast as I can, as I knew that I was pregnant with my second child. And now here we are, February 2021, my work on this thesis is finally complete!
Dear Linda - I would not be able to do it without you! I learned so much from you about being a researcher and a working mom. Your constant support and enthusiasm during the last months allowed me to grow. Your way of seeing things and connections within research still astonishes me! I am very grateful that you accepted me under your supervision.
Dear Liesbeth - you joined me on my journey at the very last moment but left a great mark. Your remarks put this thesis on another level, thank you!
Last but not least, dear Michiel - you are the reason I can be where I am now. Without your hard work and sacrifices, it would not be possible. You supported me from day one of the pre-master studies and celebrated every small success. You were very brave when I complained but always put me back on track. Thank you my lion!
I would also like to thank my family in Krak´ow, my parents and brother, who could always cheer me up on the phone and bring comfort in every situation. Also my parents-in-law, for support and taking care of Lucas so I could work. And my friends from studies, Mirna, Sanne and Casper - we met during pre-master, I am so happy to still be in touch and encourage each other.
Abstract
Human factors engineering is considered essential in the design of health infor-mation systems, because it helps them work more efficient, effective and safe. In case of eHealth applications the end-users are the patients or their health care providers. End-users’ insight is fundamental if the future application will become a success-ful one. The biggest challenge lies in finding the right potential end-users, which could provide valuable insight for the development process. This thesis explored the opportunities to gain future user’s input with the help of Facebook. First the pos-sibilities were determined based on a literature search in the form of scoping review and then put into practise in a comparison study between two most suitable meth-ods, that resulted from the scoping review. Additionally, the survey was created to investigate potential participants’ willingness to share their personal information through Facebook. It was demonstrated that under certain conditions it is effective to recruit participants for User Centered Design studies through Facebook, it espe-cially proves to be a quick process, which requires a smaller budget than traditional recruitment methods.
Keywords: Facebook, Human Factors Research, User Centered Design, Re-cruitment
Samenvatting
Human factors engineering wordt als essentieel beschouwd bij het ontwerp van gezondheidsinformatiesystemen, omdat het helpt deze effici¨enter, effectiever en veiliger te laten werken. Bij eHealth-toepassingen zijn de eindgebruikers de pati¨enten of hun zorgverleners. Om te garanderen dat een toekomstige applicatie succesvol is, zijn de inzichten van eindgebruikers van fundamenteel belang. De grootste uitdaging ligt in het vinden van de juiste potenti¨ele eindgebruikers, die waardevol inzicht kunnen leveren voor het ontwikkelproces. In deze studie zijn de mogelijkheden onderzocht om input van toekomstige gebruikers te verkrijgen met behulp van Facebook. Eerst werden de mogelijkheden bepaald op basis van een literatuuronderzoek in de vorm van een scoping review en vervolgens in praktijk gebracht in een vergelijkingsonder-zoek tussen de twee meest geschikte methoden die uit de scoping review voortkwa-men. In het vergelijkingsonderzoek is er ook gebruik gemaakt van een enquete om te onderzoeken in hoeverre potenti¨ele deelnemers bereid zijn om hun persoonlijke informatie via Facebook te delen. De resultaten van de onderzoeken tonen aan dat onder bepaalde voorwaarden effectief is om via Facebook deelnemers te werven voor User Centered Design onderzoeken; het blijkt vooral een snel proces te zijn, dat een kleiner budget vergt dan traditionele wervingsmethoden.
Trefwoorden: Facebook, Human Factors Research, User Centered Design, Re-cruitment
Contents
1 General introduction 3
1.1 Theoretical frameworks . . . 4
1.1.1 User centered design in human factor research. . . 4
1.1.2 Patients’ Willingness to Share Information . . . 5
1.2 Study objectives . . . 6
1.3 Thesis overview . . . 7
2 Possibilities of using Facebook as a tool in research: a scoping review. 8 2.1 Introduction . . . 8
2.1.1 This study . . . 8
2.2 Methods . . . 9
2.3 Results . . . 9
2.3.1 Study design, setting . . . 9
2.3.2 Category I: usage of Facebook’s post and groups . . . 11
2.3.3 Category II: usage of Facebook’s advertisements . . . 15
2.4 Discussion . . . 24
3 Recruitment of participants on Facebook: a comparison study. 26 3.1 Introduction . . . 26 3.2 Methods . . . 27 3.2.1 Recruitment . . . 27 3.2.2 Survey . . . 28 3.2.3 Theoretical framework . . . 29 3.3 Results . . . 29
3.3.1 General statistics and remarks . . . 29
3.3.2 eHealth usage . . . 30
3.3.3 Willingness to share information . . . 31
3.4 Discussion . . . 33
4 Overall discussion 35 4.1 Main findings . . . 35
4.2 Strengths and limitations . . . 36
4.3 Future research . . . 37 4.4 Overall conclusion . . . 37 A Appendix 44 A.1 Appendix A . . . 44 A.2 Appendix B . . . 44 2
Chapter 1
General introduction
Human factors engineering (HFE) is considered essential in the design of health information systems, because it helps make work more efficient, effective and safe. Applying human factors could allow eHealth tools to be designed with the user in mind, taking into account human capabilities and limitations such that people work in harmony with technology. Usability, defined as the effectiveness, efficiency and satisfaction with which users can achieve their goals [1], and acceptance of health IT by healthcare professionals and patients nowadays receive special attention from HFE researchers. Lack of attention to these issues may present obstacles to a successful implementation of new systems and applications [2]. An example where not respecting HFE lead to failure is the Computerized Physician Order Entry in a hospital in the Netherlands. Flaws in the design of the program could lead to patient missing a medication dose or physician being unable to select the number of days for medication duration. This could lead to increased workload for the physicians and endanger the health of patients [3].
Stanton et al. [4] stated in their book that there are many human factor (HF) methods that can be used in multiple design stages, even if systems are inflexible to changes. However, HF can provide the most valuable input in the very early stages of the design, when it is still quite little effort to perform modifications to the system’s prototypes. Additionally, many of these methods can be used for both predictive and evaluative purposes, which plays a key role in the early design stages. HF methods are especially being recognized as important in the development of eHealth applications, because even if the automation and usage of mobile devices increases, human skills will remain elemental for many tasks. Therefore, the coop-eration between humans, machines and systems are critical for success [5]. The use of mobile devices by healthcare professionals has changed many aspects of clinical practice, it became a common sight in many practices [6]. This led to increased growth in the development of medical software applications. By applying HF meth-ods in designing these mobile devices, software and wireless technology, it might be possible for physicians to make diagnoses quicker and with fewer errors, and for patients to enjoy better care.
In the case of eHealth applications, the end-users are the patients or their health care providers. Their insights in functionality and characteristics for a new system are fundamental if the future application wants to be considered as a successful one. The biggest challenge lies in finding correct potential end-users, which could provide valuable insights for the development process. Cheaper and more effective methods
are desired because with traditional recruitment finding the right type of end-user is cumbersome and expensive. The opportunity of social media for quantitative and qualitative researchers now arises as a new source of information. Facebook might be one of the social networking sites especially useful in connecting researchers with small specialized communities [7].
