View in EDS HTML Full Text PDF Full Text

Bots and Fake Participants: Ensuring Valid and Reliable Data Collection Using Online Participant Recruitment Methods

Saved in:
Bibliographic Details
Title: Bots and Fake Participants: Ensuring Valid and Reliable Data Collection Using Online Participant Recruitment Methods
Language: English
Authors: Roseline Jean Louis (ORCID 0000-0002-6483-7024), Lisa M. Thompson (ORCID 0000-0002-8001-2057)
Source: International Journal of Social Research Methodology. 2025 28(4):463-473.
Availability: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
Peer Reviewed: Y
Page Count: 11
Publication Date: 2025
Sponsoring Agency: National Institute of Nursing Research (NINR) (DHHS/NIH)
Contract Number: F31NR020575
Document Type: Journal Articles
Reports - Research
Descriptors: Research Problems, Recruitment, Selection Tools, Internet, Social Media, Online Surveys, Artificial Intelligence, Deception, Credibility, Research Administration, Research, Research Projects, Medical Research
DOI: 10.1080/13645579.2024.2410176
ISSN: 1364-5579
1464-5300
Abstract: Recruitment for successful health sciences research requires balancing efficiency, cost, accessibility, and reliability of available recruitment methods. Our case-control study used online recruitment methods, which broadened our reach to potential participants across the United States. However, this approach also exposed us to challenges associated with bot interference and fraudulent participation. This paper focuses on maintaining data integrity, specifically when utilizing online participant recruitment methods. Drawing from our experience, we propose The Swiss Cheese Model of Study Participant Fraud Prevention, adapted from Reason's Swiss Cheese Model, and illustrate 10 prevention and verification measures that can be taken to minimize fraud in research studies that rely on online recruitment. We emphasize the importance of a layered approach, including carefully designed recruitment media and compensation protocols, vetting of participant eligibility, and data verification protocols to ensure the validity and reliability of research findings in the digital age.
Abstractor: As Provided
Entry Date: 2026
Accession Number: EJ1495994
Database: ERIC
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
    Url: https://content.ebscohost.com/cds/retrieve?content=AQICAHj0k_4E0hTGH8RJwT4gCJyBsGNe_WN95AvKlDbXJGqwxwE5OKdk8sHgQrphzc7YWtUwAAAA4zCB4AYJKoZIhvcNAQcGoIHSMIHPAgEAMIHJBgkqhkiG9w0BBwEwHgYJYIZIAWUDBAEuMBEEDBlxmq3lv7NnfSBebAIBEICBm9fJTQM3x9injlwQPbw0O344bWsb6KJhORz0jqdl6AbZOq9bBf5VEu2XCrBeIXcnP6sfSD4O8F5qAK5tZ-0Y4HoSf76TC6TZxpVKhKszkGRgw-Wm6yYk6GWcSK3yX7o6x3oNVPef_KcWmriCK0k2cPnl3KKXlK5ouqXSLx2nb0XYS7lB5sRmWg_xVNwCwgDlHJ7VT2uig7hg517p
Text:
  Availability: 1
  Value: <anid>AN0185986534;9eb01jul.25;2025Jun19.01:57;v2.2.500</anid> <title id="AN0185986534-1">Bots and fake participants: ensuring valid and reliable data collection using online participant recruitment methods </title> <p>Recruitment for successful health sciences research requires balancing efficiency, cost, accessibility, and reliability of available recruitment methods. Our case-control study used online recruitment methods, which broadened our reach to potential participants across the United States. However, this approach also exposed us to challenges associated with bot interference and fraudulent participation. This paper focuses on maintaining data integrity, specifically when utilizing online participant recruitment methods. Drawing from our experience, we propose The Swiss Cheese Model of Study Participant Fraud Prevention, adapted from Reason's Swiss Cheese Model, and illustrate 10 prevention and verification measures that can be taken to minimize fraud in research studies that rely on online recruitment. We emphasize the importance of a layered approach, including carefully designed recruitment media and compensation protocols, vetting of participant eligibility, and data verification protocols to ensure the validity and reliability of research findings in the digital age.</p> <p>Keywords: Online recruitment; social media recruitment; bot infiltration; fraudulent participants; data validity and reliability</p> <hd id="AN0185986534-2">Introduction</hd> <p>Recruitment for successful health sciences research requires a delicate balancing act of efficiency, cost, accessibility, and reliability of available recruitment methods. Although face-to-face recruitment has traditionally been considered the most reliable method for assessing participants' eligibility, it is often impractical due to high costs, time constraints, and the necessity to restrict recruitment to specific geographic areas accessible to the study team. In today's digital era, there are expedient and effective means to recruit potential participants on a wider scale (Couper, [<reflink idref="bib5" id="ref1">5</reflink>]; Toepoel, [<reflink idref="bib26" id="ref2">26</reflink>]). Using online recruitment strategies, such as paid and unpaid digital advertisements, as well as posting flyers on various online platforms, presents avenues to engage a larger target population (de Leeuw, [<reflink idref="bib6" id="ref3">6</reflink>]). These approaches not only save time and resources but also facilitate recruitment of demographic groups who are typically harder to reach. Nonetheless, researchers must exercise caution since online recruitment methods can potentially compromise the collection of accurate and valid data (Waggoner et al., [<reflink idref="bib28" id="ref4">28</reflink>]; Xu et al., [<reflink idref="bib31" id="ref5">31</reflink>]).</p> <p>The Respect Black Mamas Study used online recruitment methods to conduct a case-control study to examine the effects of racially discriminatory and disrespectful maternity care by healthcare professionals (nurses, midwives, and physicians) on severe maternal morbidity outcomes among Black birthing people in the United States. We chose online platforms to broaden our potential reach to Black birthing people of reproductive age across the US, but it exposed our study to bots and fraudulent participants who, in most cases, submitted unreliable and false responses to our surveys. Drawing from our experience, the purpose of this paper is to delineate the specific steps in our online recruitment strategies and data collection processes that led to the identification of this issue during the implementation of our study. Furthermore, we introduce our Swiss Cheese Model of Study Participant Fraud Prevention, which we adapted from Reason (Reason, [<reflink idref="bib20" id="ref6">20</reflink>]). This model outlines recommendations for minimizing fraud in studies that rely on online recruitment.