EVERYTHING IS PERFECT, AND WE HAVE NO PROBLEMS: DETECTING AND LIMITING SOCIAL DESIRABILITY BIAS IN QUALITATIVE RESEARCH Bergen, N.;Labonté, R.; ; © 2021, BERGEN, N. This work is licensed under the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/legalcode), which permits unrestricted use, distribution, and reproduction, provided the original work is properly credited. IDRC Grant: 108028-002-Promoting Safe Motherhood in Jimma Zone, Ethiopia (IMCHA) https://creativecommons.org/licenses/by/4.0/legalcode https://doi.org/10.1177/1049732319889354 Qualitative Health Research 2020, Vol. 30(5) 783 –792 © The Author(s) 2019 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/1049732319889354 journals.sagepub.com/home/qhr Methods Introduction Social desirability bias refers to the tendency to present oneself and one’s social context in a way that is per- ceived to be socially acceptable, but not wholly reflec- tive of one’s reality. In research, the bias denotes a mismatch between participants’ genuine construction of reality and the presentation of that reality to research- ers. Researchers have noted a distinction between social desirability responding as a personality attribute (inflated, but honestly held, positive self-perception) and social desirability responding as a response style (a tendency to do impression management, or alter how others perceive one’s reality) (Pauls & Stemmler, 2003), cautioning that the measurement or correction of social desirability bias in research should focus on the latter (Paulhus, 1984, 2001). Early explorations of the construct of social desir- ability bias derive from psychology, attending to matters related to response tendencies and intentionality. Based on the use of standardized surveys, social desirability bias is considered to stem from the sensitive nature of an answer, and is distinct from sensitivities associated with the intrusiveness of a question or the respondent’s per- ceived risk of disclosure (Krumpal, 2013). The rationale for answering a question in a particular way has links to rational choice theory—that is, to maximize utility and goal achievement—and the subjective expected utility theory—considering the risks, losses, and outcomes associated with the decision to admit to a sensitive behavior or not (Krumpal, 2013). Researchers under- score the distinction between response styles (consistent across time) and response sets (short-lived responses that may be influenced by situational factors such as the presence of an interviewer) (D. N. Jackson & Messick, 1958; Paulhus, 2001). In research, social desirability tendencies tend to emerge in characteristic ways. They are more common in research on issues that participants find sensitive or con- troversial, and in situations where there are widely accepted attitudes, behaviors, or norms (Grimm, 2010). Social desirability bias has been found to present differ- ently depending on the affluence of the country (Johnson 889354QHRXXX10.1177/1049732319889354Qualitative Health ResearchBergen and Labonté research-article2019 1University of Ottawa, Ottawa, Ontario, Canada Corresponding Author: Nicole Bergen, University of Ottawa, 75 Laurier Ave East, Ottawa, Ontario, Canada K1N 6N5. Email: nicolejbergen@gmail.com “Everything Is Perfect, and We Have No Problems”: Detecting and Limiting Social Desirability Bias in Qualitative Research Nicole Bergen1 and Ronald Labonté1 Abstract Many qualitative research studies acknowledge the possibility of social desirability bias (a tendency to present reality to align with what is perceived to be socially acceptable) as a limitation that creates complexities in interpreting findings. Drawing on experiences conducting interviews and focus groups in rural Ethiopia, this article provides an empirical account of how one research team developed and employed strategies to detect and limit social desirability bias. Data collectors identified common cues for social desirability tendencies, relating to the nature of the responses given and word choice patterns. Strategies to avoid or limit bias included techniques for introducing the study, establishing rapport, and asking questions. Pre-fieldwork training with data collectors, regular debriefing sessions, and research team meetings provided opportunities to discuss social desirability tendencies and refine approaches to account for them throughout the research. Although social desirability bias in qualitative research may be intractable, it can be minimized. Keywords qualitative research; social desirability bias; Ethiopia; methodology https://us.sagepub.com/en-us/journals-permissions https://journals.sagepub.com/home/qhr mailto:nicolejbergen@gmail.com http://crossmark.crossref.org/dialog/?doi=10.1177%2F1049732319889354&domain=pdf&date_stamp=2019-12-13 784 Qualitative Health Research 30(5) & Van de Vijver, 2002), prevailing cultural norms (Johnson & Van de Vijver, 2002) such as the collectivist versus individualist nature of the society (Lalwani et al., 2006), and individual characteristics such as social posi- tion (Johnson & Van de Vijver, 2002), gender (Kelly et al., 2013), and personality traits (Johnson & Van de Vijver, 2002). Social desirability bias is problematic because it can lead to overestimation of the positive and diminished het- erogeneity in responses, resulting in a questionable appearance of consensus. Researchers have suggested different approaches to minimize or account for the bias. Some advocate the collection of data through self-admin- istered questionnaires, positing that the absence of another person removes the normative pressure to answer in a certain manner (Grimm, 2010; Johnson & Van de Vijver, 2002). Participant observation techniques have been recommended to triangulate other forms of data col- lection (Freeth et al., 2012; Harvey, 2018), while face-to- face interviews permit trained interviewers to take measures to probe and clarify participant responses (Kelly et al., 2013; Mooney et al., 2018). This article pertains to social desirability bias in a qualitative research study, providing an empirical account of how one research team developed and employed strat- egies to detect and limit social desirability bias. We draw from our own research study about maternal, neonatal and child health (MNCH) in Ethiopia, where we con- ducted in-depth interviews and focus group discussions with community stakeholders to learn more about their