Research Participants’ “Irrational” Expectations: Common or Commonly Mismeasured?

Alongstanding concern of research ethics is that some clinical trial participants may confuse the goals of treatment with the goals of research and thus fail to understand certain key design elements of research that distinguish it from medical care.¹ One such element of research is random assignment to receive either the experimental or control intervention. Although most lay people understand randomization,² the findings of studies of prospective and actual participants’ understanding of randomization is mostly discouraging, if somewhat mixed. In one study, only 50% of parents interviewed after informed consent discussions about enrolling their child in oncology randomized clinical trials reported that they understood the concept of randomization.³ In another study of oncology trial participants, only 23% understood that they would be randomized to either the control or experimental intervention.⁴ On the other hand, Simon and colleagues found that despite the fact that only 69% of adult oncology patients were explicitly told about randomization in the informed consent conversation, 93% understood that the intervention assignment would be randomized.⁵

These and other studies suggest that while some prospective and actual research participants may not understand randomization, actually measuring understanding about randomization is not straightforward, either. Indeed, some studies reveal that even when individuals seem to understand key concepts closely tied to randomization, they still seemed confused about its meaning.⁶ Traditionally, many have worried that even if research participants intellectually comprehend the concept of research and its elements, their intense desire for a cure or the habitual perception that medical procedures are therapeutic (i.e., their therapeutically oriented mindset) may interfere with applying those facts to their own situation⁷—what Appelbaum and Grisso have described as the distinction between understanding and appreciation. As they point out, “understanding requires an ability to grasp the meaning of the information disclosed,” and “appreciation involves the relevance of that information for one’s own situation.”⁸ Indeed, in the first paper published on the concept of therapeutic misconception, Appelbaum and colleagues reported a dramatic and disturbing case of a woman who seemed to understand yet failed to appreciate that study participants received an intervention on the basis of randomization.⁹ More recently, certain commentators have proposed that some research participants’ “unrealistic optimism” about clinical trial participation—the expectation of benefit that cannot be statistically true—may be a form of lack of appreciation, in which understanding does not result in adequate appreciation of the probabilities of therapeutic benefit.¹⁰

In a recent study we conducted that examined how individuals made their decision about whether to enroll in a Parkinson’s disease clinical trial that involved randomization to a gene-transfer or sham surgery arm, we included questions to examine the relationship between understanding and appreciation of randomization. We wanted to know whether prospective participants understood and appreciated the probability of being assigned to the sham or the gene transfer arm. Sham surgery controlled trials involving Parkinson’s disease are particularly important and apt settings to examine the potential influence of therapeutic orientation on individuals’ understanding and appreciation of research, for several reasons: 1) individuals with Parkinson’s disease are older and have a serious and incurable illness, raising vulnerability concerns;¹¹ 2) a neurosurgical intervention has particularly elaborate trappings of “treatment;” 3) sham surgical controls in research continue to be controversial;¹² and 4) randomization involving a sham surgical arm is perhaps the quintessential “research design” feature that distinguishes research from treatment.

Study Methods

Participants in our study were individuals considering enrollment in an early-phase gene transfer study for Parkinson’s disease: the Phase 2 Safety and Efficacy Study Evaluating Glutamic Acid Decarboxylase Gene Transfer to Subthalamic Nuclei in Subjects with Advanced Parkinson’s Disease (GAD 2 trial), sponsored by Neurologix, Inc.¹³ Enrollment in the gene transfer study meant being randomized at a one-to-one ratio to receive either injections of the study agent or sham surgery placebo. Sham surgery involved bilateral partial thickness burr holes and injection of saline into the burr holes, just under the skin.

The GAD 2 trial assessed 66 individuals for eligibility across seven research sites. These individuals met with the GAD 2 trial team to discuss the study, signed the informed consent form, and then underwent screening for the trial (informed consent was a prerequisite to full screening procedures for eligibility for the GAD trial). Ultimately, 45 individuals underwent surgery (22 received gene transfer surgery, and 23 received sham surgery). We recruited participants for our study from five of the seven sites and conducted interviews with 29 of the 45 (64.4%) persons screened for enrollment at those sites. The interviews were conducted by telephone and took place after individuals signed the consent form to participate in the GAD 2 trial. Although our goal was to interview everyone prior to surgery, due to scheduling difficulties we had to interview five of our participants after their surgery. Five additional participants were found to be ineligible for the GAD 2 trial following screening, and one subject ultimately declined participation; however, their interviews were conducted prior to these events at a time when they were actively considering participation in the trial. All interviews were audio-recorded and transcribed. The interview typically took 40–60 minutes.