</p> <hd id="AN0185986534-3">Recruitment design and methods</hd> <p>This Study employed a case-control study design with a nested qualitative component to collect both quantitative and qualitative data from 100 self-identified Black birthing people who gave birth in the U.S. Inclusion criteria included English-speaking birthing people aged 18 or older, self-identifying as Black or African American, and having delivered a baby in a U.S. hospital within the past year. We excluded those who did not receive prenatal care. For recruitment, we utilized an internet-based nonprobability convenience sampling approach to identify eligible participants for the quantitative portion of the study. Following this, we employed purposive sampling to select 15–20 birthing individuals based on high scores on the Mothers on Respect (MOR) Index (indicating low to very low levels of respect) and low scores on the Discrimination in Medical Settings scale (DMS) (indicating high levels of discrimination) for one-on-one online interviews (Peek et al., [<reflink idref="bib18" id="ref7">18</reflink>]; Vedam et al., [<reflink idref="bib27" id="ref8">27</reflink>]). Given that our target population, Black birthing people, may bear an additional economic burden for research participation (Hughes et al., [<reflink idref="bib12" id="ref9">12</reflink>]), we offered monetary compensation for their time in the form of electronic gift cards: $25 for completing the online survey and an additional $25 for participating in a qualitative interview. Acknowledging that participants in our study were self-selected, we are aware of the potential bias introduced by economic and topic interests (Hsieh & Kocielnik, [<reflink idref="bib11" id="ref10">11</reflink>]). Nevertheless, to increase participant engagement in this on-line study, we prioritized compensating individuals for their time, without exerting undue influence.</p> <p>The study team designed a digital recruitment flyer and distributed it across targeted social media platforms and groups. The flyer contained an explanation of the study's purpose, detailed inclusion and exclusion criteria, the expected participant tasks, detailed information about compensation, a clickable web link, and a QR code that linked to REDCap™, a secure, cloud-based platform housing all the study consent forms and surveys (Harris et al., [<reflink idref="bib9" id="ref11">9</reflink>], [<reflink idref="bib8" id="ref12">8</reflink>]).</p> <p>When a participant either clicked on the live link or scanned the QR code, they were directed to a screening consent form followed by a screening survey. The REDCap™ system was configured to pause participants if they provided answers that would make them ineligible, notifying them of their ineligibility. However, participants had the opportunity to correct their previous responses in case they made an error. To streamline the enrollment process, we automated participant enrollment through REDCap™ based on preset screening criteria, enabling them to proceed with the remaining study surveys after they responded to screening questions for eligibility.</p> <hd id="AN0185986534-4">Wave one recruitment</hd> <p>The digital study flyer was disseminated on 29 August 2022, at 10 AM through established social media channels on Facebook, LinkedIn, and Twitter (now named 'X'). By 4 PM on the same day, 27 participants consented. By midnight, 59 participants consented, and by 30 August 2022, we met our planned study sample of 100 participants. All 100 participants met the inclusion criteria and were invited to enroll in the study. Out of the 100 invited to enroll, 91 participants responded with complete survey data.</p> <p>The entire process, from posting the recruitment flyer to participants completing all the study surveys, took place in less than a day and a half. Because we had so easily met our target sample, which seemed plausible since we used platforms that could reach a wide swath of our proposed target population, we modified our IRB protocol to expand recruitment to up to 500 participants, secured funds to pay them, and reopened recruitment with the intention of manually sending invitations to enroll after IRB approved the changes. While awaiting approval for our IRB modification, an additional 114 participants completed the screening survey, all providing answers that qualified them to participate in the study.</p> <p>In the first week of data collection, we began the process of data cleaning and observed discrepancies. Participant names displayed variations of the study team's names as listed on the study consents. Moreover, many email addresses followed a similar pattern of combining first and last names ending with four to six random digits, but not resembling a possible date of birth. Upon close examination of survey responses, it became apparent that individual participants provided conflicting information within their survey responses. For instance, in the MOR index, a respondent indicated receiving care from a physician but not a midwife, while in a DMS survey, they stated the opposite. Furthermore, Likert scale responses frequently remained constant throughout the entire scaled survey across respondents, lacking the expected response variability.</p> <p>After labeling the data inconsistencies, the study team implemented a data cleaning protocol. This protocol involved contacting the 91 participants who had responded with complete data through both telephone calls and emails to verify the discrepancies. From this group, only 8 participants (8.8%) were identified as valid participants with usable data. Among the remaining 83 respondents with data discrepancies, 61 (73%) provided disconnected phone numbers, while 22 (27%) provided Google Voice numbers that went unanswered.</p> <p>We then began to verify the eligibility for the 114 participants who had met eligibility through the screening procedures, but had not yet been invited to enroll, as well as the original 9 participants who had been automatically invited but did not respond. All 123 of these participants provided disconnected or Google Voice phone numbers that went unanswered. At this point, recruitment for the study was temporarily paused until we could establish enhanced verification measures.</p> <hd id="AN0185986534-5">Wave two recruitment</hd> <p>Given that social media-based recruitment was found to be unreliable, the study team implemented two major changes to our recruitment methods. First, we added a requirement for a Zoom interview with participants, during which their camera had to be on, to confirm eligibility before extending an invitation to enroll. We removed flyers from social media platforms and began recruiting through ResearchMatch, a national health volunteer registry that was created by several academic institutions and supported by the U.S. National Institutes of Health as part of the Clinical Translational Science Award (CTSA) program (ResearchMatch, [<reflink idref="bib21" id="ref13">21</reflink>]). ResearchMatch has a large population of volunteers who have consented to be contacted by researchers about health studies for which they may be eligible. Once researchers register their IRB-approved studies with ResearchMatch, they can search for potential participants based on study criteria.