Asking women to estimate the amount of food they gave their child in the 24-hour dietary recall in terms of cups or bowls was considered illogical since roti (flatbread), a common food, does not fit into cups
Survey tool component assessed . | Explanation . | Example . |
---|---|---|
Word choice | Words used in the survey questions may not be understood by respondents, may have unintended alternative meanings, may be overly vague or specific or may be less natural than alternative words | When translating surveys from English to Hindi, we found that professional translators and Hindi-speaking researchers with experience in rural areas often selected formal Hindi words that were unfamiliar to rural women |
Syntax | Sentences in survey questions may be too complex or too long, reducing respondent capacity to retain key features of the question | The question ‘During your time in the health facility did the doctors, nurses, or other health care providers introduce themselves to you when they first came to see you?’ contained too many words and clauses. By the time the researcher finished reading it, the respondent lost track of the core question |
Sequencing | The order of questions may be inappropriate. Placing sensitive or emotionally charged questions too early in the survey can be uncomfortable for respondents and damage respondent–enumerator rapport, reducing the likelihood of a respondent providing a truthful and complete response | A survey on respectful maternity care initially asked post-partum women if they were verbally or physically abused during childbirth within the first few survey questions, to ensure that this crucial question was answered before any respondent fatigue set in. However, cognitive interviews revealed that women were uncomfortable with the question and unlikely to disclose abuse without first establishing rapport through a range of less emotionally intense questions |
Sensitivity | Questions or response options may be too direct or include topics that are insufficiently contextualized, leading to respondent and enumerator discomfort and eroding rapport | When asking women about their birth companions, they found it strange and uncomfortable to be probed about whether male family members were with them |
Response options | Response options may be insufficient to capture the actual range of responses or may be incomprehensible or uncomfortable for respondents | Likert scales with more than three response options were incomprehensible to most rural Indian women we interviewed. Asking women to estimate the amount of food they gave their child in the 24-hour dietary recall in terms of cups or bowls was considered illogical since roti (flatbread), a common food, does not fit into cups |
Resonance with local worldviews and realities | Questions may ask about domains of importance to the research team but that do not resonate with respondent views or realities | ‘Being involved in decisions about your health care’ is a domain of global importance in respectful maternity care. However, in rural India, the concept of healthcare workers involving the patient in healthcare decisions was unfamiliar and, when explained, considered undesirable |
Cognitive mismatch | Questions may access respondent cognitive domains that do not map on to the domains intended by the researchers | Women were asked whether they would recommend the place where they gave birth to a friend, as a proxy for quality of care. However, women frequently responded ‘no’ because they did not have friends, did not want to tell other women what to do or did not think they should make recommendations for other people—which was unrelated to their maternity care experiences |
Memory | Questions or response options may seek to access respondent memories in ways that are too cognitively demanding | Recalling specific post-partum practices from many months ago may not be possible for some respondents |
While it is usually possible to identify and remedy linguistic and syntax issues in survey questions, cognitive interviewing cannot always solve deeper problems with survey research. Cognitive interviews may illuminate question failures arising from a mismatch between the underlying concepts that the survey attempts to measure and the concepts that resonate with the respondent’s worldview and reality ( Schuler et al. , 2011 ; Scott et al. , 2019 ). In these cases, question revision will not achieve cognitive alignment between researcher and participant. Instead, researchers must drop questions from the survey and potentially generate new items.
There are many terms and approaches used for strengthening surveys, some of which may encompass cognitive interviewing or include components of it without applying the label ( Table 2 ). We argue however, that cognitive interviewing should be a standalone approach integrated into a larger process of survey tool development.
Approaches to strengthening surveys
Approach . | Description . | Comparison to cognitive interviewing . | Issue . |
---|---|---|---|
Expert review | Subject area experts review the survey tool and judge how well each questionnaire item truly reflects the construct it is intended to measure | Experts are unable to predict how the survey respondents will interpret the questions | |
Respondent-driven pretesting | A small group of participants with the same characteristics as the target survey population complete the survey. Researchers elicit feedback during the survey or at the end through debriefings. Feedback elicitation can include targeted probes about questions that appeared problematic, in-depth exploration of each question, probing on a random sub-set of questions, or asking participants to rate how clear the question was | ) | Low methodological clarity: can be the same as cognitive interviewing or quite different |
Translation and back translation | After translating a survey from the origin to the target language, a different translator ‘blindly’ translates the survey back. Differences are then compared and resolved ( ) | Involves bilingual translators whose world view and experience do not match the target population’s, making them unable to comment on the tool’s appropriateness | |
Pilot testing | Enumerators administer the survey to a small group of participants with the same characteristics as the target survey in as close to real world conditions as possible | Focuses on the mechanics of implementation while cognitive testing focuses on the survey questions achieving shared understanding between researcher intent and respondent interpretation |
Approach . | Description . | Comparison to cognitive interviewing . | Issue . |
---|---|---|---|
Expert review | Subject area experts review the survey tool and judge how well each questionnaire item truly reflects the construct it is intended to measure | Experts are unable to predict how the survey respondents will interpret the questions | |
Respondent-driven pretesting | A small group of participants with the same characteristics as the target survey population complete the survey. Researchers elicit feedback during the survey or at the end through debriefings. Feedback elicitation can include targeted probes about questions that appeared problematic, in-depth exploration of each question, probing on a random sub-set of questions, or asking participants to rate how clear the question was | ) | Low methodological clarity: can be the same as cognitive interviewing or quite different |
Translation and back translation | After translating a survey from the origin to the target language, a different translator ‘blindly’ translates the survey back. Differences are then compared and resolved ( ) | Involves bilingual translators whose world view and experience do not match the target population’s, making them unable to comment on the tool’s appropriateness | |
Pilot testing | Enumerators administer the survey to a small group of participants with the same characteristics as the target survey in as close to real world conditions as possible | Focuses on the mechanics of implementation while cognitive testing focuses on the survey questions achieving shared understanding between researcher intent and respondent interpretation |
Survey tool development starts with item generation, which may include a variety of approaches, including in-depth interviews with respondents, review of literature and existing survey tools, and expert review. This is followed by translation, cognitive interviewing, content modification, and then pilot testing ( Figure 1 ).
Situating cognitive interviewing within the larger process of tool development
What is Kilkari? Kilkari is India’s flagship direct-to-beneficiary messaging programme. Pregnant and post-partum women receive one weekly phone call containing a short (1.5 minute) pre-recorded health message on topics including preparing for childbirth, caring for newborns, IYCF, and FP.
Kilkari evaluation: The Kilkari evaluation ( Lefevre et al. , 2019 ) was a randomized controlled trial in rural Madhya Pradesh, India. In 2018, 5095 pregnant women were enrolled and randomized to receive Kilkari or not. An endline survey in 2020, when the study participants were 12–17 months post-partum, assessed whether receiving Kilkari changed women’s knowledge or practice.
Endline Kilkari evaluation survey tool: The draft endline survey included 12 modules to assess study participants’ knowledge and self-reported practice on topics covered by Kilkari, as well as information on socio-economics, decision-making power in the household, interaction with community health workers, exposure to Kilkari, and media consumption patterns. Draft questions were drawn from a mix of tools identified in the literature, including the Demographic and Health Survey and Multiple Indicator Cluster Surveys, and developed by other academic teams.