attitudes, perceptions, and experiences related to MNCH, including barriers and enablers of health service use (Bergen, 2018). In Ethiopia, a highly collectivist society, the prevailing cultural, social, and political environments encourage high conformity and compliance to institu- tional norms and values (Maes, 2017; Østebø et al., 2018). Previous qualitative studies about health services in Ethiopia have acknowledged the possibility of social desirability bias as a research limitation (Girma et al., 2011; R. Jackson et al., 2017; King et al., 2016), yet few elaborate on steps taken to detect and limit the bias. In this article, we describe how social desirability bias emerged as a consideration during the data collection stage of our study, and how researchers worked alongside a team of data collectors to characterize social desirabil- ity tendencies in participants, and then develop/imple- ment approaches to minimize the bias. Methods Context The Safe Motherhood Project is a collaboration between researchers at Jimma University and the University of Ottawa. In 2016, in the preliminary stages of this project, we undertook a qualitative rapid assessment study to explore community perceptions and experiences related to the following: health (generally); MNCH service use; MNCH traditions, practices, and beliefs; stakeholder roles in promoting MNCH; and MNCH knowledge acquisition and sources of information (Asfaw et al., 2019; Bergen et al., 2018; Mamo et al., 2019). The pri- mary purpose of the rapid assessment study was to inform the design and delivery of the two Safe Motherhood Project interventions (information, education and com- munication activities, and improved infrastructure and functionality of maternal waiting areas); the findings from the rapid assessment study were also used to develop survey tools for use in the project’s baseline data collec- tion in 2017–2018. As part of the qualitative rapid assessment study, we conducted a series of 24 in-depth interviews and 12 focus group discussions in rural areas of Jimma Zone, Ethiopia. The participants included various stakeholders in MNCH. Health extension workers (HEWs) (locally recruited, female community health workers comprising the most decentralized component of the formal health workforce in Ethiopia), members of the Women’s Development Army (a volunteer-based community group that supports the work of HEWs), members of the Male Development Army (a volunteer-based community group focusing on aspects of community development related to physical infrastructure or agriculture), and religious leaders (pre- dominantly Muslim imams) participated in in-depth inter- views, while focus group discussions with 6–12 participants were held with female and male community members. The focus group discussions and in-depth inter- views posed similar questions, with less specificity in the focus group discussions and more detailed probes in the in-depth interviews. The study took place at six rural or semirural health posts (basic health facilities staffed by 2–3 HEWs), located in three districts of Jimma Zone. The six health posts were invited to participate due to their accessibility from Jimma, established contact between health post personnel and members of the research team, and willingness of at least one HEW to participate. The HEWs at each of the six health posts, under guidance from the Safe Motherhood Project coordinator, recruited other participants for the in-depth interviews and focus group discussions. Participants received a small financial com- pensation to help cover transportation costs. The Safe Motherhood Project researchers worked closely with a team of data collectors from Jimma University, who assisted with fieldwork tasks (namely, conducting the in-depth interviews and focus group dis- cussions) and post-fieldwork tasks (namely, preparing written transcripts of the in-depth interviews and focus group discussions in English). Other activities, including Bergen and Labonté 785 keeping field notes, participating in a series of debriefing and reflection sessions, and participating in research team meetings, were completed concurrently (see the “Data Sources” section for more details). Ethical clearance for this research was obtained from Jimma University College of Health Sciences Institutional Review Board and University of Ottawa Health Sciences and Science Research Ethics Board; the research was conducted in accordance with the prevailing ethical principles. Working With Data Collectors The study employed nine data collectors (7 male and 2 female) who were graduate students and/or staff at Jimma University, and fluent in English and Afan Oromo (the local language). They each had previous experience or exposure to qualitative research. The data collectors par- ticipated in a two-day induction workshop facilitated by three of the Safe Motherhood Project researchers and the project coordinator. The induction workshop introduced the overall aims of the Safe Motherhood Project and the rapid qualitative assessment study; researchers facilitated sessions about ethical considerations, confidentiality, and field notes. The group worked through an extensive review of the research instruments, including discussions about translation into Afan Oromo, and minor modifica- tions to the in-depth interview and focus group discussion guidelines. Fieldwork took place over 6 days, with one research site visit per day. Seven or eight data collectors (including at least one female) were in attendance each day, and at least two Safe Motherhood Project researchers from Jimma University and/or the University of Ottawa were present in a supervisory capacity. Focus group discus- sions were facilitated by teams of two data collectors, and there was one data collector per in-depth interview. Data collectors obtained written or oral informed consent for participation from each participant at the beginning of the in-depth interview or focus group discussion. In most cases, oral consent was obtained due to low literacy lev- els and cultural considerations about the acceptability of signing consent forms. Data Sources We captured data collectors’ and project researchers’ reflections and experiences about social desirability bias through three main sources: field notes, an internal field report, and relevant email exchanges between the Safe Motherhood Project researchers. •• Project researchers and