We used a conditional probe interview, which is a semistructured qualitative interview guide that places a strong emphasis on allowing individuals being interviewed to follow their own narrative. With this approach, we could elicit a chronological narrative of how individuals came to make their decision concerning participation in the trial while exploring the following domains: 1) individuals’ understanding of and reaction to the purpose and design of the GAD 2 trial, including, among other topics, their understanding and opinions regarding a sham surgical arm and the possibility of being randomized to a sham surgery placebo group; 2) individuals’ motivations for participating in the GAD 2 trial; and 3) individuals’ perceptions and expectations concerning direct personal benefit from participating in GAD 2, as well as their perceptions and attitudes toward research risks and potential societal benefits of the trial. Each section began with open-ended stem questions, and the interviewer encouraged spontaneous responses to these. Follow-up probes were used to help clarify responses. The instructions and training of interviewers emphasized that because the interview approach was designed to be qualitative and oriented toward the participant, the interview should remain flexible enough to allow the respondents to speak openly and share information important to them. Thus, the interviewer was permitted to probe but not “force” a response as with the case of quantitative close-ended question, and the interviewer was instructed to allow individuals to respond as they desired, in order to ensure that the interview was not overly directive and allowed unexpected ideas to emerge.¹⁴

Since we were interested in learning about understanding and appreciation of randomization, individuals were asked about the probability of being assigned to the placebo arm versus the gene transfer arm if they enrolled in the GAD 2 trial. This understanding question was asked in the third person: “What are the chances of a participant being assigned to the sham surgery group [or gene transfer group]?” This question was intended to assess intellectual comprehension of the study design regarding the probability that any given participant would be randomly assigned to receive the placebo versus the gene transfer intervention. We then asked a second question intended to assess application of this fact (the appreciation question) by asking in first person: “What do you think your chances are of being assigned to sham surgery [or gene transfer group]?”¹⁵ Given that the randomization ratio was one to one, answers such as “50%” or “50/50” were counted as correct quantitative responses.

In analyzing the two questions of interest, the focus was on the relationship between the responses to the two questions—i.e., whether respondents could apply the understanding of facts to their own situation and whether they made any comments that might explain how they answered the questions. We unexpectedly found, however, that not all of the respondents provided a quantitative answer to the appreciation question, although they did respond to that question with various comments and explanations. We thus developed a categorization scheme combining the quantitative responses to the two questions with an analysis of the qualitative comments (Tables 2 and 3).

Table 2.

Respondent Groups According to Relationship between Responses to Understanding and Appreciation Questions

Group	Description	n (%)1
1	Correct response to understanding and appreciation questions	12 (46.2%)
2	Correct response to understanding question but provides comments rather than quantitative response to appreciation question	10 (38.5%)
3	Incorrect response to understanding question but consistent response to appreciation question	2 (7.7%)
4	Incorrect or “not sure” response to understanding question, “other” response to appreciation question2	2 (7.7%)

¹Numbers do not add to 29 because we are missing data for three participants.

²One participant’s response to the understanding question was “not positive” between “one in four chance or one in two chance” of placebo assignment, yet the participant responded “50%” to the appreciation question; the other participant provided a personal theory about why he believes he received gene transfer in response to the appreciation question.

Table 3.