1.1
Theoretical frameworks
1.1.1
User centered design in human factor research.
In user centered design (UCD), a deep understanding of users, tasks and environ-ment should form the foundation of a project. It is crucial to apprehend and then put in motion the whole user experience. Therefore, including end-users as well as professionals from across multiple disciplines determines a successful design process. The iterative process starts with research on context of use of future application, continues to specifying user requirements, then to producing design solution and fol-lows to the evaluation against requirements (see Figure 1.1). The process continues until the evaluation results are satisfactory for end-users and designers. End-users should be especially involved in the evaluation of the requirements process, for ex-ample by usability testing, engaging them in every stage of the process can bring a valuable feedback on how well the design is understood, and can consequently lead to a better end product [8]. Advantages of involving users in the design and devel-opment of a new system are an improvement the system’s quality and a reducement of the development time as potential usability problems can be resolved before the launch of the final product [9]. Applying UCD to the development of health IT applications for patients improves functionality and usability, which have a direct influence on the acceptance of the final system [10].
In this thesis, the main focus will lie on the first phase of the UCD, which is un-derstanding and specifying the user context. For this phase, finding a representative group of end-users is crucial for the success of any project. It is reasonable to assume that if we will find potential participants to get their insight at the very early stage of the process, and they will be willing to share their thoughts and experiences, then it will be more effortless to include them in further stages of the design.
Figure 1.1: The user-centered design process [11].
1.1.2
Patients’ Willingness to Share Information
Zhu et al. [12] have established an influence model of Online Information Support (OIS) and privacy concerns on patients’ Willingness to Share Information (WSI). In their model, the intention to share the information by patients is the dependent variable, privacy concerns and OIS are independent variables, and disease severity and information sensitivity are moderator variables. Additionally, gender, age and education level are control variables. Feelings and behaviors can differ according to gender, age influences physical condition and life experience, and education level is strongly related with strength of judgment and cognitive competence (see Figure 1.2).
Researchers [12] proved that privacy concerns have a negative impact on users’ WSI, and the OIS has a positive impact on the users’ WSI. Furthermore, information sensitivity negatively moderates the impact of OIS on WSI, whereas disease severity positively moderates the impact of privacy concerns.
In this thesis willingness to share information of the potential participants will be analyzed in order determine how challenging it can be to find participants on an online social networking platform who want to participate in a research and share their opinions during various stages of UCD process.
Facebook will be used for this purpose, because their users may have concerns about the controversial privacy policy, including inadequate safeguards against com-panies engaging in data harvesting [13], it is important to determine the impact of these concerns on the participants’ WSI in the recruitment process.
Figure 1.2: Model of information-sharing intentions of patients in online health communities [12].
1.2
Study objectives
This thesis aims to answer the following research question:
• What is the best method to find potential participants who are willing to share their opinions for UCD studies on Facebook?
Additional sub-questions of interest for this master thesis were asked:
• What are the best practices currently used in medical research studies on Facebook and what are the challenges connected with these practices? Is there any evidence on recruitment and its effectiveness for UCD studies for medical research conducted with help of Facebook?
• How effective are two recruitment methods via Facebook (advertisement and private groups), to recruit potential participants for UCD studies?
• What is the difference in the willingness to share personal information and opinions about eHealth between the potential participants for UCD studies found through two recruitment groups on Facebook?
1.3
Thesis overview
This study is divided into two main parts. These two parts will be elucidated in chapter two and three.
In the first part, in chapter two, a scoping review on the current use of Face-book in research was performed to assess evidence in the effectiveness of different communication and recruitment methods with potential end-users for user centered design research.
Chapter three will explore the possibilities to recruit participants that were found in chapter two, with the additional focus on reaching out to potential end-users of a future eHealth studies. A comparison study between an advertisement (paid method) and posting on Facebook groups (unpaid method) was conducted in order to test which of these is more promising for recruitment of the most suitable sample for the future use in user centered design studies. Additionally, in this part, the willingness to share information of potential participants was tested in the form of a survey.
In chapter four an overall discussion and advice for using Facebook in UCD research will be presented.
Chapter 2
Possibilities of using Facebook as a
tool in research: a scoping review.
2.1
Introduction
Human factors methods can be used at early stages of the design process of the various systems and applications [4], also in medical informatics. User centered design plays a special role in this process as direct insight from the future end-users can bring large perspective, which can only benefit the development. Applying UCD improves the functionality and usability of the applications, which have a direct influence on the acceptance of the final system by its end-users [10].
Searching for participants for various types of research can be a long-lasting and expensive process. In the time of the pandemic, with social distancing influencing our everyday life, new methods of connecting with future research participants can meet even more obstacles than before. Therefore, it is necessary to look for new research methods, which could save time and be more cost-effective, and additionally do not involve direct face-to-face contact between the research team and participants or patients.
Over the last ten years, use of social media, including Facebook, Twitter, Insta-gram and YouTube, has grown tremendously. As of September 2019, the number of monthly active users of Facebook has passed over 2.45 billion worldwide [14]. Con-sidering the popularity and reach of Facebook, it could become an interesting tool to connect with individuals, in order to gain insight into their opinions and experiences and involve them in the research and development processes. Facebook has several features which can be used by researchers: paid and unpaid advertisements, private closed groups containing only selected individuals, creating private sub-websites in order to promote own business or research or publishing simple post on open groups, which everyone could join. It is therefore important to explore the possibilities to use Facebook in UCD studies and to develop a guideline for future UCD researchers how they can use Facebook in the most productive way.
2.1.1
This study
The present study aims to summarize the evidence available in the recently published literature regarding the possibilities of using Facebook as a tool for health research. Ultimately, the study intend to answer the following research questions: 1) How
and when can researchers use Facebook as a research tool?, 2) What are the main barriers and facilitators for using Facebook in research?, 3) Is there any evidence on recruitment and its effectiveness for UCD studies for medical research conducted with help of Facebook?.
A scoping review was conducted to examine the current state of knowledge and identify the key concepts regarding the use of Facebook as a research tool.
2.2
Methods
A literature search of two databases, namely PubMed and EMBASE, was performed in November 2019. The search was restricted to full text articles published in 2015-2019, because during this period the popularity of Facebook in medical research peaked. A combination of keywords was applied as a search strategy within title and abstract, the search query can be found in the Appendix A.
All identified papers were exported to Rayyan (http://rayyan.qcri.org) [15], ci-tation management application, 15 duplicates were deleted. The inclusion criteria were as follows: English language, Facebook had to be used as a research tool in an eHealth or health related study, participants/patients had to be present and active during research, research objective and outcome measures had to be clear. The following exclusion criteria were applied: non-English language, no patients or par-ticipants appeared in the paper, no clear research objectives and outcome measures, conference papers, posters, other systematic reviews, usage of other social media (not Facebook).