Greater uptake of cognitive interviewing and explicit description of the process would be a strong contribution to improving the validity of survey research in this field. In this section, we discuss the steps in conducting rigorous cognitive interviews: defining scope, selecting researchers, training, sampling, data collection, and analysis. We draw illustrative examples from our experience with cognitive interviews to refine survey content for the Kilkari evaluation ( Box 2 ).
In an ideal scenario, almost all questions in a data collection tool would be tested. However, time and available resources often limit how much of the survey can be tested. Examining a survey question during a cognitive interview takes far longer than asking the same question during the field survey itself. In a cognitive interview, each survey question must first be asked and answered in a quantitative manner and then discussed in an in-depth qualitative manner through a series of probes to determine how the respondent interpreted the question.
Multiple cognitive interview guides can be developed to examine sub-components of the survey questions. Thus, a cognitive interview guide can be developed to assess one portion of the survey’s questions with one set of participants, while a second interview guide can be developed to assess a different set of questions from the survey with a different set of participants, and so on. But even with multiple cognitive interview guides, researchers will likely still have to prioritize a sub-sample of questions. Selecting which questions to test is a judgement decision that can be guided by focusing on the questions most central to measuring the key outcomes of interest and the questions that are new, conceptually complex, or have never been applied to this respondent population. It is also important to keep blocks of questions (e.g. subject modules) together since they build on and relate to one another. Box 3 presents an illustrative example drawn from our team’s process of defining the scope of cognitive interviews in the Kilkari evaluation.
Priority areas of the tool selected for CI: The draft endline survey tool contained 180 questions and was to take 90 minutes to administer in the field, at about 30 seconds per question. Although we wanted to test each question, it was not feasible to do so. While the time required to test each question in the cognitive interview varies widely, we found it appropriate to allocate each draft survey question at least 3 minutes in the cognitive interview: about 30 seconds to simply ask the question and attempt to record an answer mimicking the survey data collection, and then an additional 2.5 minutes for cognitive probing. Attempting to test each question of a 180-question long survey would require an (impossibly long) nine-hour cognitive interview.
Since the Kilkari evaluation’s priority outcomes were infant and young child feeding and the use of modern contraception, we focused on cognitive interviews for the questions on these topics. We went through the draft survey tool and identified all the questions on these two topics, which were spread across modules on knowledge, practice, decision-making power and discussion. There were approximately 60 questions, which would still take 3 hours to cover in one cognitive interview. We thus decided to split them into two separate sets ( Figure 2 ).
How much of the survey can you test through cognitive interviews?
Even with limiting the number of survey questions in our cognitive testing guide to just 30, many interviews still had to wrap up before completing all 30 questions. Sometimes respondents had to leave early or were distracted. Many times the researchers found that initial questions took longer than anticipated and thus had to end the interview before completing the guide. This was particularly the case with the least educated respondents and the earlier draft version of questions. In these cases, participant comprehension was very low and thus the researchers spent a long time explaining questions to participants and seeking to understand the various ways in which questions failed.
Key learning: You can test far fewer questions in a cognitive interview than you can cover in a survey of comparable duration (usually 1.5 hours). Multiple cognitive interview sets may be required to test all priority survey questions.
Cognitive interviewing requires high-level analytical understanding, linguistic insight, and collaboration among researchers. Researchers must be fully proficient in the language spoken by the target population, as well as the original language of the draft questions, in cases where translation is involved. This proficiency is vital so that each researcher understands the nuances of the original questions and can carefully adapt the phrasing of the questions to ensure local understanding. The effects of similarities and differences in interviewer/participant’s gender, age, sexuality, class, and ethnicity have been considered in the qualitative research methodological literature ( Fontana and Frey, 1994 ; Hsiung, 2010 ). For cognitive interviewing, the same considerations apply, wherein interviewer identity must be considered in light of the topic being studied and the research context ( Hsiung, 2010 ). Ideal researchers are strong qualitative interviewers, so that they can undertake appropriate probing and use verbal and nonverbal approaches to setting respondents at ease and handling respondent discomfort with the potentially unfamiliar process of the cognitive interview. They should also be familiar with quantitative survey research, so that they can understand how quantitative enumerators will administer questions and seek to determine appropriate answers. Box 4 discusses team composition for the Kilkari endline survey cognitive interviews.
Team structure: Our eight-person research team consisted of five female researchers, a male research team manager, a male logistics coordinator and a female team lead. The five researchers were all master’s level social scientists with prior qualitative research experience and training. The researchers worked in pairs, with the fifth and most junior researcher providing backup support. The research manager worked with the logistics coordinator to handle day-to-day logistics, sampling and data management. They also handled sensitive community relationship issues, such as taking curious onlookers (particularly husbands, who were sometimes keen to jump in and answer questions for their wives) away from the cognitive interview. The research lead and the research manager conducted training, developed the research protocol, and ran debriefs.
Profile and selection of researchers: Researchers were fluent in Hindi and English, had prior qualitative experience and where possible, and had worked previously in Madhya Pradesh. Additional experience with and sensitivity to key issues relevant to successful qualitative interviewing in rural India included: awareness of gender and caste power dynamics; knowledge of rural Indian community dynamics including health system dynamics; capacity to effectively probe; and understanding of rapport-building.
Key learnings: Experienced researchers, ideally with an understanding of both quantitative and qualitative data collection, are required and should work in pairs (one to lead the interview and one to take detailed notes). Fluency in the survey’s starting language and target language are vital in cases where the survey is translated.
Training: Training was composed of the following modules:
Overview of the entire Kilkari evaluation;
Overview, objectives and sampling for cognitive testing within the Kilkari evaluation;
Principles of cognitive interviewing;
Findings from earlier cognitive interviewing on other topics to showcase the types of issues identified through this research process and how these cognitive failures were resolved to strengthen another survey tool;
Principles of qualitative interviewing;
Research ethics and consent processes;
Data management, cover sheets, field logistics, and safety;
In-depth lecture and discussion on research topic 1: infant and young child feeding including recommended practices on exclusive breastfeeding, what exclusive breastfeeding means, and recommended practices on complementary feeding;
Question-by-question examination of the cognitive interview guide on infant and young child feeding (IYCF) to ensure that each researcher understood the underlying intent of each question;
Role play to practice cognitive interviewing;
In-depth lecture and discussion on research topic 2: family planning;
Question-by-question examination of the cognitive interview guide on family planning to ensure that each researcher understood the underlying intent of each question.
Modules 9 and 12 required two days each. We examined each survey question to be tested in the field—including the answer options—to thoroughly understand the question’s intent, assess the Hindi translation, and hypothesize potential areas of confusion that could arise. We also examined the pre-developed cognitive probes for each survey question and edited them where necessary (more on this in Box 7 ). We furthermore discussed how we could handle potential participant reactions to the questions.
We covered Modules 1–10 in the first week-long training. We then conducted our data collection on IYCF. Only once that data collection process was complete did we proceed to Modules 11 and 12. In separating exposure to the two different cognitive interview topics and guides, we ensured that the team was immersed in and focused on only one area at a time, and did not forget the family planning content while working on IYCF. While in many cases piloting is indicated, the richness of initial data exceeded expectations and the resulting interviews were included in final analyses.