</p> <p>On 16 February 2023, we conducted a ResearchMatch search, yielding 142 seemingly eligible participants. Over the next week, 20 eligibility screening interviews were completed by the primary author. All 20 potential participants were found to be ineligible for various reasons. Some participants were unable to provide consent due to a lack of proficiency in English, which was one of the eligibility criteria. Additionally, more than half of the potential participants declined to turn on their cameras, as stipulated in the informed consent process. Lastly, some individuals struggled to recall basic information, such as the infant's delivery date or the U.S. hospital where they delivered their infant child – details they had previously provided in the screening survey before the call.</p> <p>Given our observation of potential fraudulent participation via ResearchMatch, we once again paused recruitment before relaunching the survey. This time, we utilized a combination of internet-based personal networks and employed the snowball method for recruiting, with enhanced screening verification measures in place. Drawing from our experience, we offer best practices for researchers intending to utilize online surveys for their research studies.</p> <hd id="AN0185986534-6">The Swiss cheese model of study participant fraud prevention: lessons learned and future reco...</hd> <p>The Swiss Cheese model, initially conceptualized by James Reason, serves as a powerful illustration of how employing multiple layers of defense can effectively thwart accidents or failures (Reason, [<reflink idref="bib20" id="ref14">20</reflink>]). In this model, each defensive layer is metaphorically depicted as a slice of Swiss cheese, characterized by inherent vulnerabilities or 'holes' that may permit potential hazards to pass through. During our initial recruitment phase, we lacked measures to deter fraud. Even as we implemented additional layers of protection in later attempts, we failed to consider how these layers might align in a way that still allowed for potentially fraudulent activity. Recognizing that no single method for preventing fraud is entirely foolproof and acknowledging that efficacy can vary depending on the specific nature of the research study, we found it prudent to adopt a combination of strategies. In the following sections, and based on our experience, we summarize our recommendations for a fraud deterrence plan based on the Swiss Cheese Model of Study Fraud Prevention model to filter out bots and fraudulent participants (Figure 1). There are six prevention measures that can be taken at points of recruitment and data collection and four verification measures that can be taken during and after data have been collected. These steps are based on our experience and offer suggestions for researchers to consider for their own research studies.</p> <p>Graph: Figure 1. The Swiss cheese model of study participant fraud prevention.</p> <hd id="AN0185986534-7">Carefully craft recruitment media</hd> <p>In our initial approach, we focused on crafting a detailed digital recruitment flyer that aimed to equip potential participants with comprehensive information to help them make an informed decision about their interest in the study. We recognized the importance of presenting enough information to capture genuine interest and establish trust and credibility. However, we also observed that providing too much detail inadvertently enabled fraudulent participants and bot software to correctly respond to all eligibility screening questions. We recommend that recruitment media provide a broad overview of the study, highlighting only a specific subset of the inclusion criteria rather than all of them. This balanced approach allows researchers to strike a balance between generating genuine interest and deterring potentially fraudulent participation.</p> <hd id="AN0185986534-8">Consider the risks of online recruitment platforms</hd> <p>It is crucial to choose the platforms where you advertise the study and keep a record of dates that recruitment flyers were posted to each site. While social media platforms grant access to a wide audience of potential participants, we observed that these platforms are susceptible to infiltration by bots and fraudulent participants or scammers, especially on Facebook and Twitter (now called 'X') (Whitaker et al., [<reflink idref="bib30" id="ref15">30</reflink>]). If sharing on social media platforms, consider posting to closed targeted groups, restricting sharing capabilities of the post and incorporate a screening question to your survey asking participants where they saw the flyer (Xu et al., [<reflink idref="bib31" id="ref16">31</reflink>]). While this may not eliminate the possibility of participants sharing a screenshot of the flyer, the study team should remain vigilant if a participant mentions a platform or group that was not initially posted by the study team.</p> <hd id="AN0185986534-9">Create comprehensive compensation protocol accounting for fraud</hd> <p>While compensating participants for their time is essential, particularly for Black-identifying individuals who may have valid concerns about enrolling in research due to a history of exploitation, it is vital to carefully phrase the compensation amounts and participation procedures (Hughes et al., [<reflink idref="bib12" id="ref17">12</reflink>]; Langer et al., [<reflink idref="bib13" id="ref18">13</reflink>]; Nadeem, [<reflink idref="bib16" id="ref19">16</reflink>]; National Academies of Sciences et al., [<reflink idref="bib17" id="ref20">17</reflink>]; Scharff et al., [<reflink idref="bib22" id="ref21">22</reflink>]). Listing dollar amounts can potentially motivate respondents to complete surveys even if they do not meet the eligibility criteria.</p> <p>Given our experiences, if participants are to be fairly compensated for their participation, we recommend establishing clear and detailed procedures for participant remuneration and maintaining close coordination with Institutional Review Boards about modifications to research protocols to ensure data integrity. This should include clauses specifying that payment may be denied if the provided data is deemed unusable and cannot be verified. Additionally, mailing remuneration to physical addresses within the geographic region covered by the study can help reduce the risk of fraudulent claims.</p> <p>In our study, we did not specify the reasons why a participant might not receive compensation. This led to instances where fraudulent participants contacted the study team immediately after completing surveys, demanding immediate payment, despite the consent form indicating a 30-day turnaround. Additionally, we encountered 12 fraudulent participants who provided disconnected phone numbers and refused to address data discrepancies when requested to provide a valid contact number via email. These individuals then contacted the Institutional Review Board to ask for payment, asserting that the study team should have thoroughly verified their eligibility for study inclusion before permitting them to complete the surveys. This suggests that fraudulent participants are targeting studies that lack thorough verification protocols to receive compensation they are not entitled to. This is a cautionary lesson for new and experienced researchers, as well as members of Institutional Review Boards.