One reviewer assessed titles and abstracts to ensure both the inclusion and ex-clusion criteria were met. All included articles were then reviewed based on full text version and and subjected to further analysis. An extraction form was created in order to analyze the articles in systematic order: author, year of publication, the country where the study was conducted, research objective, patients/population, clinical domain, used methods, results, discussion points or suggestions from the author of a given article.
Initially, open coding of the data was performed in order to create general cate-gories of where and when Facebook can be used in research. The open coding was based on which feature of Facebook was used by the authors. Afterward, potential grouping based on purpose of the studies was performed. Finally, in one of the themes it was possible to carry out further division based on research objectives and variables (cost of advertisement per participant, per survey and per click).
2.3
Results
2.3.1
Study design, setting
Literature search The search performed in November 2019 on the online databases (PubMed and EMBASE) identified 273 relevant publications (see Figure 2.1). After removal of duplicates (n=15), initial screening was performed on the titles and abstracts of 263 remaining articles. After the first analysis, 210 articles were removed, as they did not comply with the inclusion criteria. From the remaining 53 articles, it was not possible to access the full text for seven of them and therefore
they were excluded from the final set. Further, 12 articles did not meet inclusion criteria after the screening of the full-text and therefore were also excluded. Finally, 34 articles were included and analysed.
Figure 2.1: Flowchart describing the systematic literature review process. Structure of results The included studies are discussed in two different cat-egories: I) using Facebook’s posts and groups and II) using Facebook’s advertise-ments. Categories were created based on the feature of Facebook which was used by the researchers in their studies. Then the open coding of the data was performed by the researcher and supervisor in order to create general themes, based on the purpose of the studies. The first category includes three themes and the data from eight articles will be presented. The second category contains two themes where the use of Facebook’s advertisements will be discussed, based on data extracted from 26 articles. Theme four in the second category is further divided into four classes, based on the research objective and the variables explicitly stated by the authors of the included studies. A structure of the division performed on the included articles can be seen in Figure 2.2. Additionally, in every category common barriers and facilitators will be presented.
Figure 2.2: Graph describing the division of the included articles by categories, themes and classes.
2.3.2
Category I: usage of Facebook’s post and groups
Eight articles were found in this category and were further divided into three themes. Theme 1: sharing knowledge
Four articles belonged to this theme [16], [17], [18], [19]. A summary of all articles can be found in Table 2.1.
This theme contains the biggest diversity in country of origin compared to all other themes: two from Latin America, one from Asia and one from Europe. The majority of studies included young participants, with an age difference between 15 and 34 years old. Interestingly, two of them focus on epidemiology (HPV vaccination and arboviral diseases) and they both conclude that Facebook is an effective medium for sharing this knowledge. Two articles used Facebook’s private groups and two created posts on a dedicated Facebook page. Groups proved to be successful, but creating posts with purpose of sharing knowledge only worked well in a single case. In general, all articles conclude that it is possible to use Facebook in sharing knowledge interventions, like implementing a Facebook-assisted teaching method or running campaigns to create awareness about diseases. Only one study suggests that old fashioned leaflets might still be more powerful.
Theme 2: gathering opinions
Two articles were found to represent this theme [20], [21]. A summary of all articles can be found in Table 2.2.
Both studies used the private group feature of Facebook to collect opinions from participants. Online focus groups were conducted on Facebook by having modera-tors ask questions and collect answers together with likes and shares. Both studies
Author Year Country Objective Clinical Domain Methods Results Discussion/Conclusions Lai [16] 2015 Taiwan To investigate the
effectiveness of a Facebook (FB) assisted teaching method on fe-male adolescents’ knowledge and attitudes. nursing/ health sciences
Time series study, 200 students divided in two groups, one with FB-assisted teaching on a pri-vate FB group. Measure-ments done in T0, T1 two weeks after and T2 in 8 weeks after.
In the experimental group scores for knowledge, at-titude and vaccination in-tention were higher than in the control group for both T1 and T2 .
FB is a good place to dis-cuss sensitive issues and might create space private enough to make partici-pants feel safe. The FB-assisted teaching method is superior to the tradi-tional teaching method in facilitating female adoles-cents’ knowledge and atti-tudes. Friedman [17] 2016 Puerto Rico To examine an ini-tiative created to raise awareness of Alzheimer Disease (AD). neurology/ health promo-tion AD cafeterias meetings with discussion panels and FB page with code book with crucial information about AD. Evaluation of the engagement of the population and an analy-sis of content of FB posts.
Participants demon-strated improved knowl-edge of AD following the education sessions
Combining personal ap-proach and FB page can improve community knowledge about AD.
Gamboa [18] 2019 Dominican Republic To evaluate FB as a peer-to-peer health commu-nication tool to promote education and encourage pre-ventable behaviors.
epidemiology Three private FB groups, two with intervention (weekly posts from mod-erators, free thoughts exchange from partici-pants), one without. Self-administrated pre- and post-surveys on knowl-edge and preventable behaviours.
Survey knowledge scores, subjective self-rated knowledge scores and knowledge on preventa-tive behaviours increased significantly in the inter-vention groups.
FB group can be effective in improving knowledge and prevention practises, but participants must be active. Agha- Mir-Salim [19] 2019 UK To evaluate the effectiveness of Facebook leaflets in raising awareness of melanoma and harmful sun-related behaviour. cancer preven-tion Comparison of leaflets and FB based on a pre-and post-intervention questionnaire.
Following the SunSafe intervention, the mean knowledge score out of 25 improved in both groups. A two-sample indepen-dent t-test demonstrated that the improvement in knowledge score of the leaflet arm was sig-nificantly greater than the improvement in the Facebook arm of the trial.
Despite growing popular-ity of social media, public health campaigns should still consider old fashioned (printed) methods. A printed version is more personal, so participants put more attention to it and learn better.
Table 2.1: Summary of articles evaluating possibilities of sharing knowledge on Facebook (FB).
have similarities, with (1) population 52 participants in Thrul et al. (2017) and 63 participants in Holton et al. (2016), (2) age of participants, where the mean age was 20 years old and 30 years old respectively and (3) origin from highly developed countries (the USA and Australia).
Both articles conclude that using Facebook to gather participants’ opinions is possible and effective.
Theme 3: evaluating engagement and satisfaction
Two articles were found to represent this theme [22], [23]. A summary of all articles can be found in Table 2.3.
In both studies, a similar technique to gather data was used. Participants were invited to private Facebook groups and then surveys together with Facebook mea-surements were collected and interpreted. Both studies came from USA and were published in 2018. The main differences were: (1) the time of the study: 4 weeks [23] and 9 months [22]; (2) age of participants: 18-25 year old and 50% of participant above 45 years old respectively and (3) size of the sample: 17 and 429 respectively. Despite differences in methodology, both articles conclude that using Facebook for evaluating patient’s engagement and satisfaction is possible and feasible. Ad-ditionally, Dhar et al. (2018) mention that the employment of social media into clinical practice, in form of a virtual support forum, might be beneficial for the patients.