Key learnings: Some researchers initially struggle to understand the intent of cognitive interviewing and struggle to wear two hats; they must toggle between ‘quantitative enumerator mode’ wherein they read the question as it is written and attempt to elicit a response as if conducting a quantitative survey, and ‘qualitative interviewer mode’ wherein they explore what the respondent was thinking about and draw from narrative explanations to access memories or opinions. Role play during training and close observation of the interviews is necessary to ensure the research team truly understands the intent of cognitive interviews and are capable of implementing this data collection methodology.
Participants in cognitive interviews must be drawn from the same profile as the intended survey respondents. It is difficult to predict how many participants will need to be sampled in order to capture all the cognitive failures with the survey questions. A relatively small number of well-conducted cognitive interviews can yield an enormous amount of rich information, particularly when there is a large cultural and linguistic gap between the researchers and respondent population. Our research has found reasonable evidence of saturation with a total of 20–25 participants over three rounds, which broadly aligns with recommendations from high-income countries, such as the US Center for Disease Control and Prevention’s guidance of 20–50 respondents ( CDC/National Center for Health Statistics, 2014 ) and lead American cognitive interviewing methodologist Gordon Willis’s 8–12 subjects per round, multiplied by 1–3 rounds ( Willis, 2014 ). Direct comparison of sample sizes across studies is often inappropriate because they involve testing different numbers of questions over different numbers of rounds, with varying types of respondent groups, and some involve translation while others do not. Published literature highlights a range of sample sizes, including 10 urban Greenlandic residents for a sexual health survey available in three languages ( Gesink et al. , 2010 ); 15 people per language group per round (four languages, two rounds and total of 120 interviews) for a women’s empowerment in agriculture survey in Uganda ( Malapit, Sproule and Kovarik , 2016 ); 24 people across seven ethnic groups for a mental health survey in the UK without translation ( Care Quality Commission, 2019 ); 34 people stratified for age, gender, education level and location in rural Bangladesh for assessing the cultural suitability of a World Health Organization (WHO) quality of life assessment ( Zeldenryk et al. , 2013 ); 20 women and 20 men to improve a healthcare report card in Tajikistan ( Bauhoff et al. , 2017 ); and 49 women in rural Ethiopia to assess the resonance of a WHO question on early initiation of breastfeeding ( Salasibew et al. , 2014 ). The following three suggestions can help guide sampling in an efficient and rigorous manner.
First, focus on sampling participants from within the survey’s target population who are most likely to struggle with the survey; these are usually the least educated and most marginalized people ( Box 6 ). Interviewing these participants will reveal the weaknesses of the survey most rapidly and create the opportunity to adapt the survey to be comprehensible to the entire target population. While some researchers recommend engaging a range of respondents ( Willis, 2005 ), our experience in rural India found that participants with higher than average education and exposure were far less useful in identifying issues than participants with the lowest education and exposure.
Cognitive interview population: Since the endline survey would be conducted among the 5095 women enrolled in our evaluation 1–1.5 years after they gave birth, our cognitive interviews were conducted among participants with a similar profile: rural women in Madhya Pradesh who had access to mobile phone and who are mothers to a child between 12 and 17 months in age. Within this profile, we skewed our sample towards women with low levels of education, from marginalized castes, and in lower socio-economic strata.
Sample: We conducted two sets of cognitive interviews: one on IYCF ( n = 21) and one on family planning (FP) ( n = 24) ( Table 3 ). Each set required three rounds of interviews, with the questions revised twice: after Round 1 and after Round 2. Round 1 and Round 2 involved a higher number of respondents because we sought sufficient data to understand and document the range of cognitive failures associated with the draft survey questions. Round 3 required fewer respondents because by that stage we were generally confirming that the questions were working as intended. Concentrating on lower literacy and marginalized women was highly efficient at exposing problems with the draft questions and enabling us to reduce the number of interviews necessary. With family planning, we set out to test a portion of questions that were to be asked only of women who had become pregnant in the year since the birth of their previous child. However, this event was relatively rare and we considered it inappropriate to screen women at enrolment for this event, so we oversampled in hopes of including at least a few women who would not skip out of this portion of questions.
Sample of participants for cognitive interviewing in Kilkari
Topic of survey questions . | Round 1 . | Round 2 . | Round 3 . | Total . |
---|---|---|---|---|
Set 1: IYCF | 7 | 8 | 6 | 21 |
Set 2: FP | 13 | 6 | 5 | 24 |
Topic of survey questions . | Round 1 . | Round 2 . | Round 3 . | Total . |
---|---|---|---|---|
Set 1: IYCF | 7 | 8 | 6 | 21 |
Set 2: FP | 13 | 6 | 5 | 24 |
Key learnings: We found that as few as six respondents per round was sufficient to expose cognitive failures and enable revision for the next iteration. Focus on sampling the respondents who are most likely to struggle to comprehend the survey questions—generally those within your target population who have the lowest exposure or education. You can probably conduct an unlimited number of cognitive interviews and continue identifying small potential improvements. However, returns diminish—we found three rounds to be sufficient.
Second, think about sampling in terms of iterations. Cognitive interviewing requires an initial round of interviews using the first version of the survey. Then, the team will revise the survey based on detailed debriefing and take version two to the field for another round of cognitive interviews. Additional rounds of revision are required until the survey achieves cognitive match between researcher intent and participant comprehension ( Figure 1 ). The number of respondents required may reduce from iteration to iteration as the survey questions become increasingly more appropriate to the local context.
Third, aim to include participants whose experiences exhaust the domains covered by the survey ( Beatty and Willis, 2007 ). If all of your participants skip out of a certain section of the survey, you will not be able to test the questions in this section. Ideally, your recruitment strategy can pre-identify people who will complete specific sections of the survey. However, if it is difficult to pre-identify respondents who have experienced specific domain of interest that have low prevalence, you will have to increase your sample size.
Cognitive interviewing begins with first asking the original survey question exactly as it is written and recording the respondent’s answer using the original response options. The interview then proceeds by eliciting feedback from the respondent to understand how they interpreted the question and why they gave the response provided. Two main approaches have been used for eliciting this feedback: (1) probing and (2) ‘think aloud’ ( Table 4 ).
Approaches to eliciting feedback in cognitive interviews
Approach . | Description . | Benefits . | Drawbacks . |
---|---|---|---|
Think aloud | Participant talks through their mental processes and memory retrieval as they interpret questions and formulate answers ( , 2007) | ( ) | ( ; , 2013) |
Probing | : Researcher uses pre-developed questions to interview participant about their interpretation of the question, such as ‘what does [word] mean to you?’ or ‘why did you say [answer]?’ ( ) : Research formulates probes during the cognitive interview to explore emergent issues, such as ‘Earlier you said you had never done [activity] but now you said you had completed [sub-activity]. Why did you say that?’ ( ; , 2013) | ( , 2013) |
Approach . | Description . | Benefits . | Drawbacks . |
---|---|---|---|
Think aloud | Participant talks through their mental processes and memory retrieval as they interpret questions and formulate answers ( , 2007) | ( ) | ( ; , 2013) |
Probing | : Researcher uses pre-developed questions to interview participant about their interpretation of the question, such as ‘what does [word] mean to you?’ or ‘why did you say [answer]?’ ( ) : Research formulates probes during the cognitive interview to explore emergent issues, such as ‘Earlier you said you had never done [activity] but now you said you had completed [sub-activity]. Why did you say that?’ ( ; , 2013) | ( , 2013) |
Probing requires that enumerators use a combination of scripted and unscripted prompts to guide the directionality of the interview, while ‘think aloud’ asks the respondent to verbalize their thoughts while interpreting the question and formulating their response ( Willis, 2005 ). While some researchers find success incorporating the think aloud approach ( Ogaji et al. , 2017 ), probing is gaining consensus as the ideal method ( Willis, 2014 ) and is particularly appropriate in global public health when working with respondents who find it easier to answer questions than verbalize their thought process ( Zeldenryk et al. , 2013 ). Even with probing, careful effort must be made to manage the shortcomings of this method. Researchers must provide clear explanation that the exercise seeks participant feedback on the questions and that any confusion or incomprehension is entirely the research team’s fault, not the participant’s. Box 7 provides an example cognitive interview question and reflections on the data collection process.