</p> <hd id="AN0185986534-10">Manually enroll eligible participants</hd> <p>Data collection systems like Qualtrics and REDCap™ offer the option of partial or complete automation in data collection, allowing participants to screen in and complete surveys without embedded verification checks. Hence, it is crucial to use these features judiciously and with precautionary measures. First, we recommend avoiding automatic stops for screening questions and permitting potential participants to revisit previous screening questions until they answer them 'correctly' (albeit untruthfully). While this may entail additional effort for the study team initially, it ensures that participant eligibility is determined by the investigator.</p> <hd id="AN0185986534-11">Use captcha and IP address tracking</hd> <p>Most data collection systems are equipped with built-in security features like CAPTCHA (acronym for Completely Automated Public Turing Test to Tell Computers and Humans Apart) and IP address tracking (Brazhkin, [<reflink idref="bib3" id="ref22">3</reflink>]; Waggoner et al., [<reflink idref="bib28" id="ref23">28</reflink>]). Bot software is often linked to a single IP address often based in low to middle income countries (Asare-Marfo et al., [<reflink idref="bib1" id="ref24">1</reflink>]; Pozzar et al., [<reflink idref="bib19" id="ref25">19</reflink>]; Singh, [<reflink idref="bib23" id="ref26">23</reflink>]). Consequently, multiple responses from the same IP address or from IPs associated with VPN servers, hosting sites, or data centers should be flagged for automatic review (Singh, [<reflink idref="bib23" id="ref27">23</reflink>]). CAPTCHA serves as a challenge-response mechanism to ascertain whether the user is human, thus deterring bot attacks (Gutub & Kheshaifaty, [<reflink idref="bib7" id="ref28">7</reflink>]; Hidalgo & Alvarez, [<reflink idref="bib10" id="ref29">10</reflink>]).</p> <hd id="AN0185986534-12">Conduct thorough eligibility interviews</hd> <p>Given the persistence of fraudulent participants and sophisticated capabilities of advanced bot software, it is crucial to acknowledge that traditional measures such as CAPTCHA and IP address verification may not suffice as foolproof safeguards. To further strengthen the screening process, we propose that researchers consider virtual eligibility interviews conducted via video conferencing. In large studies, this may be a random selection of 10% of those who are eligible to assure data quality. While we recognize the increased burden virtual eligibility interviews place on both researchers and research participants, health sciences researchers must prioritize the collection of valid and reliable data. This becomes particularly important when developing interventions, based on these data, that aim to improve health outcomes among minority groups disproportionately affected by health inequities.</p> <p>Even after screening surveys are reviewed for eligibility, the absence of verification methods (e.g. face-to-face interviews) leaves room for eligible participants to potentially exploit the system by enrolling multiple times using distinct email addresses. While this adds to research costs, our experience has shown that some individuals who do not meet screening eligibility criteria are still willing to engage in video screening. This underscores the importance of employing multi-faceted verification methods to ensure the integrity of the participant pool.</p> <hd id="AN0185986534-13">Conduct quality checks</hd> <p>In addition to the above-mentioned measures, there are additional steps that can be taken to prevent fraud during data collection. One effective strategy is to assess the time it takes for participants to complete each survey included in the study. If a survey is designed to take, for example, 20 minutes to complete, but is finished in 5 minutes, these submissions should be flagged for review (Brazhkin, [<reflink idref="bib3" id="ref30">3</reflink>]; Callegaro et al., [<reflink idref="bib4" id="ref31">4</reflink>]). Furthermore, we recommend conducting checks for data consistency such as performing cross-field validation (Little et al., [<reflink idref="bib14" id="ref32">14</reflink>]). An illustrative example from our study revealed instances where fraudulent participants provided contradictory responses between surveys including demographic details, such as age and the date of birth of their child. Data inconsistencies should trigger a careful examination of the data for potentially fraudulent activity.</p> <hd id="AN0185986534-14">Use variety of survey question types to assess concordance</hd> <p>Another approach is to ask several variations of the same question to assess concordance. In our surveys we asked participants about the trimester of pregnancy, weeks of gestation, and number of months pregnant in different sections of the surveys. In one section, we required a text response; in another, a categorical response to a range of months; and in a third, we utilized a simple radio button. In our study data from both real and fraudulent participants appeared similar between groups (Table 1); this initially made it difficult to distinguish whether responses to online surveys are from humans or bots (Bowen et al., [<reflink idref="bib2" id="ref33">2</reflink>]). However, upon closer examination of a free-text question, 'At which hospital did you deliver your baby?' a few differences emerged between the responses of real and fraudulent participants (Table 1). Notably, 14.3% of the fraudulent sample provided names of hospitals that do not exist in the United States. An even larger percentage (18.7%) listed facilities that weren't hospitals (such as OBGYN offices or those lacking labor and delivery units). Of those who stated that they delivered in the Northeast, the majority delivered in one hospital in New York City. This difference indicates that fraudulent participants likely resorted to searching for hospital names online to answer this specific question, unlike birthing people who typically know the name of the hospital where they delivered.</p> <p>Table 1. Demographic details for study participants and fraudulent participants.</p> <p> <ephtml> <table><thead><tr><td /><td>Study participants</td><td>Fraudulent participants</td></tr><tr><td>N (Percent)</td><td><italic>N</italic> = 88 (100%)</td><td><italic>N</italic> = 91 (100%)</td></tr><tr><td>Variable</td><td>n (column percent)</td></tr></thead><tbody><tr><td><bold>Severe Maternal Mortality</bold></td></tr><tr><td>Yes</td><td>24 (27.3%)</td><td>36 (39.6%)</td></tr><tr><td>No</td><td>64 (72.7%)</td><td>55 (60.4%)</td></tr><tr><td><bold>Age</bold></td></tr><tr><td>Average maternal age (<35 years)</td><td>63 (71.6%)</td><td>85 (93.4%)</td></tr><tr><td>Advance maternal age (≥35 years)</td><td>25 (28.4%)</td><td>6 (6.6%)</td></tr><tr><td><bold>Income</bold></td></tr><tr><td>≥$50,000</td><td>52 (59.1%)</td><td>79 (86.8%)</td></tr><tr><td><$50,000</td><td>36 (40.9%)</td><td>12 (13.9%)</td></tr><tr><td><bold>Education</bold></td></tr><tr><td>College degree</td><td>62 (70.5%)</td><td>75 (82.4%)</td></tr><tr><td>No college degree</td><td>26 (29.5%)</td><td>16 (17.6%)</td></tr><tr><td><bold>Insurance</bold></td></tr><tr><td>Private</td><td>52 (59.1%)</td><td>42 (46.2%)</td></tr><tr><td>Public</td><td>36 (40.9%)</td><td>49 (53.8%)</td></tr><tr><td><bold>Birth Mode</bold></td></tr><tr><td>Vaginal or scheduled cesarean section</td><td>67 (76.1%)</td><td>77 (84.6%)</td></tr><tr><td>Emergency cesarean section</td><td>21 (23.9%)</td><td>14 (15.4%)</td></tr><tr><td><bold>Self-reported pre-pregnancy health</bold></td></tr><tr><td>Good prenatal health</td><td>82 (93.2%)</td><td>88 (96.7%)</td></tr><tr><td>Poor prenatal health</td><td>6 (6.8%)</td><td>3 (3.3%)</td></tr><tr><td><bold>In which trimester did you start prenatal care?