Barriers and facilitators
Barriers Bringing new technology into research might be risky and in the case of Facebook, the authors mentioned encountering barriers in their research. Firstly, one of them is a self-selection of the sample [20], researchers have limited influence on who will participate in the study. Second is possible response bias, connected with self-reports [18], where researchers have to believe in what participants say, with no possibilities to confirm their statements through other sources such as medical records. Additionally, some studies had problems with small sample sizes ([18], [23]), which can influence the generalizability of the results. Country of origin might also have an influence on that, as societies differ from each other, and results from Latin America might not be duplicated in Europe. Furthermore, one of the authors suggests a lower attachment to the study of participants via Facebook than via more old fashioned methods [19], and therefore advise to combine these methods if possible.
Facilitators Using Facebook as a research tool brings some advantages to research methodology. It allows to gather a large, geographically and demographically diverse sample [21], also on a national level in big countries as the USA. Additionally, it allows participants to join the research at a time which is convenient for them [20], which has an impact on the level of engagement and possibly on sample size. What can be specifically important in research on sensitive topics, such as connected with gender, is that Facebook gives users a sense of privacy and hence their answers will be more trustworthy and participation will be more active [16].
Author Year Country Objective Clinical Domain Methods Results Discussion/Conclusion Holton
[20]
2016 AUS To identify public opinion about sex-ual and reproduc-tive health. public health/obstetrics Advertisement of FB to attract participants to a private FB discussion group. Questions formed in 4 themes, answers tran-scribed and systematically coded.
It was possible to include a diverse sample of women and men via FB. There was enough activity on the discussion group to gain an insight in participants opinions.
FB was a cost-effective and efficient way of en-abling people from diverse geographical locations to participate in the group discussion at times that were convenient for them. Thrul
[21]
2017 USA To inform health behavior change in-terventions.
psychiatry Advertisements on FB to attract to a secret FB group, where 90 min on-line focus groups (OFG) were conducted. Engage-ment, including number of posts, comments, com-ment responses and word counts were analyzed.
Participant experiences with the Facebook OFG procedures were positive overall
It is feasible to conduct synchronous FB OFGs and this data collection method may be partic-ularly appropriate to inform development of social media interven-tions for health behavior change.
Table 2.2: Summary of articles evaluating gathering of users’ opinions on Facebook (FB).
Author Year Country Objective Clinical Domain Methods Results Discussion/Conclusion Dhar [22] 2018 USA To investigate how
implementation of social media may affect patient en-gagement and satis-faction.
transplantology A FB support group for liver transplant pa-tients with health care providers. A survey after 9 months of group existence to gain insight in patient’s satisfaction. Analysis of activity levels and post’s content of group users.
98% of posts were reacted to or commented on by other group members. Pa-tients were the most ac-tive users. A total of 95% of survey respondents reported that joining the group had a positive im-pact on their care.
An incorporation of Face-book into clinical practice is not only feasible but can provide a substantial ben-efit to patients.
Stapleton [23]
2018 USA To evaluate engage-ment with and ac-ceptability of using a secret FB group to deliver a healthy body image inter-vention.
dermatology/ cancer preven-tion
A secret educational FB group with daily post for 4 weeks. Pre- and post-intervention survey (ac-ceptability). Engagement data extracted from FB (engagement).
An average of 91% of participants viewed each post. The highest amount of likes for inspirational and psycho-educational posts. A post-survey showed favorable ratings for intervention: interest-ing, understandable and useful.
The labelling of posts as discussion posts may be helpful for soliciting en-gagement. Potential use of FB in group-based in-terventions in various geo-graphic locations.
Table 2.3: Summary of articles evaluating user’s engagement on Facebook (FB).
2.3.3
Category II: usage of Facebook’s advertisements
26 articles were found in this category and were further divided into two themes. Theme 4: conducting recruitment
Recruitment of participants is the most popular reason to use Facebook’s adver-tisements in research. 24 out of 34 included articles were evaluating this feature of Facebook. A summary of all articles included in this theme can be found in Table 2.4.
It was possible to distinguish 4 main groups of research objectives used by re-searchers: effectiveness, feasibility, utility and viability of recruiting participants through Facebook. Additionally, for each group variables were extracted in order to systematize the knowledge. It has to be mentioned that in some cases, under the term of feasibility, researchers were describing cost-effectiveness of the method, which could be added to the group of effectiveness, but the decision was made to keep the classification based on the author’s description of research objective in the study.
Research objective Variable Number of articles
Effectiveness
Cost per participant 10 ([24], [25], [26], [27], [28], [29], [30], [31], [32], [33]) Cost per survey 2 ([34], [35])
Cost per click 4 ([36], [37], [38], [39])
Feasibility Cost per participant 6 ([40], [41], [42], [43], [44], [45]) Utility Cost per participant 1 ([46])
Viability Cost per participant 1 ([47])
Table 2.4: Summary of research objectives and variables used in articles evaluating recruitment through Facebook.
Class A: evaluating effectiveness with costs per participant The most pop-ular way of appraisal of Facebook recruitment possibilities was through judging its effectiveness with costs per recruited participant, 10 out of all 24 articles using Face-book’s advertisements were using this evaluation. A summary of articles using this method can be found in Table 2.5.
Articles are published in the years 2016-2019. Four studies were performed in the USA, three in Australia, one in Canada, New Zealand and Denmark. Face-book recruitment was used most in psychology (depression and smoking cessation), epidemiology and oncology.
Researchers tried to recruit participants from a variety of populations: young [31], middle age [32], hard-to-reach ([33], [32]), woman only [24], men only [31].
In general online recruitment methods (Facebook) were more cost-effective than traditional methods ([24], [25], [27]), however most researchers did not take the cost of staff salary into consideration while calculating costs for the traditional method. In one case Facebook recruitment was faster but not more cost-effective than tradi-tional methods [29].
Randomized controlled trials often require highly specific inclusion and exclusion criteria. The targeting used then needs to be more precise and this raises the costs ([28], [30]).
Based on the conclusions from articles included in this section, it can be said that recruitment of participants through Facebook is effective and costs can fit the strict budget of researchers, unless there are highly specific inclusion and exclusion criteria.
Class B: evaluating effectiveness with costs per survey Two articles deter-mined the effectiveness of Facebook advertisement in recruitment based on the final costs per survey ([34], [35]). A summary of these articles can be found in Table 2.6. Both studies were performed in the USA, in 2018 [35] and 2019 [34], in areas of psychology and transplantology respectively. Both teams were able to reach a low cost (between 2,2 USD and 19 USD). Teo et al. used 15 different advertisements (a combination of three images and five text versions referring to potential motivations of participants) in order to find the best and most cost-effective recruitment method. Costs were 19 USD in general per completed survey and only 4,8 USD for the best combination of headline and image. Half of the survey participants were recruited with just 2 of the 15 advertisements.
In both cases, it was said that Facebook is an effective way to recruit participants for surveys and it is economically justified to use this recruitment method.