Data collection tool: The data collection tool consisted of each draft survey question (and its answer options) followed by suggested scripted probes and comments about areas to explore. Researchers were also strongly encouraged to use emergent probes based on specific information that arose during their interview. Figure 3 provides an example question from the IYCF cognitive interview guide.
Example question from the IYCF cognitive interview guide
Data collection experience: We found that some participants struggled to understand the purpose of the interview and felt embarrassed or annoyed by the probes. Some researchers initially struggled to move beyond the scripted probes, at times failing to probe on issues that arose and demanded attention. During debriefs, we dissected each question and subsequent probing and emphasized the importance of unscripted probing to better understand respondent’s cognitive processes. For instance, if a respondent said that a baby should be breastfed 3 hours after birth, and we knew from other questions in the survey that the respondent had a caesarean delivery, the interviewer had to probe to determine whether the participant experienced delayed breastfeeding due to her post-operation recovery and whether her reply on when babies should breastfeed was actually her description of when her baby was breastfed. For another example, if a participant said she had not heard of condoms when directly asked in the knowledge section and then later said that she had used condoms when asked about her use of birth control, the researcher had to have the insight to circle back to the initial knowledge question about condoms to determine if the initial response was driven by low comprehension (maybe we used an unfamiliar word for condom) or discomfort (shyness) or another factor.
Key learnings: Researchers need deep familiarity with the tool and subject matter to successfully combine scripted and emergent probing. Rapport must be developed with the research participants and they must be regularly reassured and reminded that any confusion caused by the survey questions is the research team’s ‘fault’ and not theirs.
Data collection ideally requires two researchers to conduct every interview: one to lead the questioning and one to take responsibility for notetaking and to support the questioning. While in some qualitative interviews junior researchers can serve as appropriate notetakers, for cognitive interviewing experienced, trained researchers should perform this role. Notetaking is essential in cognitive interviewing because debriefs and rapid revision of survey questions depend on detailed notes being available for analysis after the interview. Notetakers must be trained to record the various forms of cognitive failure that occur during the interview as well as non-verbal feedback from the respondent. While recordings of the interviews can also be reviewed during the debriefs, reviewing the whole audio file or generating and reading a complete transcript requires far more time than is available in the field. High-quality notes are the backbone of the rapid field-based analysis, discussed next, that enables production of revised survey questions.
To ensure the robustness of findings, cognitive interviewing requires alternating data collection with team debriefing sessions from which revisions to the tool may emerge. This process is iterative with resulting changes to language and tool content requiring additional testing until the final round of cognitive interviews showcase high comprehension amongst respondents.
Box 8 showcases the field work approach from the Kilkari endline survey cognitive interviewing, which found that 1.5 days of debriefing time was required for every 7–8 interviews conducted. For every survey question being tested, researchers will need to budget at least 3 minutes during the interview and an hour during the debrief to discuss what each respondent said for that question and consider how to improve the question. Simple questions that work well take the least amount of time—perhaps less than 3 minutes during the interview to establish that the respondent interpreted the question as expected and another 10 minutes during the debrief to clarify that all interviews had similar success. Questions with translation problems demand more time. Questions with deeper conceptual issues, such as entire concepts failing to resonate with respondent worldview, take the most time. The schedule for data collection and the steps used in debriefing are discussed in Box 8 .
Phase 1. Revising the survey tool: We aimed to conduct approximately six cognitive interviews in one day of fieldwork. After collecting data through about eight interviews on the first version of the instrument, we required 1.5 days for debrief and revision before returning to the field to test the second iteration of the tool ( Table 5 ).
Illustrative schedule of 1 month of cognitive interview field work
Day 1 . | Day 2 . | Day 3 . | Day 4 . | Day 5 . | Day 6 . | Day 7 . |
---|---|---|---|---|---|---|
Training, including topic lecture and discussion on IYCF and detailed review of the IYCF cognitive interview guide | Break | |||||
5 cognitive interviews (CIs) on IYCF version 1 | Morning: 3 CIs on IYCF version 1 Afternoon: Debrief | Continue debrief and revise survey questions, create IYCF version 2 | 5 CIs on IYCF version 2 | Morning: 2 CIs on IYCF version 2 Afternoon: Debrief | Continue debrief and revise survey questions, create IYCF version 3 | Break |
6 CIs on IYCF version 3 | Last debrief and revisions to create final version of IYCF questions | Topic lecture and discussion on FP and detailed review of the FP cognitive interview guide | 6 CIs on FP version 1 | 7 CIs on FP version 1 | Break | |
Debrief and revise survey questions | Additional debrief and revision, create FP version 2 | 6 CIs on FP version 2 | Debrief and revise survey questions, create FP version 3 | 5 CIs on FP version 3 | Last debrief and revisions to create final version of FP questions | Break |
Day 1 . | Day 2 . | Day 3 . | Day 4 . | Day 5 . | Day 6 . | Day 7 . |
---|---|---|---|---|---|---|
Training, including topic lecture and discussion on IYCF and detailed review of the IYCF cognitive interview guide | Break | |||||
5 cognitive interviews (CIs) on IYCF version 1 | Morning: 3 CIs on IYCF version 1 Afternoon: Debrief | Continue debrief and revise survey questions, create IYCF version 2 | 5 CIs on IYCF version 2 | Morning: 2 CIs on IYCF version 2 Afternoon: Debrief | Continue debrief and revise survey questions, create IYCF version 3 | Break |
6 CIs on IYCF version 3 | Last debrief and revisions to create final version of IYCF questions | Topic lecture and discussion on FP and detailed review of the FP cognitive interview guide | 6 CIs on FP version 1 | 7 CIs on FP version 1 | Break | |
Debrief and revise survey questions | Additional debrief and revision, create FP version 2 | 6 CIs on FP version 2 | Debrief and revise survey questions, create FP version 3 | 5 CIs on FP version 3 | Last debrief and revisions to create final version of FP questions | Break |
Components of debriefs:
All researchers who conducted and took notes during the cognitive interviews attend, with the lead researcher serving as facilitator.
The debrief begins with collecting consent forms, uploading of audio recordings, and entering basic data information (assigning each interview a unique ID and documenting the duration, location, respondent age, etc.) into a data management system.