data collectors partici- pated in daily debriefing sessions after each day of fieldwork to discuss key findings, challenges, and other issues and to share strategies to enhance data collection. These sessions took place directly after the data collection activities to capture the group’s impressions. Typically, they were held in a private area at the field site, during transporta- tion home from the field site, or in a private area of a coffee shop. The sessions became more detailed and longer in duration as the study pro- gressed. To document their experiences and impressions, and to capture the issues raised dur- ing the debriefing sessions, data collectors and researchers kept extensive field notes, which were submitted to the research team. The protocol for field notes was discussed in the pre-fieldwork training session, based on a field note guide devel- oped by the research team. •• Midway through the fieldwork, a larger meeting facilitated by a senior researcher on the Safe Motherhood Project and attended by project researchers, the project coordinator, and data col- lectors, provided an opportunity to elaborate on issues identified in the daily debriefing sessions and brainstorm possible solutions. This meeting was held on the university campus and lasted for about 2 hr. The minutes and notes from this meet- ing were summarized in the internal field report. •• In the 2 weeks following the completion of field- work, project researchers held individual reflec- tion sessions with each data collector to inquire about the data collection activities, and their impressions of the findings; the salient aspects of these discussions were summarized as field notes and served as a major input for the internal field report. These sessions took place on the university campus and lasted about half an hour each. They were administered by two members of the research team, and though not mandatory for the data col- lectors, all data collectors agreed to participate. Throughout all stages of the study, the project researchers in Ethiopia and Canada communicated through email about the progress of the research, highlighting salient observations and reflections. Some aspects of these exchanges were similar to the project researchers’ field notes, though the questions and prompts from off-site project researchers generated new avenues of thought. Data Analysis Data analysis drew from the content of field notes, the internal field report, and relevant email exchanges between Safe Motherhood Project researchers. In total, 51 documents were imported into Atlas.ti software for 786 Qualitative Health Research 30(5) analysis. We coded the material by two major themes: how social desirability bias may have presented in the research, and strategies adopted by the research team to navigate these influences. Then, we summarized emergent ideas within each theme. These ideas tended to be present across multiple sources of data. For example, issues raised in field notes were often dis- cussed through email exchanges and highlighted in the internal field report. We consulted the 36 in-depth interview and focus group discussion transcripts gen- erated for the study to cross-reference examples men- tioned in the documents (when provided), or to identify examples that illustrated ideas raised in the documents (when examples were not provided). When possible, we used the date from the field note to obtain a tran- script from the same day to find an example that illus- trated the issue identified by the data collector or researcher. Results Detecting Social Desirability Tendencies All data collectors employed in the study believed that some participant responses reflected social desirability tendencies. As university students or staff with prior experience in community-based public health research, the data collectors relied on their preexisting exposure to the research context as well as their interactions with the participants to identify common cues suggesting social desirability tendencies in in-depth interview and focus group discussion participant answers. These included: denial of (already known) problems, challenges, or short- comings; providing only partial or vague answers (palter- ing); excessive and repeated praise for government initiatives; nervous facial expression and other body lan- guage cues; and inconsistent use of advanced vocabulary related to the study topic. The data collectors provided examples of instances when participants exhibited these behaviors in the study interviews (Table 1). Data collectors did not think that all participants demonstrated social desirability tendencies, and they also did not think that all responses were biased. Data collectors discussed how they distinguished between participants’ authentic beliefs (e.g., that there were no problems, or that the government deserved excessive praise) versus participants’ presentation of their reality in ways that suggested a social desirability bias (e.g., immediate praise or denial of problems without time for reflection). Some data collectors noticed a pattern whereby participants displayed more social desirability tendencies at the onset of the in-depth interview or focus group discussion and the behavior diminished as the encounter progressed. They noted that there were sometimes inconsistencies in a participant’s responses within a single in-depth interview or focus group dis- cussion. For example, while some participants expressed that home births never happen, when they were later asked to state where they had given birth to their own children, the majority indicated that the births had taken place at home. This contradiction flagged the possibility of incongruence between their expressed beliefs and practices regarding place of birth. Some data collectors noted that participants in one-on- one interviews were more likely to respond in a socially desirable manner if another individual was within ear- shot, especially if the individual was an HEW or an influ- ential community member. In focus group discussions, if one of the participants held a position of power within the community (e.g., a community leader or relative of a community leader), the other participants would some- times let them speak first, and then agree with their opinions. Strategies to Minimize Social Desirability Bias Throughout the data collection activities, the data collec- tors developed strategies to avoid