Class C: evaluating effectiveness with costs per click Measuring the effec-tiveness of Facebook’s advertisements is also possible by evaluating costs per click. Four articles were found and analysed in this section. Summary of these articles can be found in Table 2.7.
All articles are recently published, from the years 2018 and 2019. Three of them were published in the USA and one in Brazil. In all articles, researchers managed to perform the recruitment with less than 1 USD per click (for simplicity: 0,09 BRL equals 0,021 USD).
Articles of Lam et al. [38] and Cowie et al. [36] both conclude that recruiting through Facebook is suitable for an older population (above 45 years old).
It was noticed that more responses was received from women [39], and that women join research to support a higher cause [36], so they prefer more ’altruistic’ approach in proceeding with advertisement campaigns.
Based on articles authors’ experience, it can be said that recruitment through Facebook is effective, also for an older population.
Class D: evaluating feasibility, utility and viability with costs per partic-ipant The decision was made to merge the results of evaluating feasibility, utility and viability, because for each in the last two categories only one article was found, and therefore it would be difficult to draw conclusions. For each category cost per participant was a variable, which makes the merge justified. Summary of the articles can be found in Table 2.8.
Eight articles were evaluated in this section, four were published in the USA and four in Australia, both developed countries. Half of them are from the year the 2019, so they contain the most up-to-dated knowledge. The main focus of the research was on psychology and psychiatry.
The cost of advertisements varies from 1,86 AUD to 593 USD, where the 593 USD was used for recruitment for a randomized controlled trial [43]. It was proven that recruitment for a randomized controlled trial is possible through Facebook, however
costs are always higher than in any other type of studies, because of the precise inclusion and exclusion criteria (and therefore usage of expensive precise targeting). It is feasible to use Facebook in the recruitment of hard-to-reach populations ([40], [45]) and also in case of sensitive topic of the study [43].
Interestingly, three out of eight articles offered additional reward after completing the study ([40], [44], [45]). It had a possible effect on positive and faster response from potential participants.
It has to be mentioned that low cost of the recruitment does not necessarily relate to a high quality of the results/data [47]. In general, based on articles in this section it can be concluded that Facebook can be successfully used as recruitment tool in research in various clinical domains, e.g. psychology, oncology or epidemiology. Theme 5: promotion
Another way of using Facebook’s advertisements is to promote health-related ac-tions. Both articles from this section focused on public health and its aspects ([48], [49]). A summary of these articles can be found in Table 2.9.
The authors do not agree on the effectiveness of using Facebook’s advertisements on large populations. Alley et al. [48] claim that using them can be too costly (42 AUD). Graham et al. [49] state that it is suitable but then advertisement has to be well targeted to collect the correct sample size with lower costs.
Both researchers used targeted advertisements and they both concluded that they were cost-effective. In one article [48], cost per sign up was used as an outcome measure and in the second [49], cost per click. This makes them difficult to com-pare within this category, but if one specifically looked at cost of advertisements in recruitment, the promotion costs appear similar.
Barriers and facilitators
Barriers Unfortunately, researchers who would like to use Facebook in their re-search may encounter some obstacles. The first obstacle is the possible occurrence of a skewed population ([36], [27], [28], [38]). Some authors experienced over-recruiting white people and women without specific targeting, which has to be taken into con-sideration for creating future research. Additionally, Facebook tends to recruit a younger population in general [30], which can be caused by the trend that younger people spend more time on social media than older generations. Another problem connected with a population is that even with direct targeting, it is difficult to reach ethnic groups through Facebook advertisements, as some of these groups still prefer face-to-face contact [33].
A right balance has to be found between targeted and non-targeted advertise-ments, as the first are found to be more successful than the second [48], but they are also more expensive ([26], [39]), which brings the need of a higher budget. How-ever, at the early stages of a study, it is difficult to see who was missed by the advertisement [31], so it might be challenging for a beginner to correctly target the advertisement on the first try. Furthermore, budget and time of the year for publication of the advertisement have direct impact on its success [48].
Using other authors’ results in order to create and successfully carry out the advertisement process may be troublesome as Facebook is continually changing the
algorithms which conduct targeting ([24], [41], [29]), so reproducibility of results may be impossible to achieve.
Facilitators Probably the biggest advantage of using Facebook’s advertise-ments is the possibility of adjusting them during a trial ([24], [28]). This opens up great possibilities for researchers as they can react to changes in advertisement responses nearly in real-time.
Additionally, it allows to save money by being more precise in targeting a certain population [48]. Furthermore, Facebook allows to advertise in shorter time and with totally lower cost in comparison with traditional methods, which sometimes also requires more staff effort, so a larger budget is necessary ([24], [29]).
Another opportunity can be seen in the ease of spreading the news about the research via already interested participants [37]. Facebook’s features grant users options to share the link through private messages or on their private page, which will be then seen on other users’ feed. Furthermore, Facebook’s advertisements can deliver the information without actively seeking for it on the user’s side [49]. Facebook’s advertisements’ algorithms are tailoring and targeting the content of the feed to the personal preference of the users. Personal details or previous interest in a particular topic will lead to advertisements being displayed that lie within the user’s interests.
Author Year Country Clinical domain Methods Cost per participant/Results Discussion/Conclusions Adam
[24]
2016 Canada obstetrics Comparison of traditional meth-ods and targeted FB advertise-ments; measurements on cost per eligible participant; performance of FB ads was monitored using Facebook Ads Manager.
20 CAD for FB, 24 CAD for tra-ditional methods.
FB shortened recruitment time by almost a half. Similar demo-graphics for both groups.
Costs were similar but creating FB ad takes less time so costs per staff were cut; costs of ad may vary because of changes in FB al-gorithms, what makes future and past studies difficult to compare
Christen-sen [25]
2017 Denmark epidemiology Comparison between online (FB and other websites) and offline methods (press releases, posters, and flyers). For each recruitment method, costs per enrolled par-ticipant were calculated.
3,44 EUR nationwide for FB, 22 EUR from one city for FB. FB was second cheapest method of recruitment.
Offline methods are more labor-intensive for the research team - this factor was not taken into account in the costs reported in our study. Online methods seem more familiar to a young target population.
Derrick [26]
2017 USA psychology Comparison between two paid FB advertising tactics (“Pro-mote your Page” and “Send Peo-ple to Your Website”) and tar-geted direct mail recruitment.
181 USD per couple.
Demographic characteristics dif-fer per recruitment method. “Send People to Your Website” tactic was most cost-effective and most demographically di-verse.
Recruitment strategies that tar-get large populations tend to be more cost-effective. Highly spe-cific and restrictive eligibility cri-teria tend to make studies more expensive.
Frandsen [27]
2016 AUS psychology Comparison between traditional methods and advertisements on FB, (combination of words and pictures). Participants minimum 18 years old, living a maximum 25km from the recruitment site.
103 AUD per completed partici-pant.
Compared with individuals re-cruited through social media, a greater proportion of those who became aware of the study via traditional media were eligible, enrolled into, and completed the study.