A spreadsheet is then used to document and systematize the debrief. Our team found success using rows to list the quantitative survey questions and response options and columns to list each interview by unique ID.
The team proceeds survey question by survey question. For each survey question, the researchers draw from the interview notes and occasionally from reviewing the audio recordings to document and discuss how the participant(s) that they interviewed responded to the question, what answer they provided (if any), and what additional information the cognitive probing elicited. The researchers and/or facilitator write extensive notes in the spreadsheet for each question to summarize what was said for each participant.
Revisions—rewording, rewriting, reordering, or removing questions—are then made by the entire team to attempt to resolve issues. These revisions are documented in the spreadsheet by adding a new column next to the original question.
After each question has been discussed, each interview participant’s responses have been shared, and each revision has been formulated in the spreadsheet, a revised cognitive interview guide is developed, with updated probes as needed.
The debrief closes with a discussion of research challenges, logistical considerations for the next day’s fieldwork, and participant sampling.
Phase 2. Preparation of peer review manuscripts (optional)
Often the end goal of cognitive interviewing is the production of a revised survey instrument that is valid and locally grounded. If this is the case, by the end of the final debrief, the team is finished. However, in cases where additional dissemination of the results of the cognitive interviewing is warranted, the cognitive interviews should be transcribed for analysis. When necessary, they also have to be translated, which demands extreme care. Significant portions of text in the original language must be retained to capture nuance in meaning around vocabulary words. The researchers should themselves carry out or at least check the translations. Thematic analysis can then be used to classify the text segments in the transcripts according to the cognitive failures exemplified.
Key learnings: Each cognitive interview generates an enormous amount of data that must be documented and used for subsequent revisions in the field. Cognitive interviewing data collection must be balanced with extensive time allocated for debriefing and revision.
Once the survey has achieved strong performance in the target population, cognitive interviewing is complete. The larger survey enumeration team will be formed and trained. It is useful to send some or all of the cognitive interview researchers to support the quantitative enumerator training and pilot testing. The cognitive interviewers can explain to the quantitative survey team why questions are worded and ordered the way they are and how to handle the types of responses that may arise in the population.
The devil is in the details when it comes to cognitive interviewing—in terms of both the quantitative survey that this method hones and the cognitive interviewing method itself. Cognitive interviewing focuses on getting each detail of a survey question right to ensure valid data collection. Each word chosen, the exact syntax used, the response options provided (yes/no, Likert, etc.), the addition or removal of examples, the question styles selected (such as hypotheticals and true/false statements), and the resonance of the underlying constructs being assessed will all influence the alignment of researcher intent and respondent interpretation and response. The application of cognitive interviewing also demands careful attention to detail. Researchers must allocate adequate time and attention to each stage of the process and must ground difficult decisions in strong methodological logic. Cognitive interviewing demands tough decisions on which questions to test to ensure that the scope of the exercise is appropriate. The research team must be carefully selected and trained so that they can set respondents at ease and probe effectively to identify and document cognitive failures in real time. Research participants most likely to yield rich data must be sampled. The cycles of interviews, debriefing, analysis, and revision must be structured, meticulous and well documented.
As the methodology of cognitive interviewing continues to evolve in this field, the recommendations in Table 6 to ensure quality can help develop standards for research rigour.
Recommendations to ensure the quality of cognitive interviews
Component . | Recommendations . | Rationale for recommendations . |
---|---|---|
Scope of survey tool tested | ||
Developing the cognitive interview guide | ||
Recruiting and training researchers | ||
Participant sample characteristics | ||
Conducting interviews | ||
Debrief and analysis | ||
Supporting quantitative survey enumerator training |
Component . | Recommendations . | Rationale for recommendations . |
---|---|---|
Scope of survey tool tested | ||
Developing the cognitive interview guide | ||
Recruiting and training researchers | ||
Participant sample characteristics | ||
Conducting interviews | ||
Debrief and analysis | ||
Supporting quantitative survey enumerator training |
Survey research is fundamental to shaping our understanding of health systems. Surveys may aim to measure a range of outcomes, including population health, practices, care-seeking, attitudes towards services, and knowledge on health issues and topics. Researchers, practitioners, and policymakers rely on survey data to assess the scope of health or health system problems, prioritize the distribution of resources, and evaluate the effectiveness of programmes and interventions. It is thus crucial that researchers ensure that survey instruments are valid, i.e. that they truly measure what they intend to measure. Cognitive interviewing must be recognized as a fundamental validation step in survey development ( Beatty and Willis, 2007 ; Willis and Artino, 2013 ), alongside literature review, expert consultation, and drawing from previously developed survey tools ( Sullivan, 2011 ).
Ultimately, the need for cognitive interviewing in global public health arises from a gap, whether linguistic, cultural, or socioeconomic, between researchers and respondents. The greater this gap, the more space there is for cognitive mismatch to occur, leading to invalid research findings, and the more important cognitive interviewing will be in reducing this divergence. At the low-risk end of the spectrum are those with the smallest gap, such as surveys that are developed, administered, and analysed by the same population that completes them, as may occur during participatory action research. At the high-risk end of the spectrum with the greatest gap between researchers and respondents are surveys developed by researchers in one setting and adapted by another research group for use in a different setting, such as surveys developed in English in Western contexts and then translated to other languages and administered in non-Western settings. Cognitive interviewing may identify the most egregious survey question failures in the latter instance but can also identify surprising gaps between researcher intent and respondent interpretation even in the former.
The data underlying this article will be shared on reasonable request to the corresponding author.
This work was funded by the Bill and Melinda Gates Foundation (OPP1179252).
This work was made possible by the Bill and Melinda Gates Foundation, and we particularly acknowledge the support of the following people there: Diva Dhar, Suhel Bidani, Rahul Mullick, Suneeta Krishnan, Neeta Goel and Priya Nanda. We futher thank BBC Media Action, Dan Harder at the Creativity Club for edits to the overarching framework, and the qualitative researchers Manjula Sharma, Dipanwita Gharai, Bibha Mishra, Namrata Choudhury, Aashaka Shinde, Shalini Yadav, Anushree Jairath, and Nikita Purty.
The examples used in this methodological musing article were drawn from research that received ethical approval at Sigma, Delhi, India (10041/IRB/D/17-18) and Johns Hopkins University, Baltimore, USA (00008360).
The authors declare that they have no conflict of interest.
Agency for Healthcare Research and Quality in the Department of Health and Human Services . 2019 . Conducting Cognitive Testing for Health Care Quality Reports . https://www.ahrq.gov/talkingquality/resources/cognitive/index.html , accessed 1 April 2021 .
Bauhoff S , Rabinovich L , Mayer LA . 2017 . Developing citizen report cards for primary health care in low and middle-income countries: results from cognitive interviews in rural Tajikistan . PLoS One 12 : 1 – 14 .doi: 10.1371/journal.pone.0186745 .
Google Scholar
Beatty PC , Willis GB . 2007 . Research synthesis: the practice of cognitive interviewing . Public Opinion Quarterly 71 : 287 – 311 .doi: 10.1093/poq/nfm006 .
Care Quality Commission . 2019 . 2019 Community Mental Health Survey: Quality and Methodology Report . pp. 1 – 20 . https://www.nhssurveys.org , accessed 15 March 2021 .