or mitigate social desirability tendencies. When setting up the in-depth interviews or focus group discussions, data collectors tried to ensure that the location was private and not within earshot of others. At the beginning of each in- depth interview or focus group discussion, the data col- lectors confirmed that the participants met the criteria to participate in the study. For example, development army leaders and HEWs, though part of the community, were not permitted to participate in focus group discussions for community members. While conducting the in-depth interviews or focus group discussions, data collectors used various approaches to establish rapport with participants, including the use of humor, self-disclosure, and/or making displays of respect. Given that data collectors were outsiders to the community, these strategies helped to put participants at ease and make them feel comfortable. Data collectors came to appreciate the importance of clearly explaining details about the study to participants, including the purpose of the overall study, how the data would be used, and confidentiality and anonymity procedures. Data collectors felt that with a better understanding of the purpose of the research (i.e., to learn about MNCH perspectives and experiences, positive and negative, and support inter- ventions to promote MNCH) the participants were less likely to view the encounter as an audit or evaluation of their performance. As they became more familiar with the in-depth inter- view and focus group discussion guides and more Bergen and Labonté 787 experienced in the field, data collectors developed ways of asking questions to minimize social desirability biases (Table 2). These techniques attempted to minimize par- ticipants’ social desirability bias and to promote more candid and detailed responses. Data collectors noted that if they suspected a response to reflect social desirability tendencies, they would remain nonconfrontational and respectful, and try one of these techniques to try to elicit a more authentic response. In a few cases where data col- lectors suspected that the participant had been primed about what to say (e.g., in a preinterview briefing by the HEW), they listened as the participant expressed these ideas, and then proceeded to ask questions about the par- ticipants’ own experiences and perspectives. Table 1. Cues for Detecting Social Desirability Tendencies. Behavior Example Denial of any problem, challenge, or shortcomings A major government initiative in Ethiopia is for all women to give birth at a health facility. When asked about usual place of delivery, women in a focus group discussion insisted that there was complete adherence to the facility birth initiative, with comments like: “No women ever give birth at home. We are perfect in this kebele.” Providing partial or vague answers (paltering) Religious leaders are prominent opinion leaders in many communities and are sometimes called upon to reinforce health messages. When probed about his role in promoting health, one religious leader responded: “We have a role. It is important.” Excessive and repeated praise for government initiatives The Ethiopian Ministry of Health has introduced a series of major primary health care reforms since 2004. When asked about maternal health challenges in their community, participants in a focus group discussion answered as follows: “The government has helped us all. After the government plan the health extension workers were with us, and mothers did not encounter problems. Thanks to our government.” Nervous facial expression and body language Data collectors asked participants about what actions would be needed to be taken by various stakeholders to further improve health in their communities. Focus group discussion participants looked to others for affirmation of their opinions or appeared nervous, especially if prompted to comment on the roles of stakeholders with political connections. Inconsistent use of vocabulary Data collectors noted that some participants used advanced vocabulary (in the local language) when discussing certain topics, which was inconsistent with their education level and other use of language. One participant used advanced terms to refer to phases of antenatal care in a manner that the data collector suspected resulted from a preinterview briefing by the health extension worker (HEW) (i.e., where the participant was advised how to respond to certain questions). Table 2. Techniques for Asking Questions in a Manner That Limits Social Desirability Responses, With Examples. Technique Original Approach Improved Approach Indirect questioning Posing questions directly Example: What harmful traditional practices are done in your community? Posing indirect questions about the past, or the behaviors of others Example: What harmful practices do you know about that existed in the past? Do you think these practices might exist today in a hidden or open manner? Providing assurances Asking questions and waiting for participants to respond Responding to hesitant participants by assuring them that their opinions are not wrong, and asking them to please speak freely Briefly explaining the confidentiality and anonymity procedures at the beginning of the in-depth interview or focus group discussion Thoroughly explaining the confidentiality and anonymity procedures at the beginning of the in-depth interview or focus group discussion, and then offering reminders throughout the encounter, especially preceding sensitive questions Probing for more information Accepting generic or incomplete responses Asking follow-up questions, or prompts Example: Can you explain more about why you feel this way? Requesting stories or examples Accepting generic or incomplete responses Requesting that participants provide a story or example to illustrate their response Example: Can you tell me about a time that you experienced this? Prefacing the question Posing questions directly and with little context Example: Why do women attend (or not attend) antenatal care visits? Providing context when asking questions, acknowledging that all communities have challenges and that people have diverse experiences Example: We know that some women attend antenatal care visits and others do not. What do you think are the reasons why women do (and do not) attend? 