A combination of strategies gives best results (traditional plus FB); using only social media gives a skewed population. Cost of traditional methods might be underestimated (unknown cost for staff, only cost of newspaper was known).
Juraschek [28]
2018 USA oncology Comparison between traditional methods and paid advertisement on FB. First just general, second targeted on particular zip codes (for Afro Americans) and on men only.
794 USD per enrolee for general advertisement and 1426 USD for targeted one.
Community fairs costed 917 USD, direct mail 799 USD and periodicals 436 USD.
Small changes in title and text might have a big impact on re-cruit rates. Micro-trials allow to adjust the ad to the audience and therefore gain more participants. Simply changing target audience is not enough to overcome re-cruitment barriers.
Kayrouz [29]
2016 AUS psychology Comparison between traditional methods and 4 FB strategies to advertise.
37 AUD
Two FB strategies (boosting posts and promoting FB public fan pages) seemed to have the largest impact on recruitment rates. FB recruitment was 2.5 times faster but not more cost-effective than traditional meth-ods.
FB provides an opportunity for potential applicants to become better informed about the re-search and the opportunity to develop a sense of the credibility and trustworthiness. The algo-rithms used by FB in promoting pages are known to change regu-larly.
Nash [30] 2017 AUS cardiology Traditional methods compared with advertisements on FB (right side panel and panel in main stream on university page).
81-176 AUD (depending on the city)
FB was successful in middle-to-older age participants, but vary by location.
FB was able to recruit a slightly younger population than tradi-tional methods (2 years younger)
Reiter [31]
2017 USA epidemiology Targeted FB advertisements with different text and photos.
2,76 USD
Advertisements that included an image of a couple or a group were better than ones that included an image of a single person.
It is important to pay attention to content of the advertisement and they should feature individu-als who are racially or ethnically diverse. FB advertisements are a convenient and cost-efficient strategy for recruiting young gay and bisexual men.
Tsai [32] 2019 USA oncology 5 FB advertisements in 5 differ-ent languages for 2 months, tar-geting Chinese, Korean and Lat-inx cancer survivors, residents of the USA. Completing the sur-vey resulted in Walmart 10 USD eGift card
47 USD
The East Asian and Latinx can-cer survivors did not significantly differ in age, years lived in the USA, education level, generation status, and time since diagnosis.
FB was successful in recruiting these ethnic minor populations, from different parts of the USA. Skewed population: more lati-nos with low income and more women with breast cancer. Volkova
[33]
2017 New Zealand
nutrition Traditional methods (print me-dia, team networks and radio) compared with advertisements on FB.
5 NZD.
Facebook resulted in the highest absolute registration, randomisa-tion and study complerandomisa-tion num-bers, both overall and for each ethnic group.
FB advertising was most effec-tive and least costly. Traditional methods might still be necessary for ethnical groups; combining traditional methods and online methods might be the best solu-tion.
Table 2.5: Summary of the articles evaluating effectiveness of Facebook (FB) recruitment with costs per participant.
Author Year Country Clinical domain Methods Cost per survey/Results Discussion/Conclusions Chung
[34]
2019 USA transplantology Using 3 FB approaches to re-cruit: regular ad, support groups and FB page of the study.
2,2 USD per ad, 2,9 USD per pro-moting study page
Costs were economically feasible even for those with limited funds.
As the goal was to recruit for the survey, it is better to do surveys on PC rather than phones. Teo [35] 2018 USA psychology 15 versions of advertisement on
FB (3 different pictures and 5 different headlines for motives of taking the survey. Advertise-ment appeared only in news feed on computers.
19 USD in general per completed survey and only 4,8 USD for the best combination of headline and image.
Half of survey participants were recruited with just 2 of the 15 ads.
FB advertisements were effec-tive in rapid and cost-effeceffec-tive veterans enrollment. Images and headlines help with target-ing particular sub-groups.
Table 2.6: Summary of the articles evaluating effectiveness of Facebook (FB) recruitment with costs per survey.
Author Year Country Clinical domain Methods Cost per click/Results Discussion/Conclusions Cowie
[36]
2018 USA geriatrics Advertisements on FB: two dif-ferent types of approach, one ’typical’ recruit and one ’altru-istic’, so targeting people which want to help for bigger cause.
0,91 USD for typical approach, 1,27 USD for altruistic one. Suitable for 60+ years old popu-lation.
Tendency to over-recruit white people. Women slightly fa-vored the altruistic campaign, and men favored the typical cam-paign. People aged 55-64 are twice as likely to engage with sponsored Facebook advertising than younger adults.
Isaacson [37]
2018 USA neurology Targeted advertisements on FB (US population more than 25 years old, who previously liked the study page).
0,62 USD
Quick and cost-effective in re-cruiting for educational pur-poses.
Recommended to ask users to share and forward the posts. Testing on a broad range of de-vices and software is essential for positive user experiences. Lam [38] 2018 USA geriatrics Targeted advertisement on FB
to attract users to a video on Youtube.
0,06 USD
FB was able to attract the cor-rect population (Chinese elderly) at low costs.
Facebook could be an alternative way to reach older population de-spite the common belief that the geriatric population does not use social media
Machado [39]
2019 Brazil psychology Two advertisements on FB; one with loss-framed messages and one with gain-framed messages about smoking.
0.12 BRL per gain-framed ad and 0.09 BRL per loss-framed ad.
Loss-framed ads reached more users and got more clicks.
More specifically targeted ads lead to higher costs. Usu-ally higher response rates among women. Higher interest among older population as they really want to quit smoking.
Author Year Country Clinical domain Methods Cost per participant/Results Discussion/Conclusions Akard
[40]
2015 AUS oncology Targeted advertisements on FB which lead to a survey on an-other website. Chance to win an Apple gift card.
17 AUD
From 1050 clicks on the adver-tisement, 106 participants met inclusion criteria and 45 of them completed whole survey. Process took 10.5 weeks.
FB eligible for different types of studies, particularly for rare populations. FB increases sam-ple diversity and by so generaliz-abililty of study results. FB re-quires minimal staff which cuts costs.
Bennetts [41]
2019 AUS nursing and parenting
Comparison of 20 paid and free advertisements on FB, managed via ad manager, which allows targeting. Unique url for each strategy.
2,32 AUD
Paid FB ads recruited nearly 75% of the sample. The combi-nation of paid and free FB ads proved to be an effective strat-egy for recruiting a large sample of employed parents.
Some FB users impulsively clicked on the link, compared to traditional method when people need to take more action; it also means less engagement for the future follow-ups.
Das [42] 2017 AUS epidemiology Targeted (only men) advertise-ment on FB with redirection to a special website to express in-terest and if so then survey on an external website.
48 AUD
Feasible and efficient strategy for recruitment.
FB could be used for monitoring the impact of HPV vaccination. Higher cost than for female stud-ies.
Guthrie [43]
2019 USA gynecology Comparison of 3 different FB ad-vertisements and regular mailing in recruitment women with vagi-nal problems
593 USD per randomized partic-ipant.
Considering more barriers than other studies (sensitive topic), researchers still managed to re-cruit participants. Rere-cruitment to a clinical trial testing is feasi-ble through social media adver-tising.