CDC/National Center for Health Statistics . 2014 . National Center for Health Statistics: Cognitive Interviewing . https://www.cdc.gov/nchs/ccqder/evaluation/CognitiveInterviewing.htm , accessed 1 April 2021 .
Drennan J . 2003 . Cognitive interviewing: verbal data in the design and pretesting of questionnaires . Journal of Advanced Nursing 42 : 57 – 63 .doi: 10.1046/j.1365-2648.2003.02579.x .
Fontana A Frey JH . 1994 . Interviewing: the art of science. In: Denzin NK , Lincoln YS (eds). The Handbook of Qualitative Research . Thousand Oaks, CA: SAGE , pp. 361 – 76 .
Google Preview
Gesink D , Rink E , Montgomery-Andersen R , Mulvad G , Koch A . 2010 . Developing a culturally competent and socially relevant sexual health survey with an urban arctic community . International Journal of Circumpolar Health 69 : 25 – 37 .doi: 10.3402/ijch.v69i1.17423 .
Hall J , Barrett G , Mbwana N et al. 2013 . Understanding pregnancy planning in a low-income country setting: validation of the London measure of unplanned pregnancy in Malawi . BMC Pregnancy and Childbirth 13 : 1.doi: 10.1186/1471-2393-13-200 .
Hsiung P-C . 2010 . Reflexivity: The Research Relationship, Lives and Legacies: A Guide to Qualitative Interviewing . https://www.utsc.utoronto.ca/~pchsiung/LAL/reflexivity/relationship , accessed 1 April 2021 .
Knafl K , Deatrick J , Gallo A et al. 2007 . The analysis and interpretation of cognitive interviews for instrument development . Research in Nursing and Health 30 : 224 – 34 .doi: 10.1002/nur.20195 .
Lefevre A , Agarwal S , Chamberlain S et al. 2019 . Are stage-based health information messages effective and good value for money in improving maternal newborn and child health outcomes in India? Protocol for an individually randomized controlled trial . Trials 20 : 1 – 12 .doi: 10.1186/s13063-019-3369-5 .
Malapit H , Sproule K , Kovarik C . 2016 . Using Cognitive Interviewing to Improve The Women’s Empowerment in Agriculture Index Survey Instruments. Evidence from Bangladesh and Uganda . Poverty, Health and Nutrition Division, International Food Policy Research Institute .
Miller K . 2003 . Conducting Cognitive Interviews to Understand Question-Response Limitations . American Journal of Health Behavior 27 : S264 – 72 .
Ogaji DS , Giles S , Daker-White G , Bower P . 2017 . Development and validation of the patient evaluation scale (PES) for primary health care in Nigeria . Primary Health Care Research and Development 18 : 161 – 82 .doi: 10.1017/S1463423616000244 .
Ruel E , Wagner WE , Gillespie BJ . 2016 . The Practice of Survey Research: Theory and Applications. Thousand Oaks, CA: Sage Publications .
Salasibew MM , Filteau S , Marchant T . 2014 . Measurement of breastfeeding initiation: Ethiopian mothers’ perception about survey questions assessing early initiation of breastfeeding . International Breastfeeding Journal 9 : 1 – 8 .doi: 10.1186/1746-4358-9-13 .
Schuler SR , Lenzi R , Yount KM . 2011 . Justification of intimate partner violence in rural Bangladesh: what survey questions fail to capture . Studies in Family Planning 42 : 21 – 8 .
Scott K , Gharai D , Sharma M et al. 2019 . Yes, no, maybe so: the importance of cognitive interviewing to enhance structured surveys on respectful maternity care in northern India . Health Policy and Planning 35 : 67 – 77 .doi: 10.1093/heapol/czz141 .
Sullivan GM . 2011 . A primer on the validity of assessment instruments . Journal of Graduate Medical Education 3 : 119 – 20 .doi: 10.4300/jgme-d-11-00075.1 .
Weeks A , Swerissen H , Belfrage J . 2007 . Issues, Challenges, and Solutions in Translating Study Instruments . Evaluation Review 31 : 153 – 65 .
Willis G . 2014 . Pretesting of health survey questionnaires: cognitive interviewing, usability testing, and behavior coding. In: Johnson TP . (ed). Handbook of Health Survey Methods , First Edition, New York : John Wiley & Sons , Inc. pp. 217 – 42 .doi: 10.1002/9781118594629.ch9 .
Willis GB . 2005 . Cognitive Interviewing: A Tool for Improving Questionnaire Design . Thousand Oaks, CA : SAGE .
Willis GB , Artino AR . 2013 . What do our respondents think we’re asking? Using cognitive interviewing to improve medical education surveys . Journal of Graduate Medical Education 5 : 353 – 6 .doi: 10.4300/jgme-d-13-00154.1 .
Zeldenryk L , Gordon S , Gray M et al. 2013 . Cognitive testing of the WHOQOL-BREF Bangladesh tool in a northern rural Bangladeshi population with lymphatic filariasis . Quality of Life Research 22 : 1917 – 26 .doi: 10.1007/s11136-012-0333-1 .
Month: | Total Views: |
---|---|
May 2021 | 691 |
June 2021 | 328 |
July 2021 | 205 |
August 2021 | 150 |
September 2021 | 91 |
October 2021 | 257 |
November 2021 | 88 |
December 2021 | 74 |
January 2022 | 51 |
February 2022 | 80 |
March 2022 | 87 |
April 2022 | 113 |
May 2022 | 64 |
June 2022 | 123 |
July 2022 | 53 |
August 2022 | 52 |
September 2022 | 76 |
October 2022 | 97 |
November 2022 | 64 |
December 2022 | 63 |
January 2023 | 49 |
February 2023 | 97 |
March 2023 | 131 |
April 2023 | 112 |
May 2023 | 112 |
June 2023 | 81 |
July 2023 | 82 |
August 2023 | 99 |
September 2023 | 106 |
October 2023 | 108 |
November 2023 | 106 |
December 2023 | 86 |
January 2024 | 97 |
February 2024 | 116 |
March 2024 | 113 |
April 2024 | 125 |
May 2024 | 161 |
June 2024 | 127 |
July 2024 | 126 |
August 2024 | 46 |
Citing articles via.
Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide
Sign In or Create an Account
This PDF is available to Subscribers Only
For full access to this pdf, sign in to an existing account, or purchase an annual subscription.
Presenter(s): olivia sexton, sophie pilley, jo d’ardenne and richard bull.
This resource introduces the cognitive interviewing method, a qualitative method which explores an individual’s thought processes when presented with a task or information. Cognitive interviews traditionally involve administering a survey questionnaire whilst collecting additional verbal and non-verbal information about how individual’s respond to the question. The types of verbal and non-verbal information which are typically collected during a cognitive interview are participant observations, any naturalistic verbalisations from participants, and responses to semi-structured interviews. This additional verbal information is used to evaluate whether the question is measuring what the researcher intends it to. Cognitive interviewing is traditionally used in social research to pre-test survey questions.
We will explain:
Cognitive interviewing is a qualitative research method used to explore how individuals perceive stimuli and information. Typically, when we design surveys and questionnaires, we design them to be used in quantitative research projects. When designing surveys and questionnaires, we want the questions that we use to be able to generate good quality data, in that our data is valid, reliable, sensitive, unbiased and complete.