788 Qualitative Health Research 30(5) Discussion To date, little research attention has been given to devel- oping practical strategies for how to identify and mini- mize social desirability during qualitative fieldwork. This article proposes approaches for researchers to work together with data collectors to identify context-specific forms of social desirability bias in health research and to develop protocols to strengthen data collection practices in ways that expose social desirability tendencies and facilitate open communication within the research team. Although our findings are not an attempt to measure social desirability bias per se, they do provide an account of how the implications of social desirability bias in qual- itative research can be better understood and addressed. Our findings add to a small body of literature that explores social desirability in community-based qualita- tive health research in Ethiopia. Social environments within Ethiopia, a highly collectivist society, tend to reinforce conformity and compliance to established norms and values. Certain strategies that we identified for limiting social desirability bias (i.e., detailing the purpose of the study and providing assurances) have been applied by other Ethiopian-based research groups. King et al. (2016), in their qualitative study with HEWs about maternal health service use, counteracted sus- pected social desirability bias by communicating the benefits of the research and assuring participants of the confidentiality measures (King et al., 2016). R. Jackson et al. (2017), in effort to limit social desirability bias, clarified that the purpose of their study was not to evalu- ate job performance of the HEWs that they interviewed but to better understand delays in access to care (an important distinction, which was also noted in our find- ings). In both articles, detailed commentary about the deployment and/or success of these strategies was, understandably, outside the scope of reporting. Personality Attribute or Response Style? When data collectors suspected that participants were exhibiting social desirability tendencies, they primarily relied on their intuition, knowledge, and previous expo- sure to distinguish impression management response styles from personality attributes associated with exag- gerated self-regard. The data collectors attest to this judg- ment sharpening as they became more immersed in the research and were generally very confident in their ability to detect instances of social desirability bias; however, beyond describing the cues for detecting social desirabil- ity bias and patterns in their occurrence, they could offer little explanation about how they made the distinction. Psychology research has demonstrated that certain condi- tions and preparations can enhance an individual’s ability to understand and make judgments about the behaviors and interactions of others (Shaw et al., 2013; ten Brinke et al., 2016). These findings lend support to the notion that the data collectors can become more effective at detecting social desirability bias as a result of group dis- cussions and debriefings. Researchers engaged in in- depth interviews or focus group discussions may benefit from training programs that integrate a focus on detecting social desirability bias. Data collectors in our study relied on behavioral cues to detect social desirability, including both verbal and nonverbal cues. Previous research supports the idea that individuals may be more effective in detecting bias when considering both types of cues (Vrij et al., 2000). We note, however, that while nervousness may be a cue of social desirability bias, it is also a common occur- rence in many social interactions, especially if the indi- viduals are not familiar with one another, or if they are not used to sharing opinions in a group setting. The data collectors in our study suggested that they sometimes based their suspicion of social desirability bias on con- tradictions and inconsistencies in participants’ responses. We interpret this finding with caution, given that participants in research studies may change their answers, may misinterpret questions, and/or may genu- inely have conflicting or complex views. Data collectors may, however, become familiar with the sensitivity of particular questions over time, and thus be able to better ascertain patterns of social desirability bias (replicating aspects of social desirability measurement tools applied in psychology; Furnham, 1986). Questioning Techniques, Rapport Building, and Power Differentials The wording and delivery of questions by data collectors during in-depth interviews and focus group discussions might steer participants toward biased or unbiased responses. Fisher (1993), who examined the indirect questioning technique on social desirability bias in mar- keting, found a difference between the mean scores of socially sensitive questions when asked directly versus indirectly; the difference was not apparent for socially neutral questions. Fisher (1993) further distinguishes between questions that address personal outcomes (based on internalized values that are independent of social con- siderations) versus normative constructs (motivate by the expectations and approval/disapproval by others), hypothesizing that the former line of questioning is less subject to social desirability bias than the latter. Thus, participants may be less likely to engage in impression management when their responses do not require them to make social judgments. Techniques such as indirect Bergen and Labonté 789 questioning and prefacing the question (whereby the data collector asserts the social judgment) may be useful options to reorient the nature of the question and thereby reduce social desirability bias, encouraging participants to speak freely, without judgment. Highly collectivist societies tend to make sharper in-group and out-group distinctions and are less forthcoming to those perceived to be part of the out-group (Johnson & Van de Vijver, 2002). Thus, researchers engaged with participants in these settings may require deliberate efforts to minimize the extent to which they are viewed as the out-group, especially in interactions with individuals who hold less influential social positions. Rapport building techniques such as humor, self-dis- closure, and researcher–participant matching by sociode- mographic characteristics have been applied in other settings. Vallano and Compo (2011), studying the accu- racy