Oesterle [44]
2019 USA psychology Advertisement on FB with redi-rection to a different website for eligibility testing. Chance to earn up to 60 USD for taking 3 surveys.
42 USD per consented partici-pant.
Equal to better results than tra-ditional methods but did not achieve a bigger population di-versity.
Problem-focused language (e.g., “mental health problem”) per-formed better than using posi-tive language. Ads placed in the context of a current and locally relevant issue are probably more effective than generally framed ads.
Thornton [45]
2019 AUS psychology Recruitment through an adver-tisement on FB, a local research register and university psychol-ogy courses. Additionally a chance to win an iPod.
1,86 AUD
There were significantly fewer women in the Facebook sample.
The chance to win an iPod may have contributed significantly to the success of the campaign. FB is potentially efficient and cost-effective for large samples and hard-to-reach populations.
Author Year Country Clinical domain Methods Cost per participant/Results Discussion/Conclusions Bold [46] 2016 USA psychology Advertisement on FB,
redirec-tion to another website for sur-vey.
35 USD
Highly cost-effective, but ques-tionable diversity of the sample.
Advertisement were cost-effective and successful for reaching adult smokers in-terested in treatment. High potential also for smokers who heavily drink alcohol.
Crosier [47]
2016 USA psychiatry Advertisement on FB, redirec-tion to another website for sur-vey. Additional constant adjust-ment of the content of the adver-tisement.
8 USD
Good for reaching hard-to-reach population, scalable and efficient method.
Low costs does not mean high quality data. Systematic test-ing can fine-tune the ad strategy. Suggestion for ’carousel style’: constant change in image on the ad.
Table 2.8: Summary of the articles evaluating feasibility, utility and viability of Facebook (FB) recruitment with costs per participant.
Author Year Country Clinical domain Methods Cost per participant/Results Discussion/Conclusions Alley [48] 2016 AUS physical
educa-tion
Comparison of Web-based and print-based methods to attract people to physical activity. For first four month FB ad was not targeted, then targeted.
42 AUD for targeted ad, 68 AUD for not targeted one. The targeted Facebook ad-vertisements were the next most cost-effective method and reached a large number of sign-ups
FB might not be suitable for large population-based interven-tions. Ads more effective when targeting was used. Smaller ads more effective than these on feed.
Graham [49]
2019 Canada obstetrics/public health
Evaluation of usage of advertise-ments on FB and Google. Cam-paign in 3 phases, 8 weeks each (in order to avoid fatigue and overexposing).
0,35 CAD for FB ad, 0,76 CAD for Google ad
Hiring a vendor rises costs but can be beneficial for those with no experience in FB ads. Good to made campaign in phases, that gives possibility of adjusting and building best targeted ad.
Table 2.9: Summary of the articles using advertisements on Facebook for promotion
2.4
Discussion
Main findings This research focused on examining the possible usage of Face-book as a research tool within different fields of studies, from randomized controlled studies in cardiology to psychiatry interventions on young military veterans. This scoping review resulted in the observation that Facebook can be used in several ways, mainly through advertisements and creating dedicated groups and posts. The analysis performed by researcher and supervisor led to five main themes, where Facebook was used in sharing knowledge, gathering opinions, evaluating engage-ment and satisfaction, recruitengage-ment of study participants and promotion of health related actions. Furthermore, several barriers and facilitators were discovered. The main barriers were: possible imbalance in population demographic features, higher costs for more specific studies, troubles with relying on past studies due to constantly changing Facebook algorithms. Despite the barriers, facilitators look promising and encouraging: flexibility of the usage (also during on-going trials), cost-effectiveness, ease of use, accessibility and a possibility of gathering big samples. Unfortunately, none of the found articles touched on the subject of recruitment for UCD studies for medical research.
The potential of Facebook and other social media should not be dismissed. They have a global range, with a large and diverse population. Currently, almost every-body is using some type of social media on their smartphone, what makes the pos-sible communication even easier to establish. It all creates the perfect opportunity for researchers.
The possibilities of Facebook in research are growing with time. In the begin-ning, the most obvious function was to recruit and advertise about the studies [50]. With the development of Facebook and growing needs of scientists, more ways were found to use it as a research tool. Nowadays, researchers use Facebook to simply communicate with patients, ask for their needs and answer them [51], gather quan-titative and qualitative data [52]. However, this review showed that recruitment is still the most popular usage of Facebook an also the best examined ([53], [54]).
In general, it can be concluded that Facebook is a promising channel for reaching participants or patients, in very different forms. For the purposes of this scoping review, 70% of found articles debate on the recruitment of the participants for clinical trials, interviews, surveys. However, bringing Facebook in research can lead to the addition of selection bias and self-selection bias. Using the option to target on Facebook might be the first solution, with collecting a bigger sample being the second. But for both solutions, there is one more underlying problem, the authenticity of the profiles. Researchers have to fully rely on information given by Facebook users, which sometimes might dishonestly provide them, for example to protect their privacy. This aspect of using Facebook in research needs further investigation, in order to produce more accurate results.
Another aspect of using Facebook that needs additional examination, as only 23% of found articles debate on gathering opinions, sharing knowledge and eval-uating patients’ engagement and satisfaction. More evidence is needed to make researchers more keen on using Facebook as a supportive tool in research.
It is also interesting that the majority of studies were conducted in the USA and Australia. Only two out of 34 articles were conducted in Europe and also two in South America. That makes the generalization of result more difficult for these
regions. Additionally, in 2018 the General Data Protection Regulation (GDPR) came into force in European Union, which may had influenced ongoing research, and possibly fewer publications from Europe in that period of time. The question should be answered whether or not GDPR has an influence on using Facebook as a tool in Europe.
It has to be mentioned that Facebook is not the only type of social media, which can be used in research. Other authors are mentioning also blogs and microblogs (like Twitter), content communities (like Instagram) or virtual game-worlds (like World of Warcraft) among others [7]. This variety creates possibilities for researchers to choose the media which suits their studies’ needs best.
Strengths and limitations This study’s value lies in its fresh approach, trying to summarize the knowledge in a systematic way. The scoping review provided a helpful overview for current and future researchers who would like to use Facebook in their research. However, some limitations are worth mentioning. Only one reviewer conducted the search, inclusion and exclusion process and final summary, what may have influence on objectivity of the results. The used query was largely open and could overlook some important articles and findings. At the same time, only two databases were explored, which may limit the number of found relevant articles. Implications The results of this study can be used by future researchers as a preparation for their prospective work with Facebook, for those who need some basic guidance in this topic and also for those who would like to continue studying the usage of Facebook in health care research, as a baseline for their investigation.
Chapter 3
Recruitment of participants on
Facebook: a comparison study.