</bold></td></tr><tr><td>Within first trimester</td><td>70 (79.5%)</td><td>66 (72.5%)</td></tr><tr><td>After first trimester</td><td>18 (20.5%</td><td>25 (27.5%)</td></tr><tr><td><bold>Did you deliver a healthy baby?</bold></td></tr><tr><td>Yes</td><td>80 (90.9%)</td><td>87 (95.6%)</td></tr><tr><td>No</td><td>8 (9.1%)</td><td>4 (4.4%)</td></tr><tr><td><bold>Level of chronic discrimination</bold></td></tr><tr><td>High Low</td><td>57 (64.8%) 31 (35.2%)</td><td>66 (72.5%) 24 (26.4%)</td></tr><tr><td><bold>Delivery hospital region</bold></td></tr><tr><td>Midwest</td><td>14 (15.9%)</td><td>3 (3.3%)</td></tr><tr><td>Northeast</td><td>26 (29.5%)</td><td>26 (28.6%)*</td></tr><tr><td>Southeast</td><td>34 (38.6%)</td><td>9 (9.9%)</td></tr><tr><td>Southwest</td><td>7 (8.0%)</td><td>10 (11.0%)</td></tr><tr><td>West</td><td>7 (8.0%)</td><td>6 (6.6%)</td></tr><tr><td>Refused to answer</td><td>N/A</td><td>7 (7.7%)</td></tr><tr><td>Hospital named does not exist</td><td>N/A</td><td>13 (14.3%)</td></tr><tr><td>Hospital named does not have labor services</td><td>N/A</td><td>17 (18.7%)</td></tr></tbody></table> </ephtml> </p> <p>1 *81% of participants giving birth in the Northeast named one single well-known hospital in New York City</p> <p>Additionally, we included an open-ended text response question asking participants to share details about their birth experiences and paid special attention to the details provided or omitted. For example, some participants gave a simple answer such as 'It was good.' or 'It was hard.' While others gave more complex answers, but that still lacked depth and details that would be expected when recalling a birth story (Takehara et al., [<reflink idref="bib24" id="ref34">24</reflink>]; Waldenström, [<reflink idref="bib29" id="ref35">29</reflink>]). These unexpected responses prompted us to take a second look at the data provided. While this measure can help reduce fraudulent participants, it is important to note that given the recent widespread use of AI software such as ChatGPT, this method is not foolproof. AI can generate plausible, yet fictitious, responses based on simple prompts (Table 2). Therefore, exercise caution when relying solely on open-ended test questions for fraud detection.</p> <p>Table 2. Example of prompts and ai-generated responses using chat-GPT-3.5.</p> <p> <ephtml> <table><thead><tr><td>Prompt</td><td>ChatGPT Response</td></tr></thead><tbody><tr><td>Prompt: "Describe in 50 words an adverse perinatal outcome in first person"</td><td><italic>I experienced an adverse perinatal outcome when I gave birth. It was a harrowing experience as my baby faced complications during delivery, causing distress and requiring immediate medical intervention. This challenging journey was filled with anxiety and uncertainty, but the resilience of both my baby and me ultimately led to a positive outcome.</italic></td></tr><tr><td>Prompt: "Tell me in 100 words about a birthing experience that turned out bad because of an experienceof discrimination from a nurse during labor"</td><td><italic>My birthing experience turned nightmarish due to discrimination from a nurse during labor. Despite my vulnerability, she treated me callously, dismissing my concerns and pain. Her negligence delayed critical medical interventions, worsening the situation. The emotional distress and humiliation intensified my pain. Eventually, another compassionate nurse intervened, highlighting the stark contrast. The incident scarred what should have been a joyous moment. It highlighted the urgent need for empathy and sensitivity in healthcare, reminding me of the systemic issues faced by marginalized individuals. Despite this ordeal, my resilience prevailed, and with the support of other medical staff, I emerged stronger, albeit deeply affected.</italic></td></tr><tr><td>Prompt: "Tell me in 50 words about a stillbirth you experienced as a Black woman"</td><td><italic>As a Black woman, the pain of stillbirth cut deep, echoing generations of loss. My heartache was woven with the historic trauma of my community. The unspoken burden of racial disparities in healthcare intensified the grief. Despite this, my resilience grew, a testament to the strength that runs through generations, even in the face of profound sorrow.</italic></td></tr></tbody></table> </ephtml> </p> <hd id="AN0185986534-15">Collect and verify contact information</hd> <p>Originally, our aim in collecting participant contact information was solely to facilitate the scheduling of qualitative interviews once their survey data had been reviewed. However, after we realized that our study may be infiltrated by bots and fraudulent participants, this extra step yielded additional advantages. First, it enabled us to assess the authenticity of provided phone numbers. The overwhelming majority of our potential participants either provided a disconnected or Google Voice phone number. It also provided a means to cross-verify any discrepancies in the data for participants that provided real phone numbers. Collecting phone numbers also allows researchers to confirm the target geographic regions are represented in the study, with the understanding that some individuals may possess cell phones with area codes that are different from their area of residence. Therefore, we highly recommend that researchers collect and verify the authenticity of provided phone numbers.</p> <p>There are several methods researchers can use to verify whether a telephone number is a dedicated wireless, landline, or computer-generated source. Google Voice requires callers to verify their identity before connecting, allowing the recipient to identify the caller, a feature not found on standard landline or wireless phones. Additionally, online phone carrier lookup services can be used. These services identify the carrier associated with a given number. For Google Voice and computer-generated numbers, 'Bandwidth.com' is typically listed, whereas dedicated wireless and landline numbers display their respective carriers. However, these verification methods have limitations. Google Voice users can disable caller verification, and while Google Voice numbers often appear under 'Bandwidth.com,' not all Google Voice numbers are listed. This step is crucial as bots and deceptive respondents often use these types of numbers to avoid detection (Teitcher et al., [<reflink idref="bib25" id="ref36">25</reflink>]). For larger-scale studies, researchers can conduct quality checks (e.g. calling a randomly selected 10% of provided phone numbers).