of eye-witness accounts, found that verbal rapport building through interviewer self-disclosure and express- ing interest in the respondent improved recall accuracy of past events and acted as a buffer against inaccurate reporting. Humor is another strategy that can help to alle- viate power differentials and facilitate rapport building in group-based research (Hewer et al., 2019). Matching by sociodemographic characteristics (e.g., gender, ethnicity, and age) is a common strategy to enhance relations and foster a mutual sense of understanding and empathy. In settings with prevailing ethnic tensions or strong gender norms, matching by these characteristics can be particu- larly important (Chiumento et al., 2018). In our study, data collectors used approaches that fit with their person- ality to establish rapport, which may have relied on a shared characteristic with the participant (especially lan- guage and cultural ties) but sometimes relied on a differ- ing characteristic (such as younger data collectors making gestures of respect to elderly participants). Interestingly, in the context of our research, gender matching between data collectors and participants was not a factor that data collectors considered to be important with regard to social desirability bias. The benefits of rapport building, however, have lim- its. In a study in Vietnam, Latkin et al. (2016) found that the interviewer strategy of establishing rapport with the participant did not always have the intended effect: While participants underscored the importance of feel- ing comfortable and trusting the interviewer, this famil- iarity also led to decreased reports of risk behavior. The study, which asked a subsample of participants from a previous survey to reflect on the experiences of them- selves and other participants, discovered that partici- pants sometimes tailored their answers to ensure that they met the criteria to be part of the study, or to alter the duration of the survey (Latkin et al., 2016). Researchers should be cognizant of the potential risks of rapport building and the power imbalances implicit in the researcher–participant relationship. Power differentials were apparent in our study on several levels and impacted how the research team came to understand issues related to social desirability bias. Initially, some of the data collectors in our study were hesitant to openly discuss the possibility of such bias. Given the complex role that data collectors assumed as partial insiders and outsiders both in the community and within the research team, the impetus to discuss the pos- sibility of social desirability bias was low, while the potential costs, such as reflecting upon the interviewing/ facilitation skills of the data collector or questioning the integrity of community participants in the study, were high. Lowering social costs may facilitate more open conversations (ten Brinke et al., 2016). Senior research- ers and supervisors can lower data collectors’ social costs by demonstrating an awareness of the possibility of social desirability bias, offering examples of when they have previously encountered social desirability bias, and providing necessary assurances about the inev- itability of social desirability tendencies. Open commu- nication between the researcher and data collectors can strengthen the research process, facilitate nuanced insights into the data, and enable comprehensive and transparent reporting (Bergen, 2018). Power differentials also exist within communities. The data collectors in our study mentioned that many of the HEW-recruited participants had links to political leaders, and that the participants had been selected because of these links. This recruitment method may have exacerbated social desirability tendencies through selection bias (i.e., those with certain views and experi- ences were invited to participate), preinterview brief- ings by the HEWs about how to respond, exposure to political propaganda, desire to protect the community reputation, and expectations that participants would be financially compensated. Preempting Social Desirability Bias Rather than ignoring or denying the persistence of social desirability bias, researchers should acknowledge it as a reality and take measures to account for it throughout the research process. If unattended to at previous stages of the research, social desirability bias can affect the quality of the data, leaving researchers with incomplete and shal- low contextualization of responses, missing information about weaknesses and barriers, and/or a preponderance of one-sided perspectives. Taking active measures to detect and limit biases strengthens the rigor and transparency of qualitative research and provides opportunities for reflex- ive contemplation. To this end, embracing tensions and ambiguity in research, and leaving room for flexibility, 790 Qualitative Health Research 30(5) creativity, and responsiveness can lead researchers to new insights (Simovska et al., 2019). Researchers in Ethiopia and abroad may benefit from the lessons learned in our study. In particular, we high- light the importance of planning regular debriefing ses- sions as a research team. This encourages data collectors to discuss and address challenges on an ongoing basis, helps to keep morale high, and encourages skill develop- ment. We also underscore that researchers’ engagement with data collectors and the data collection process can yield insights into possible sources of bias that may not be apparent solely by reviewing transcripts. Although these measures emanate from our work in Ethiopia, they are general in nature and may be adapted for any interna- tional setting where social desirability tendencies are apparent (see Table 1). We welcome further studies or comments on their applicability in other contexts. (As an extension of our findings, Table S1 in the supplemental file specifies measures that research teams can take to minimize social desirability bias in the pre-fieldwork stage of research.) Limitations and Future Research Directions Our findings are derived from studying field notes, obser- vations, and communications, which provided a rich array of perspectives from members of the research team. We acknowledge that our findings derive from our expe- riences working with a research team that is native to the research setting, which may not be feasible in all research endeavors. We