To be able to check that our questions will likely generate good quality data, we can use the cognitive interviewing method. Traditionally this technique has been used when pre-testing questions which are used in quantitative data collection.
As a research method, the cognitive interviewing technique is used within qualitative interviews to explore how individuals comprehend and judge questions, how they retrieve the information to answer a question and ultimately how they respond to the question.
Cognitive interviews typically occur in a one-to-one setting between an interviewer and a participant. During these interviews, the participant is asked survey questions, but the focus is on the mental processes used to come up with an answer.
There are four key elements of a cognitive interview.
The first key element is administering the survey question. When interviewers administer the survey question, it should be asked in a format which is as close as possible to how participants would complete this in the real survey context. For example, if the survey is interviewer-administered, the interviewer should read the questions aloud to the participant.
The second key element is participant observation. During a cognitive interview, interviewers should observe any non-verbal signs from participants when they are thinking about or answering the questions. For example, if you administer a question and a participant appears to hesitate, this should be noted.
The third key element is the think-aloud technique. This technique developed by Ericsson and Simon (1980) encourages participants to verbalise their thought processes as they answer the questions. As this is often not the easiest thing for participants to do, we recommend training participants to think aloud, by first demonstrating the technique yourself.
The final key element of the cognitive interview is interviewer probing, which is where interviewers, following an interview protocol, ask semi-structured questions. During our cognitive interviews, this is where we get the majority of our information about how the questions are working. There are different ways in which the probing questions can be asked. For example, they can be asked at the same time as the question being administered or can be asked at the end of the survey.
The most important part of conducting a cognitive interview as an interviewer is to not bias the participant. This means that you should avoid trying to add clarifications to questions unless there are specific interview instructions. Remember, it is important to know which questions are working well, and which questions may need amending to help with understanding.
> Download example probe sheet here.
Whilst cognitive interviewing is traditionally used to pre-test survey questions, it does have other uses as a research method. As the cognitive interview ultimately explores how individual’s respond to stimuli, the stimulus does not always have to be survey questions.
Cognitive interviewing can be used in a variety of other ways. It can be used to test advance leaflets and letters (for example informational leaflets and invitational letters), in combination with user testing of digital devices and forms and to test permission forms for data linkage.
> Download case studies document.
Cognitive interviewing adds something to the survey development process: it tells us something about what goes on in respondents’ minds when asked the survey question or to complete a questionnaire. It is particularly good at identifying problems with comprehension that relate to the ‘reasonableness’ of the task, the way in which the request is structured (syntax) and the words used. As such it is a useful tool to allow us to examine whether the question achieves construct and face validity. It can also identify other problems, related to recall, judgement, response, visual layout and clarity of objectives.
However, as with any technique, there are boundaries to using this technique.
Strengths | Limitations |
Cognitive interviewing adds something to the questionnaire development process: it tells us something about what goes on in respondents’ minds when asked the survey question or to complete a questionnaire |
Cognitive testing samples are generally small. The purpose is not statistical estimation but rather probing around potential issues. Cognitive testing sample sizes, therefore, are not large enough to supply precision in statistical estimates around the size or extent of issues observed. |
Cognitive testing is a useful tool which allows researchers to examine whether the question or materials being tested achieve construct and face validity. In other words, that it does what it is supposed to do. |
Because cognitive testing can’t always reach data saturation (due to budget and time constraints), or the point at which no new information or themes are observed in the data, it is unable to determine for certain that all problems have been mapped. |
Cognitive testing allows unforeseen issues to be identified by encouraging participants to discuss how they found using the questions or materials. |
Cognitive testing is unable to say definitively that any problems encountered are ‘real’. It may be that participant behaviour during cognitive testing does not fully replicate a real world setting, and as such they are more likely to mention ‘issues’ which under normal conditions would not be identified as such. |
Comprehension problems result when questionnaire designer and respondent’s interpretation don’t match. Cognitive testing is particularly good at identifying problems with comprehension that relate to the ‘reasonableness’ of the task, the way in which the request is structured (syntax), and the words used. |
Cognitive testing relies on the sample having been designed well: that is that it is large enough and including a range of different types of respondent so that all questions are sufficiently tested. When test questionnaires are heavily filtered (routed), there is the risk that some questions may only be asked to a small number of participants. |
Cognitive testing is able to identify if participants experience any difficulty in recalling events from a specific period by probing around they found doing this and how confident they are that their response is accurate. |
Cognitive testing is not the same as a real world survey. It is that the problems identified during cognitive testing would occur in the survey itself, rather than being an artefact of the cognitive interview process. |
Cognitive testing is useful in uncovering areas where potentially unreliable data may be collected. This could include when participants provide estimates or ‘short cuts’ to a question when a higher degree of accuracy is intended. Participants may give answers in an incorrect format, refuse to answer certain questions, or may be unable to answer the way they wish using the answer options provided. |
Whilst it is hoped that recommendations by way of changes to questions will lead to improvements in data quality, cognitive testing does not usually offer the opportunity to test these recommendations. As such, it is difficult to determine how successful these recommendations are or what other problems they may cause. |
Cognitive testing is able to identify if participants experience any difficulty in selecting a response option, and if so, if there are other response options that should potentially be included in the question. |
Dr. Olivia Sexton is Senior Researcher at the National Centre for Social Research (NatCen). NatCen is a registered charity and the largest independent and not-for-profit social research organisation in the UK. The centre is driving innovation in survey methods and methodology, providing a variety of research solutions to help clients understand their stakeholders better.
Primary author profile page
Olivia Sexton, Sophie Pilley, Jo d’Ardenne and Richard Bull. (2023). Cognitive Interviewing and what it can be used for . National Centre for Research Methods online learning resource. Available at https://www.ncrm.ac.uk/resources/online/all/?id=20816 [accessed: 15 August 2024]
⌃ BACK TO TOP
Shop New Arrivals
Rolex Datejust 116333 41MM Steel & 18k Yellow Gold, Oyster Black Arabic Dial, Rolex Box $ 11,495
Rolex Day-Date 18238 36MM 18k Yellow Gold, President Band Champagne Tapestry Dial (1994) $ 17,995
Rolex Submariner 116613 40MM Steel & 18k Gold, Oyster Band Black Chromalight Dial, B&P (2020) $ 14,995
Rolex GMT-Master 126710 40MM Red & Blue Pepsi Bezel Stainless Steel Jubilee, B&P (2022) $ 22,195
Join the product waitlist & get first dibs.
The Rolex Yacht-Master is growing in popularity among men and women. The addition of a mid-size variation, such as the reference 68623 that is offered here today, features a 35mm Oyster case and matching Oyster bracelet and Oysterlock clasp, creating the perfect size and weight for men and women. A handsome two-tone finish combining stainless steel and 18 karat yellow gold creates the perfect accessory for any occasion, whether you're setting sail in a Regatta or heading into the boardroom. A sophisticated white dial is set with highly visible black hour markers and luminous Mercedes hands. The automatically changing date display is ever-present at the 3-hour marker along with a magnifying Cyclops lens for easy viewing. Waterproof up to 330 feet, the Mid-Size Rolex Yacht-Master can accompany you virtually anywhere.
The watch is beautiful. The service was great and I received the watch within a couple days. The watch was adjusted to fit perfectly.