3.1
Introduction
Recruitment of suitable participants is one of the biggest challenges in medical stud-ies, also for those focused on user centered design, as the potential end-users are playing the key role in the whole process [55]. As it was found in several studies and summarized in the previous chapter of this thesis, paid advertisements are the most effective way to engage participants and offer the most relevant opportunity for gaining insights into the context of use in the user centered design processes. However, other methods also appeared effective enough to offer an interesting al-ternative. The studies analyzed in the previous chapter did not compare possible recruitment methods between each other, which is particularly important when try-ing to choose the most suitable method for human factor research studies in the future. Additionally, by comparing the different approaches, we can answer the question which one is most effective to apply in the first phases of user centered design and will help in reaching out to the most appropriate end-user groups.
The year 2020 will be remembered as the pandemic time all around the world. SARS-COV-2 virus influenced the lives of millions of people. The current spread of the virus also affected the usage of a broad spectrum of eHealth tools. In the Netherlands, every non-urgent visit to a general practitioner and most of the spe-cialists was conducted online or via telephone [56]. From one day to another the whole health care system had to rely on more remote methods.
Because people are gradually becoming more experienced in using eHealth tools, it might therefore be relevant to see this from the perspective of their willingness to share information. This is a fundamental part of reaching out to potential end-users of an eHealth application in user centered design. In the first phase, UCD aims to gain insight into users’ experiences and thoughts about the context of use and experiences with eHealth tools.
Trust plays an essential role in today’s research, both in the recruitment process and later on in conducting surveys, interviews and especially in medication trials. It is necessary for participants to trust the researchers in order to create a space for openness and encourage their willingness to share personal information. It is known that patients with more sensitive health information (such as HIV related records) are more reluctant to share their personal information [57]. Additionally, the lack
of transparency today, caused among others by the technical difficulty of creating systems patients can understand, has a negative influence on end-consumers and their trust to share the information [58]. This might be especially problematic in the case of Facebook, which is potentially a recruitment place for research.
The objective of this research is to compare the advertisement and direct recruit-ment via groups on Facebook in order to investigate which of them is better attuned to reach out to potential end-users and to gaining sight on information relevant for human factor research on eHealth.
The main questions of the research are: 1) What are the differences and similar-ities in population characteristics and effectiveness of two participants recruitment possibilities on Facebook: via advertisements and via direct contact on groups?, 2) How popular is actual usage of eHealth tools and how does it affect the future re-search of eHealth tools from a human factors rere-search perspective?, 3) How willing are people to share their personal information in the survey on Facebook?.
3.2
Methods
3.2.1
Recruitment
In order to answer the research questions about the effectiveness of two different recruitment methods, two main ways of contacting participants were established. Recruitment via an advertisement of Facebook (Group 1)
A private Facebook account was used to create a new sub-website (private webstie) on Facebook. This step was necessary for the creation of the advertisement. Further on, the main features of the advertisement were chosen and one of the links from the Google Forms platform were connected with the advertisement. It is important to mention that the advertisement was published only on Facebook and not on additional platforms owned by Facebook, such as Messenger or Instagram, in order to gain a clear view of the potential of the recruitment of Facebook exclusively. The advertisement which was viewed by potential participants can be seen on Figure 3.1. A budget of 10 euro per day was utilised and the advertisement was scheduled for publication. The advertisement was published twice for a period of seven days. As the first round did not gain enough answers, the decision was made to republish the advertisement for a second time with the same period of seven days. The targeting of the first round was as follows: people using Facebook in the Netherlands, age above 18 years old. For the second round, the targeting was adjusted and included a population up to 55 years old.
Figure 3.1: The advertisement placed on Facebook platform. Recruitment via direct contact on Facebook groups (Group 2)
As the goal was to gain as broad sample of the population living in the Netherlands as possible, only the groups with a nationwide reach were chosen to make possible contact with participants. Out of the ten groups that were chosen, only four posts were allowed to be published by the groups’ administrators. The main reason for rejection was the prohibition of posting any surveys, as the regular policy of the group. The post was published from a private account, with a brief description of the purpose of the study, an invitation for the survey and the second link from the Google Forms platform. In the first round, the posts were published twice, on the first and fourth day of the week. The original post was removed and replaced with an exact copy. The second round followed the same pattern.
3.2.2
Survey
For the purposes of this study, a new questionnaire was developed. It consisted of four groups of questions, focusing on: 1. demographic details of participants (nominal and open-ended questions), 2. their thoughts on eHealth (yes/no and ended questions), 3. their thoughts on research on Facebook (nominal, open-ended and Likert scale questions), 4. possible contact for purposes of next studies (yes/no and open-ended questions). For an overview of all questions see Appendix B.
Questions were designed in the way that participants’ answers could directly answer research questions stated for this research. Questions were developed by the researcher and discussed with the supervisor.
After survey development, a test phase was conducted by four individuals outside of the project to get an unbiased opinion and in order to detect possible language mistakes, interpretation of the questions and general survey flow. The revised survey
was then approved by the researcher’s supervisor. For the need to use the survey on Facebook, two copies of the survey were placed on the Google Forms platform of the researcher.
Before starting the survey, participants were informed about the content of the survey, its purpose, length and were assured that their answers will be treated completely confidential. Additionally, the private email address of the researcher was given to allow direct contact. At the beginning of the survey’s subsections, short explanations of difficult terms related to eHealth jargon were provided, together with the survey part’s purpose.
3.2.3
Theoretical framework
In order to analyze the results of the survey, the theoretical framework established by Zhu et al. was used [12], previously described in the first chapter of this the-sis. The focus was on the influence of privacy concerns on the willingness to share the information. Participants were not asked to share their medical history thus disease severity as a moderator variable will not be discussed. Also, as questions about demographics were mandatory to answer and other sensitive information was not questioned, the information sensitivity as moderator variable will also not be discussed.
In order to better understand the intentions of the survey’s participants, the possible answers/behaviors were preliminarily divided into four categories: user is willing to give the information and is giving it, user is willing to give the information but is not giving it, user is not willing to give the information but is giving it anyway, user is not willing to give the information and is not giving it. Consistency in an attitude followed by actions was one of the most important aspects taken into consideration during analyzing the responses to the survey. Additionally, the openness of users to consider Facebook as the optional contact platform played an important role in evaluating the users’ answers.
3.3
Results
3.3.1
General statistics and remarks
Group 1 As a result of the advertisement on Facebook, ten participants were recruited, 70% of them were female. Users in age category 18-25 constituted 40% of the overall group. The percentage spread of age can be seen on Figure 3.2. Seven of the participants were Dutch, one Polish, one Greek and one Czech. All participants stated that they use Facebook several times a day. 50% of the participants had bachelor degrees, 30% master degrees, one finished high school. One’s degree is not known as the participant stated only “degree”. In general, the advertisement met very unfriendly responses from some Facebook users, who left eight, highly rude comments under the advertisement. Most of the comments were directly connected with the privacy concerns about sharing medical data on Facebook.
Group 2 As a result of the recruitment via groups, thirteen participants were recruited, 77% of them were female. More than 60% of the participants were in the 18-35 age division. The percentage spread of age can be seen on Figure 3.2.