did not engage with the community-based participants in the study to explore and characterize social desirability tendencies per se. Previous research has sug- gested that early engagement with community leaders provides critical insight when preparing for subsequent interviews with community members (Jimenez et al., 2019), which could be a useful strategy to anticipate social desirability tendencies. Future research may seek to incorporate the participants’ feedback about why and when they adopt social desirability tendencies, which would triangulate our findings. Additional research is warranted to better understand the underlying reasons why (and when) individuals adopt social desirability tendencies and then, if appropri- ate, identify approaches to respond to the underlying causes that drive the behavior. The study of social desir- ability bias in qualitative research may benefit from the integration of learnings from cultural consensus theories (i.e., theories studying cultural beliefs and the degree to which individuals know or report them; Weller, 2007) and integration with approaches to quantify and account for reporting biases. We also call for more research to explore strategies to account for social desirability bias during data analysis and reporting, as well as to compare how strategies to detect and limit social desirability bias can be translated to different settings and research topics. We acknowledge that there may be an inherent circular issue in the discussion of social desirability bias, as the investigators themselves may also demonstrate desir- ability bias on the subject. Conclusion While social desirability bias is often cited as a limita- tion in qualitative research, detailed characterization of the phenomenon is lacking, especially in highly collec- tivist settings such as rural Ethiopia. Our findings, which characterized social desirability bias issues in our research about MNCH in Jimma Zone, contribute to the broader literature in two important ways. First, acknowl- edging that social desirability bias was an inherent part of our research, we describe strategies that we devel- oped to detect and limit such bias. These strategies advance current approaches to qualitative data collec- tion and stand to benefit other researchers undertaking studies of a similar nature. Second, through our process of ongoing engagement with the research team about social desirability issues, we gained insight into how research groups can foster a heightened consciousness about social desirability bias. We advocate for research teams to create opportunities to discuss social desirabil- ity bias at various stages of the research process in a trusting and confidential environment. Acknowledgments The authors would like to acknowledge members of the research team (Lakew Abebe, Muluemebet Abera, Yisalemush Asefa, Shifera Asfaw, Kunuz Haji Bedru, Gemechu Beyene, Gebeyehu Bulcha Duguma, Getachew Kiros, Manisha A. Kulkarni, Jaameeta Kurji, Abebe Mamo, Sudhakar Morankar, and Corinne Packer). A special thanks to Lakew Abebe, Shifera Asfaw, Alzahra Hudani, Getachew Kiros, Manisha A. Kulkarni, Abebe Mamo, and Sudhakar Morankar for reviewing and pro- viding comments on drafts of the manuscript. Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was carried out with the aid of a grant from the Innovating for Maternal and Child Health in Africa initiative— a partnership of Global Affairs Canada (GAC), the Canadian Institutes of Health Research (CIHR), and Canada’s International Development Research Centre (IDRC). Bergen and Labonté 791 ORCID iD Nicole Bergen https://orcid.org/0000-0002-8161-2599 Supplemental Material Supplemental Material for this article is available online at jour- nals.sagepub.com/home/qhr. Please enter the article’s DOI, located at the top right hand corner of this article in the search bar, and click on the file folder icon to view. References Asfaw, S., Morankar, S., Abera, M., Mamo, A., Abebe, L., Bergen, N., Kulkarni, M.A., & Labonté, R. (2019). Talking health: Trusted health messengers and effective ways of delivering health messages for rural mothers in Southwest Ethiopia. Archives of Public Health, 77, Article 8. Bergen, N. (2018). Narrative depictions of working with lan- guage interpreters in cross-language qualitative research. International Journal of Qualitative Methods, 17(1). https://doi.org/10.1177/1609406918812301 Bergen, N., Mamo, A., Asfaw, S., Abebe, L., Kurji, J., Kiros, G., Abera, M., Bulcha Duguma, G., Haji Bedru, K., Kulkarni, M.A., Labonté, R. & Morankar, S. (2018). Perceptions and experiences related to health and health inequality among rural communities in Jimma Zone, Ethiopia: A rapid quali- tative assessment. International Journal for Equity in Health, 17(1). https://doi.org/10.1186/s12939-018-0798-9 Chiumento, A., Rahman, A., Machin, L., & Frith, L. (2018). Mediated research encounters: Methodological consider- ations in cross-language qualitative interviews. Qualitative Research, 18, 604–622. Fisher, R. J. (1993). Social desirability bias and the validity of indirect questioning. Journal of Consumer Research, 20(2), 303–315. Freeth, D., Sandall, J., Allan, T., Warburton, F., Berridge, E., Mackintosh, N., & Abbott, S. (2012). A methodologi- cal study to compare survey-based and observation-based evaluations of organisational and safety cultures and then compare both approaches with markers of the quality of care. Health Technology Assessment, 16(25). https://doi. org/10.3310/hta16250 Furnham, A. (1986). Response bias, social desirability and dis- simulation. Personality and Individual Differences, 7(3), 385–400. Girma, F., Jira, C., & Girma, B. (2011). Health services utilization and associated factors in Jimma zone, south west Ethiopia. Ethiopian Journal of Health Sciences, 21(3), 85–94. Grimm, P. (2010). Social desirability bias. In J. Sheth & N. Malhotra (Eds.), Wiley international encyclopedia of mar- keting. https://doi.org/10.1002/9781444316568.wiem02057 Harvey, S. A. (2018). Observe before you leap: Why observa- tion provides critical insights for formative research and intervention design that you’ll never get from focus groups, interviews, or KAP surveys. Global Health: Science and Practice, 6, 299–316. Hewer, R., Smith, K., & Fergie, G. (2019). The social function- ality of humor in group-based research. Qualitative Health Research, 29, 431–444. Jackson, D. N., & Messick, S. (1958). Content