</p> <hd id="AN0185986534-16">Create comprehensive IRB protocol for removing low quality data</hd> <p>The most evident way to increase validity of study findings is to ensure that an IRB-approved protocol incorporates strategies for identifying, addressing, and excluding low-quality data (Miner et al., [<reflink idref="bib15" id="ref37">15</reflink>]; Pozzar et al., [<reflink idref="bib19" id="ref38">19</reflink>]). To minimize the possibility of incorrectly excluding real participants by misclassifying them as fraudulent participants, we suggest exclusion should not be based on a single failed prevention measure. Researchers should determine the threshold for the number of prevention measures that must be failed before considering participant removal. These criteria should be set during the development of recruitment strategies (Xu et al., [<reflink idref="bib31" id="ref39">31</reflink>]). Although our IRB mandates the use of a comprehensive protocol template, it didn't include a section dedicated to fraud prevention measures and subsequent steps to address any resulting fraud. Following an incident where fraudulent participants contacted our IRB demanding compensation for their deceptive data, our study team compiled a set of best practice guidelines (all of which are outlined in this paper) to aid our IRB in integrating such sections into future protocol templates.</p> <hd id="AN0185986534-17">Conclusion</hd> <p>Social media is a recruitment tool that offers significant advantages for researchers, particularly when targeting hard-to-reach populations and those traditionally underrepresented in research studies. However, these methods are not immune to the infiltration of bots and fraudulent respondents seeking financial incentives, even when basic built-in protections are in place. This presents a challenge to data integrity and can have a disproportionate impact on research involving minority populations. Based on our experiences with online recruitment, we have developed recommendations using the Swiss Cheese Model of Study Fraud Prevention for researchers to consider. The components of the proposed model include the adoption of proactive measures in the early design stages of recruitment and data collection, as well as during the data cleaning phase. Because no single fraud prevention method is entirely foolproof, and effectiveness may vary depending on the nature of the research study, it is advisable to employ a combination of strategies. This should include a careful assessment of the advantages and disadvantages of each method as well as an assessment of how they may potentially interact to create opportunities for fraudulent activity. While this model does not pretend to be a standard technique for social research, implementing these fraud prevention strategies can strengthen the reliability and validity of research findings.</p> <hd id="AN0185986534-18">Disclosure statement</hd> <p>No potential conflict of interest was reported by the author(s).</p> <ref id="AN0185986534-19"> <title> References </title> <blist> <bibl id="bib1" idref="ref24" type="bt">1</bibl> <bibtext> Asare-Marfo, D, Hatley, N., Lau, A., Mercer, A., Keeter, S., Ferno, J. (2020, February 18). Assessing the risks to online polls from bogus respondents. Pew Research Center Methods. https://<ulink href="http://www.pewresearch.org/methods/2020/02/18/assessing-the-risks-to-online-polls-from-bogus-respondents/">www.pewresearch.org/methods/2020/02/18/assessing-the-risks-to-online-polls-from-bogus-respondents/</ulink></bibtext> </blist> <blist> <bibl id="bib2" idref="ref33" type="bt">2</bibl> <bibtext> Bowen, A. M., Daniel, C. M., Williams, M. L., & Baird, G. L. (2008). Identifying multiple submissions in internet research: Preserving data integrity. AIDS and Behavior, 12 (6), 964 – 973. https://doi.org/10.1007/s10461-007-9352-2</bibtext> </blist> <blist> <bibl id="bib3" idref="ref22" type="bt">3</bibl> <bibtext> Brazhkin, V. (2020). "I have just returned from the moon:" online survey fraud. Supply Chain Management, 25 (4), 489 – 503. https://doi.org/10.1108/SCM-12-2019-0466</bibtext> </blist> <blist> <bibl id="bib4" idref="ref31" type="bt">4</bibl> <bibtext> Callegaro, M., Manfreda, K. L., & Vehovar, V. (2015). Web survey methodology. SAGE Publications Ltd. https://doi.org/10.4135/9781529799651</bibtext> </blist> <blist> <bibl id="bib5" idref="ref1" type="bt">5</bibl> <bibtext> Couper, M. P. (2017). New developments in survey data collection. Annual Review of Sociology, 43 (1), 121 – 145. https://doi.org/10.1146/annurev-soc-060116-053613</bibtext> </blist> <blist> <bibl id="bib6" idref="ref3" type="bt">6</bibl> <bibtext> de Leeuw, E. D. (2012). Counting and measuring online: The quality of internet surveys*. Bulletin of Sociological Methodology/Bulletin de Méthodologie Sociologique, 114 (1), 68 – 78. https://doi.org/10.1177/0759106312437290</bibtext> </blist> <blist> <bibl id="bib7" idref="ref28" type="bt">7</bibl> <bibtext> Gutub, A., & Kheshaifaty, N. (2023). Practicality analysis of utilizing text-based CAPTCHA vs. Graphic-based CAPTCHA authentication. Multimedia Tools & Applications, 82 (30), 46577 – 46609. https://doi.org/10.1007/s11042-023-15586-5</bibtext> </blist> <blist> <bibl id="bib8" idref="ref12" type="bt">8</bibl> <bibtext> Harris, P. A., Taylor, R., Minor, B. L., Elliott, V., Fernandez, M., O'Neal, L., McLeod, L., Delacqua, G., Delacqua, F., Kirby, J., Duda, S. N., & REDCap Consortium. (2019). The REDCap consortium: Building an international community of software platform partners. Journal of Biomedical Informatics, 95, 103208. https://doi.org/10.1016/j.jbi.2019.103208</bibtext> </blist> <blist> <bibl id="bib9" idref="ref11" type="bt">9</bibl> <bibtext> Harris, P. A., Taylor, R., Thielke, R., Payne, J., Gonzalez, N., & Conde, J. G. (2009). Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics, 42 (2), 377 – 381. https://doi.org/10.1016/j.jbi.2008.08.010</bibtext> </blist> <blist> <bibtext> Hidalgo, J. M. G., & Alvarez, G. (2011). Chapter 3 - CAPTCHAs: An Artificial Intelligence Application to Web Security. Advances in Computers, 83, 109 – 181. https://doi.org/10.1016/B978-0-12-385510-7.00003-5</bibtext> </blist> <blist> <bibtext> Hsieh, G., & Kocielnik, R. (2016). You get who you pay for: The impact of incentives on participation bias. Proceedings of the 19th ACM Conference on Computer-Supported Cooperative Work & Social Computing (pp. 823 – 835). https://doi.org/10.1145/2818048.2819936</bibtext> </blist> <blist> <bibtext> Hughes, T. B., Varma, V. R., Pettigrew, C., & Albert, M. S. (2017). African Americans and clinical research: Evidence concerning barriers and facilitators to participation and recruitment recommendations. The Gerontologist, 57 (2), 348 – 358. https://doi.org/10.1093/geront/gnv118</bibtext> </blist> <blist> <bibtext> Langer, S. L., Castro, F. G., Chen, A.-C.