Absolutely a wonderful experience from start to finish!!
The reviews what people are saying.
Overall great experience and Bob’s Watches made it a seamless, simple process. I sold a Rolex and worked with Max who was excellent to work with. Would consider Bob’s Watches again for sale or purchase of Rolex.
Never had purchased a pre-owned watch until Bob’s. My experience was outstanding in every way. Will use Bob’s again for my next purchase.
3rd Purchase. Once again, it was fast and friendly. 100% positive. Thank you! Jason S. Chicago Incredibly easy to work with, and Larry went above and beyond and making an easy transaction. Would highly recommend!
Do you have questions about our Steel and Gold Rolex 68623 for sale? Our team of friendly watch experts is here to assist you. Whether you're curious about its features, authenticity, or purchasing process, we're just a call or email away. Don't hesitate to reach out; your perfect Blue Rolex Yacht-Master watch is just a conversation away.
Choosing Bob's Watches for your luxury watch purchase ensures unparalleled authenticity and quality. Our stringent authentication process, performed by a team of experts, guarantees that every watch is genuine and original. We offer certified pre-owned watches that meet our high standards for condition and functionality, giving you peace of mind with your Rolex investment. Additionally, our in-house professional servicing ensures that each timepiece is meticulously maintained and performs at its best. Finally, our comprehensive warranty provides long-term support, covering parts and labor.
Rolex watches are synonymous with luxury, style, and precision. These iconic timepieces are crafted using only the highest quality materials, including precious metals and gemstones. Rolex watches are designed to withstand the test of time, with intricate movements and advanced technology ensuring unparalleled accuracy and reliability. As a status symbol of success and sophistication, Rolex watches continue to be highly sought after by collectors and enthusiasts worldwide. Learn more by reading our Rolex watches ultimate guide and discover the finest collection of Rolex for sale, embodying luxury and precision.
We are proud to present a collection of authentic, unworn or pre-owned luxury watches that have been rigorously verified for authenticity. Each timepiece is subject to an extensive authentication procedure, ensuring that every purchase is made with assurance and worn with prestige. We guarantee a shopping experience free from counterfeits.
IMAGES
VIDEO
COMMENTS
Find low prices for 20 Rolex ref. 68623 watches on Chrono24. Compare deals and buy a ref. 68623 watch.
The Rolex Yacht-Master 68623 has a last known retail price of $6,100 and trades for $7,220 on the pre-owned market. Expect to pay 6% more when buying from a secondary market dealer.
Explore pre-owned & used Rolex Yacht-Master 68623 watches at Bob's Watches. Discover our collection of certified and authentic luxury watches at the best prices.
The Rolex Yacht-Master is a two-part collection of luxury-oriented sports watches. Here is everything that you need to know before you make your purchase.
Buy a certified pre-owned Rolex Yacht-Master 68623 Steel and Gold sku 160450 White watch from our luxury collection. Save up to 50% on all used watches at Bob's Watches.
This pre-owned Rolex Yacht-Master 68623 watch is in very good condition. During our quality control check, any necessary adjustments are made to ensure the watch is functioning as intended. The watch is Crown & Caliber Certified Authentic and comes backed by our 1-year warranty.
Rolex Yacht-Master Listing: $7,818 Rolex Yacht-Master blue 68623, Reference number 68623; Gold/Steel; Condition Good; Watch with original box; Watch with origina. Skip. Over 144,000 users from around the world give us 4.8/5 stars Search through 555,631 watches from 124 countries.
Rolex Certified Pre-Owned Yacht-Master 1997, 35 mm, steel/yellow gold Reference: 68623. Free Delivery within 2-3 days.
This pre-owned Rolex Yacht-Master 68623 watch is in excellent condition. During our quality control check, any necessary adjustments are made to ensure the watch is functioning as intended. The watch is Crown & Caliber Certified Authentic and comes backed by our 2-year warranty.
Elegantly combining functionality and nautical style, the Yacht-Master is a veritable ally at sea designed for sailors and navigators. More on rolex.com.
Find low prices for 26 Rolex ref. 68623 watches on Chrono24. Compare deals and buy a ref. 68623 watch.
Certified preowned Sport Rolex Yacht-Master 68623 watch is made out of Stainless steel on a 18k & Stainless Steel bracelet with a 18k & Stainless Steel Fliplock buckle. This Rolex watch has a 35 x 35 mm case with a Round caseback and Gold Luminescent dial. It is Gray and Sons Certified Authentic and comes backed by our 24-month warranty.
Get the best deals on Rolex Yacht Master 68623 when you shop the largest online selection at eBay.com. Free shipping on many items | Browse your favorite brands | affordable prices.
The Rolex Yacht-Master ref 68623 (1997 - 1998) features a 35mm Oyster case, appealing to women and men alike. The ref. 68623 also boasts a striking two-tone finish, the collection's signature tapered lugs, a 60-minute timing bezel, and a blue dial set with contrasting gold-trimmed hour markers and Mercedes hands.
Elevate your style with the Rolex Yacht-Master 68623, a blend of fine craftsmanship and timeless luxury. Reliable timekeeping, stylish design, superb durability. Add to your collection now!
Rolex Yacht-Master Inserat: 7.705 € Rolex Yacht master 68623 white TNumber USED Boys (Unisex), Referenznummer 68623; Gold/Stahl; Automatik; Zustand Sehr gut; Standort: Japan, Chuo-ku, Tokyo
ROLEX OYSTER PERPETUAL DATE YACHT-MASTER 68623 WRISTWATCH 18K GOLD SS BOX TAGS. stlauctionco (16040) 99.7% positive; Seller's other items Seller's other items; Contact seller; US $225.00. 22 bids. ... Rolex Yacht Master 68623, Rolex Oyster Perpetual Gold Case Wristwatches, Rolex Oyster Perpetual Gold Band Wristwatches,
Rolex achieved perfection when it turned its attention to redefining the purpose and functionality of the bezel. Typically, a bezel operates independently from the internal mechanism; however, the Ring Command bezel in the Yacht-Master II regatta chronograph operates in conjunction with it.
Tomsk Oblast is a region in Western Siberia, which borders Omsk Oblast and Tyumen Oblast to the west, Khantia-Mansia to the north, Krasnoyarsk to the east, Kemerovo Oblast to the southeast, and Novosibirsk Oblast to the south.
Business, Economics, and Finance. GameStop Moderna Pfizer Johnson & Johnson AstraZeneca Walgreens Best Buy Novavax SpaceX Tesla. Crypto
The mid-size Rolex Yacht-Master was introduced to market in 1994, just two years after the iconic regatta timer debuted. The case offers a very desirable mid-size 35mm diameter, offering flattering proportions for those who seek a smaller sports watch. As a traditional Yacht-Master timepiece, the ref. 68623 also features a precious metal 60-minute bezel with raised and polished graduations, a ...
Want to know what the weather is now? Check out our current live radar and weather forecasts for Tomsk, Tomskaya oblast', RU to help plan your day
The Rolex Yacht-Master is growing in popularity among men and women. The addition of a mid-size variation, such as the reference 68623 that is offered here today, features a 35mm Oyster case and matching Oyster bracelet and Oysterlock clasp, creating the perfect size and weight for men and women.