and style in per- sonality assessment. Psychological Bulletin, 55(4), 243– 252. https://doi.org/10.1037/h0045996 Jackson, R., Tesfay, F. H., Gebrehiwot, T. G., & Godefay, H. (2017). Factors that hinder or enable maternal health strategies to reduce delays in rural and pastoralist areas in Ethiopia. Tropical Medicine & International Health, 22(2), 148–160. https://doi.org/10.1111/tmi.12818 Jimenez, M. E., Hudson, S. V., Lima, D., & Crabtree, B. F. (2019). Engaging a community leader to enhance prepa- ration for in-depth Interviews with community members. Qualitative Health Research, 29(2), 270–278. Johnson, T. P., & Van de Vijver, F. J. (2002). Social desirabil- ity in cross-cultural research. In J. Harness, F. J. Vijver, & P. Mohler (Eds.), Cross-cultural survey methods (pp. 193–202). John Wiley. Kelly, C. A., Soler-Hampejsek, E., Mensch, B. S., & Hewett, P. C. (2013). Social desirability bias in sexual behavior reporting: Evidence from an interview mode experiment in rural Malawi. International Perspectives on Sexual and Reproductive Health, 39(1), 14–21. https://doi. org/10.1363/3901413 King, R., Jackson, R., Dietsch, E., & Hailemariam, A. (2016). Utilisation of maternal health services in Ethiopia: A key informant research project. Development in Practice, 26(2), 158–169. https://doi.org/10.1080/09614524.2016 .1131244 Krumpal, I. (2013). Determinants of social desirability bias in sensitive surveys: A literature review. Quality & Quantity, 47(4), 2025–2047. https://doi.org/10.1007/s11135-011- 9640-9 Lalwani, A. K., Shavitt, S., & Johnson, T. (2006). What is the relation between cultural orientation and socially desirable responding? Journal of Personality and Social Psychology, 90(1), 165–178. https://doi.org/10.1037/0022- 3514.90.1.165 Latkin, C. A., Mai, N. V. T., Ha, T. V., Sripaipan, T., Zelaya, C., Le Minh, N., & Go, V. F. (2016). Social desirability response bias and other factors that may influence self- reports of substance use and HIV risk behaviors: A quali- tative study of drug users in Vietnam. AIDS Education and Prevention, 28(5), 417–425. https://doi.org/10.1521/ aeap.2016.28.5.417 Maes, K. (2017). The lives of community health work- ers: Local labor and global health in urban Ethiopia. Routledge. Mamo, A., Morankar, S., Asfaw, S., Bergen, N., Kulkarni, M. A., Abebe, L., Labonté, R., Birhanu, Z., & Abera, M. (2019). How do community health actors explain their roles? Exploring the roles of community health actors in promoting maternal health services in rural Ethiopia. BMC Health Services Research, 19, Article 724. Mooney, A. C., Campbell, C. K., Ratlhagana, M.-J., Grignon, J. S., Mazibuko, S., Agnew, E., & Lippman, S. A. (2018). Beyond social desirability bias: Investigating inconsisten- cies in self-reported HIV testing and treatment behaviors among HIV-positive adults in North West province, South Africa. AIDS and Behavior, 22(7), 2368–2379. https://doi. org/10.1007/s10461-018-2155-9 https://orcid.org/0000-0002-8161-2599 https://doi.org/10.1177/1609406918812301 https://doi.org/10.1186/s12939-018-0798-9 https://doi.org/10.3310/hta16250 https://doi.org/10.3310/hta16250 https://doi.org/10.1002/9781444316568.wiem02057 https://doi.org/10.1037/h0045996 https://doi.org/10.1111/tmi.12818 https://doi.org/10.1363/3901413 https://doi.org/10.1363/3901413 https://doi.org/10.1080/09614524.2016.1131244 https://doi.org/10.1080/09614524.2016.1131244 https://doi.org/10.1007/s11135-011-9640-9 https://doi.org/10.1007/s11135-011-9640-9 https://doi.org/10.1037/0022-3514.90.1.165 https://doi.org/10.1037/0022-3514.90.1.165 https://doi.org/10.1521/aeap.2016.28.5.417 https://doi.org/10.1521/aeap.2016.28.5.417 https://doi.org/10.1007/s10461-018-2155-9 https://doi.org/10.1007/s10461-018-2155-9 792 Qualitative Health Research 30(5) Østebø, M. T., Cogburn, M. D., & Mandani, A. S. (2018). The silencing of political context in health research in Ethiopia: Why it should be a concern. Health Policy and Planning, 33(2), 258–270. https://doi.org/10.1093/ heapol/czx150 Paulhus, D. L. (1984). Two-component models of social desirable responding. Journal of Personality and Social Psychology, 46(3), 598–609. Paulhus, D. L. (2001). Socially desirable responding: The evolution of a construct. In H. I. Braun, D. N. Jackson, & D. E. Wiley (Eds.), The role of constructs in psychologi- cal and educational measurement (pp. 49–69). Lawrence Erlbaum. Pauls, C. A., & Stemmler, G. (2003). Substance and bias in social desirability responding. Personality and Individual Differences, 35(2), 263–275. Shaw, J., Porter, S., & ten Brinke, L. (2013). Catching liars: Training mental health and legal professionals to detect high-stakes lies. The Journal of Forensic Psychiatry & Psychology, 24(2), 145–159. https://doi.org/10.1080/1478 9949.2012.752025 Simovska, V., Colding Lagermann, L., Salah Abduljalil, H., Lerche Mørck, L., & Kousholt, D. (2019). Inside out: What we (don’t) talk about when we talk about research. Qualitative Research, 19, 113–130. ten Brinke, L., Vohs, K. D., & Carney, D. R. (2016). Can ordi- nary people detect deception after all? Trends in Cognitive Sciences, 20(8), 579–588. https://doi.org/10.1016/j.tics. 2016.05.012 Vallano, J. P., & Compo, N. S. (2011). A comfortable wit- ness is a good witness: Rapport-building and suscepti- bility to misinformation in an investigative mock-crime interview. Applied Cognitive Psychology, 25(6), 960– 970. Vrij, A., Edward, K., Roberts, K. P., & Bull, R. (2000). Detecting deceit via analysis of verbal and nonverbal behavior. Journal of Nonverbal Behavior, 24(4), 239– 263. Weller, S. C. (2007). Cultural consensus theory: Applications and frequently asked questions. Field Methods, 19(4), 339– 368. https://doi.org/10.1177/1525822X07303502 Author Biographies Nicole Bergen MHSc, BSc is a PhD Candidate in Population Health, University of Ottawa, ON, Canada. Ronald Labonté PhD, FCAHS, HonFFPH is Distinguished and former Canada Research Chair in Globalization and Health Equity, and a professor in the School of Public Health and Epidemiology, University of Ottawa, ON, Canada. https://doi.org/10.1093/heapol/czx150 https://doi.org/10.1093/heapol/czx150 https://doi.org/10.1080/14789949.2012.752025 https://doi.org/10.1080/14789949.2012.752025 https://doi.org/10.1016/j.tics.2016.05.012 https://doi.org/10.1016/j.tics.2016.05.012 https://doi.org/10.1177/1525822X07303502 1049732319889354 d1acb38e-7501-44ff-b460-2b95fec2116c