-C., Davis, K. C., Joseph, R. P., Kim, W., SUNNY, Larkey, L., Lee, R. E., Petrov, M. E., Reifsnider, E., Youngstedt, S. D., & Shaibi, G. Q. (2021). Recruitment and retention of underrepresented and vulnerable populations to research. Public Health Nursing, 38 (6), 1102 – 1115. https://doi.org/10.1111/phn.12943</bibtext> </blist> <blist> <bibtext> Little, M. A., Varoquaux, G., Saeb, S., Lonini, L., Jayaraman, A., Mohr, D. C., & Kording, K. P. (2017). Using and understanding cross-validation strategies. Perspectives on saeb et al. GigaScience, 6 (5), 1 – 6. https://doi.org/10.1093/gigascience/gix020</bibtext> </blist> <blist> <bibtext> Miner, M. H., Bockting, W. O., Romine, R. S., & Raman, S. (2012). Conducting internet research with the transgender population: Reaching broad samples and collecting valid data. Social Science Computer Review, 30 (2), 202 – 211. https://doi.org/10.1177/0894439311404795</bibtext> </blist> <blist> <bibtext> Nadeem, R. (2022, April 7). 2. Black Americans' trust in medical scientists and views about the potential for researcher misconduct. Pew Research Center Science & Society. https://<ulink href="http://www.pewresearch.org/science/2022/04/07/black-americans-trust-in-medical-scientists-and-views-about-the-potential-for-researcher-misconduct/">www.pewresearch.org/science/2022/04/07/black-americans-trust-in-medical-scientists-and-views-about-the-potential-for-researcher-misconduct/</ulink></bibtext> </blist> <blist> <bibtext> National Academies of Sciences, E. Affairs, P. and G. Committee on Women in Science, E. Research, C. on I. the R. of W. and U. M. in C. T. and, Bibbins-Domingo, K. & Helman, A. (2022). Barriers to representation of underrepresented and excluded populations in clinical research. In Improving representation in clinical trials and research: Building research equity for women and underrepresented groups. National Academies Press (US). https://<ulink href="http://www.ncbi.nlm.nih.gov/books/NBK584407/">www.ncbi.nlm.nih.gov/books/NBK584407/</ulink></bibtext> </blist> <blist> <bibtext> Peek, M. E., Nunez-Smith, M., Drum, M., & Lewis, T. T. (2011). Adapting the everyday discrimination scale to medical settings: Reliability and validity testing in a sample of African American patients. Ethnicity & Disease, 21 (4), 502 – 509.</bibtext> </blist> <blist> <bibtext> Pozzar, R., Hammer, M. J., Underhill-Blazey, M., Wright, A. A., Tulsky, J. A., Hong, F., Gundersen, D. A., & Berry, D. L. (2020). Threats of bots and other bad actors to data quality following research participant recruitment through social media: Cross-sectional questionnaire. Journal of Medical Internet Research, 22 (10), e23021. https://doi.org/10.2196/23021</bibtext> </blist> <blist> <bibtext> Reason, J. (2000). Human error: Models and management. BMJ, 320 (7237), 768 – 770. https://doi.org/10.1136/bmj.320.7237.768</bibtext> </blist> <blist> <bibtext> ResearchMatch. (n.d.). Retrieved March 16, 2024, from https://<ulink href="http://www.researchmatch.org/">www.researchmatch.org/</ulink></bibtext> </blist> <blist> <bibtext> Scharff, D. P., Mathews, K. J., Jackson, P., Hoffsuemmer, J., Martin, E., & Edwards, D. (2010). More than Tuskegee: Understanding mistrust about research participation. Journal of Health Care for the Poor and Underserved, 21 (3), 879 – 897. https://doi.org/10.1353/hpu.0.0323</bibtext> </blist> <blist> <bibtext> Singh, Singh V.P, Pal P. (n.d.). Survey of different types of CAPTCHA - International. Retrieved October 10, 2023, from https://<ulink href="http://www.academia.edu/7053243/Survey%5fof%5fDifferent%5fTypes%5fof%5fCAPTCHA%5finternational">www.academia.edu/7053243/Survey%5fof%5fDifferent%5fTypes%5fof%5fCAPTCHA%5finternational</ulink></bibtext> </blist> <blist> <bibtext> Takehara, K., Noguchi, M., Shimane, T., & Misago, C. (2014). A longitudinal study of women's memories of their childbirth experiences at five years postpartum. BMC Pregnancy and Childbirth, 14 (1), 221. https://doi.org/10.1186/1471-2393-14-221</bibtext> </blist> <blist> <bibtext> Teitcher, J. E. F., Bockting, W. O., Bauermeister, J. A., Hoefer, C. J., Miner, M. H., & Klitzman, R. L. (2015). Detecting, preventing, and responding to "fraudsters" in internet research: Ethics and tradeoffs. The Journal of Law, Medicine & Ethics: A Journal of the American Society of Law, Medicine & Ethics, 43 (1), 116 – 133. https://doi.org/10.1111/jlme.12200</bibtext> </blist> <blist> <bibtext> Toepoel, V. (2016). Doing surveys online. SAGE Publications Ltd. https://doi.org/10.4135/9781473967243</bibtext> </blist> <blist> <bibtext> Vedam, S., Stoll, K., Rubashkin, N., Martin, K., Miller-Vedam, Z., Hayes-Klein, H., & Jolicoeur, G. (2017). The mothers on respect (MOR) index: Measuring quality, safety, and human rights in childbirth. SSM - Population Health, 3, 201 – 210. https://doi.org/10.1016/j.ssmph.2017.01.005</bibtext> </blist> <blist> <bibtext> Waggoner, P. D., Kennedy, R., & Clifford, S. (2019). Detecting fraud in online surveys by tracing, scoring, and visualizing IP addresses. Journal of Open Source Software, 4 (37), 1285. https://doi.org/10.21105/joss.01285</bibtext> </blist> <blist> <bibtext> Waldenström, U. (2003). Women's memory of childbirth at two months and one year after the birth. Birth (Berkeley, Calif), 30 (4), 248 – 254. https://doi.org/10.1046/j.1523-536x.2003.00254.x</bibtext> </blist> <blist> <bibtext> Whitaker, C., Stevelink, S, and Fear, N. (2017). The Use of Facebook in Recruiting Participants for Health Research Purposes: A Systematic Review. Journal of Medical Internet Research, 19 (8), e290. https://doi.org/10.2196/jmir.7071</bibtext> </blist> <blist> <bibtext> Xu, Y., Pace, S., Kim, J., Iachini, A., King, L. B., Harrison, T., DeHart, D., Levkoff, S. E., Browne, T. A., Lewis, A. A., Kunz, G. M., Reitmeier, M., Utter, R. K., & Simone, M. (2022). Threats to online surveys: Recognizing, detecting, and preventing survey bots. Social Work Research, 46 (4), 343 – 350. https://doi.org/10.1093/swr/svac023</bibtext> </blist> </ref> <aug> <p>By Roseline Jean Louis and Lisa M. Thompson</p> <p>Reported by Author; Author</p> <p></p> <p>Roseline Jean Louis is a Ph.D. Candidate at Emory University School of Nursing in the United States (US). Her research uses a multi-methods approach to examine the impacts of racial discrimination and disrespectful maternity care on maternal morbidity among the Black Birthing population in the US. She is an Edward A. Bouchet Graduate Honor Society Scholar. She aims to develop and implement potential strategies to improve maternal health among marginalized populations.</p> <p>Lisa M. Thompson is a tenured Professor at Emory University. The focus of her research is to evaluate effective behavioral interventions, using implementation science methods to reduce environmental exposures that can be implemented in real-world settings. She is the principal investigator of a community-level cluster randomized trial in rural Guatemala that is the first implementation science study to develop and evaluate community-level interventions that aim to reduce household burning of plastic waste. This study received approval from Emory University's Institutional Review Board (IRB STUDY00004513).</p> </aug> <nolink nlid="nl1" bibid="bib26" firstref="ref2"></nolink> <nolink nlid="nl2" bibid="bib28" firstref="ref4"></nolink> <nolink nlid="nl3" bibid="bib31" firstref="ref5"></nolink> <nolink nlid="nl4" bibid="bib20" firstref="ref6"></nolink> <nolink nlid="nl5" bibid="bib18" firstref="ref7"></nolink> <nolink nlid="nl6" bibid="bib27" firstref="ref8"></nolink> <nolink nlid="nl7" bibid="bib12" firstref="ref9"></nolink> <nolink nlid="nl8" bibid="bib11" firstref="ref10"></nolink> <nolink nlid="nl9" bibid="bib21" firstref="ref13"></nolink> <nolink nlid="nl10" bibid="bib30" firstref="ref15"></nolink> <nolink nlid="nl11" bibid="bib13" firstref="ref18"></nolink> <nolink nlid="nl12" bibid="bib16" firstref="ref19"></nolink> <nolink nlid="nl13" bibid="bib17" firstref="ref20"></nolink> <nolink nlid="nl14" bibid="bib22" firstref="ref21"></nolink> <nolink nlid="nl15" bibid="bib19" firstref="ref25"></nolink> <nolink nlid="nl16" bibid="bib23" firstref="ref26"></nolink> <nolink nlid="nl17" bibid="bib10" firstref="ref29"></nolink> <nolink nlid="nl18" bibid="bib14" firstref="ref32"></nolink> <nolink nlid="nl19" bibid="bib24" firstref="ref34"></nolink> <nolink nlid="nl20" bibid="bib29" firstref="ref35"></nolink> <nolink nlid="nl21" bibid="bib25" firstref="ref36"></nolink> <nolink nlid="nl22" bibid="bib15" firstref="ref37"></nolink>
Header DbId: eric
DbLabel: ERIC
An: EJ1495994
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Bots and Fake Participants: Ensuring Valid and Reliable Data Collection Using Online Participant Recruitment Methods
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Roseline+Jean+Louis%22">Roseline Jean Louis</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0002-6483-7024">0000-0002-6483-7024</externalLink>)<br /><searchLink fieldCode="AR" term="%22Lisa+M%2E+Thompson%22">Lisa M. Thompson</searchLink> (ORCID <externalLink term="https://orcid.org/0000-0002-8001-2057">0000-0002-8001-2057</externalLink>)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22International+Journal+of+Social+Research+Methodology%22"><i>International Journal of Social Research Methodology</i></searchLink>. 2025 28(4):463-473.
– Name: Avail
  Label: Availability
  Group: Avail
  Data: Routledge. Available from: Taylor & Francis, Ltd. 530 Walnut Street Suite 850, Philadelphia, PA 19106. Tel: 800-354-1420; Tel: 215-625-8900; Fax: 215-207-0050; Web site: http://www.tandf.co.uk/journals
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 11
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2025
– Name: SourceSuprt
  Label: Sponsoring Agency
  Group: SrcSuprt
  Data: National Institute of Nursing Research (NINR) (DHHS/NIH)
– Name: NumberContract
  Label: Contract Number
  Group: NumCntrct
  Data: F31NR020575
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Journal Articles<br />Reports - Research
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Research+Problems%22">Research Problems</searchLink><br /><searchLink fieldCode="DE" term="%22Recruitment%22">Recruitment</searchLink><br /><searchLink fieldCode="DE" term="%22Selection+Tools%22">Selection Tools</searchLink><br /><searchLink fieldCode="DE" term="%22Internet%22">Internet</searchLink><br /><searchLink fieldCode="DE" term="%22Social+Media%22">Social Media</searchLink><br /><searchLink fieldCode="DE" term="%22Online+Surveys%22">Online Surveys</searchLink><br /><searchLink fieldCode="DE" term="%22Artificial+Intelligence%22">Artificial Intelligence</searchLink><br /><searchLink fieldCode="DE" term="%22Deception%22">Deception</searchLink><br /><searchLink fieldCode="DE" term="%22Credibility%22">Credibility</searchLink><br /><searchLink fieldCode="DE" term="%22Research+Administration%22">Research Administration</searchLink><br /><searchLink fieldCode="DE" term="%22Research%22">Research</searchLink><br /><searchLink fieldCode="DE" term="%22Research+Projects%22">Research Projects</searchLink><br /><searchLink fieldCode="DE" term="%22Medical+Research%22">Medical Research</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1080/13645579.2024.2410176
– Name: ISSN
  Label: ISSN
  Group: ISSN
  Data: 1364-5579<br />1464-5300
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Recruitment for successful health sciences research requires balancing efficiency, cost, accessibility, and reliability of available recruitment methods. Our case-control study used online recruitment methods, which broadened our reach to potential participants across the United States. However, this approach also exposed us to challenges associated with bot interference and fraudulent participation. This paper focuses on maintaining data integrity, specifically when utilizing online participant recruitment methods. Drawing from our experience, we propose The Swiss Cheese Model of Study Participant Fraud Prevention, adapted from Reason's Swiss Cheese Model, and illustrate 10 prevention and verification measures that can be taken to minimize fraud in research studies that rely on online recruitment. We emphasize the importance of a layered approach, including carefully designed recruitment media and compensation protocols, vetting of participant eligibility, and data verification protocols to ensure the validity and reliability of research findings in the digital age.
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 2026
– Name: AN
  Label: Accession Number
  Group: ID
  Data: EJ1495994
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=EJ1495994
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1080/13645579.2024.2410176
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 11
        StartPage: 463
    Subjects:
      – SubjectFull: Research Problems
        Type: general
      – SubjectFull: Recruitment
        Type: general
      – SubjectFull: Selection Tools
        Type: general
      – SubjectFull: Internet
        Type: general
      – SubjectFull: Social Media
        Type: general
      – SubjectFull: Online Surveys
        Type: general
      – SubjectFull: Artificial Intelligence
        Type: general
      – SubjectFull: Deception
        Type: general
      – SubjectFull: Credibility
        Type: general
      – SubjectFull: Research Administration
        Type: general
      – SubjectFull: Research
        Type: general
      – SubjectFull: Research Projects
        Type: general
      – SubjectFull: Medical Research
        Type: general
    Titles:
      – TitleFull: Bots and Fake Participants: Ensuring Valid and Reliable Data Collection Using Online Participant Recruitment Methods
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Roseline Jean Louis
      – PersonEntity:
          Name:
            NameFull: Lisa M. Thompson
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-print
              Value: 1364-5579
            – Type: issn-electronic
              Value: 1464-5300
          Numbering:
            – Type: volume
              Value: 28
            – Type: issue
              Value: 4
          Titles:
            – TitleFull: International Journal of Social Research Methodology
              Type: main
ResultId 1