Respondent Comments Categorized According to Relationship between Responses to Understanding and Appreciation Questions

Group	Comment Codes	Subject ID
1	a. No or minimal comment	1, 3, 7, 17, 18
	b. Mentions “statistics” specifically or appeals to statistical reasoning	9, 23, 29
	c. Uses probability analogies or phrases (heads or tails, coin toss, luck of draw, etc.)	6
	d. Can’t control or influence outcome/no different than anyone else	27
	e. Hopefully better than 50% chance	9
	f. Prayer will increase chances of getting gene transfer intervention	10
	g. Says “don’t know,” “no idea,” or “who knows”	20, 23, 27
2	a. Uses probability analogies or phrases (heads or tails, coin toss, luck of draw, etc.)	4, 21, 22
	b. Can’t control or influence outcome/no different from anyone else	19, 22
	c. Has positive feeling; feels lucky	11, 12
	d. Going to be in 50% who get gene transfer	12
	e. Prayer will increase chance of getting gene transfer intervention	13, 15
	f. Hopes that positive willing will affect outcome	25
	g. Superstitious	28
	h. Says “don’t know,” “no idea,” or “who knows”	4, 19, 21
3	a. No or minimal comment	24
	b. Has positive feelings; feels lucky	14
4	a. Personal theory about own chance for gene transfer	8
	b. No or minimal comment	16

The analysis of respondents’ comments followed a general scheme for the qualitative analysis of the conditional probe interviews as a whole, modifying as needed the coding procedures used for a previous study using this interview method.¹⁶ All transcriptions were checked for accuracy prior to coding and analysis. Two members of the team read through the transcripts and developed a provisional coding scheme. These codes were then discussed and further refined in meetings involving two additional team members (the first author and a coauthor who is a qualitative analysis expert). All four team members then reviewed each interview in detail and independently coded each transcript using the coding scheme. Any coding discrepancies were resolved through discussion or by team members suggesting revisions to the coding scheme that would better capture the comments; if the latter option was chosen, it was followed by independent coding by the four members and again comparing the coding. Through this iterative process the team ensured that the coding scheme was open to change and refinement. It also allowed the team to capture new and unanticipated themes. The codes were not intended to be mutually exclusive and, given the novel nature of the topic, a nonreductive approach was taken, allowing for as many codes as necessary to capture potential themes.

Study Results

Table 1 shows that most of the individuals we interviewed were men (83%), white, and well educated, and that the mean years since diagnosis of Parkinson’s disease was 10.7 years. For the understanding question concerning the probability of assignment to the sham surgery or gene transfer group, 82.8% (24 of 29) of respondents correctly replied that there was a 50% chance of assignment to either arm. Three respondents (10.3%) gave a response different from 50% (one said he initially was told 50% but then later that he had a “one-in-four chance” of receiving placebo), and two respondents (6.9%) replied they were not sure (one thought he was told two conflicting probabilities; another said no percentage was given).

Table 1.

Respondent Characteristics (N = 29)

Characteristic	Mean (SD) or n (%)
Age in years, mean (SD)	60.3 (8.0)
Gender, n (%)
Male	24 (82.8%)
Female	5 (17.2%)
Race/ethnicity, n (%)
White	29 (100.0%)
Education, n (%)
High school or equivalent	6 (20.7%)
Some college	5 (17.2%)
College degree	12 (41.4%)
Postgraduate	6 (20.7%)
Marital status, n (%)
Married	22 (75.9%)
Divorced/single/widowed	7 (24.1%)
Years since PD diagnosis, mean (SD)	10.7 (4.9)

Because our main concern is how understanding and appreciation are related, Table 2 presents respondents according to how each answered the two questions. Group 1 consists of respondents who gave correct quantitative answers to both the understanding and the appreciation questions. Table 3 summarizes the results of the content analysis of the comments from all of the respondents. Five respondents gave the correct answer to the appreciation question with relatively little or no comment. The remainder of respondents’ comments fell into two broad domains.

First, comment types b, c, and d (group 1) can be seen as explanations for the quantitative probability response to the appreciation question. Five respondents (S6, S9, S23, S27, and S29) seemed to indicate that they were answering the appreciation question as intended, as a question of probability as applied to him or herself. Some of these respondents referred to statistics explicitly: “In my work, I’ve had to look at these kind of statistics … and I think I’m pretty confident that they’ll do a good job”(S29). Others used lay language of probabilities (S6: “Luck of the draw, I guess.”), or commented on how the laws of probability applied to him or her (S27: “I can’t change those odds. If it’s 50/50, then it’s 50/50.”). One of the respondent’s comments (S23) resembled the gambler’s fallacy—e.g., the belief that if in coin tosses the coin repeatedly comes up tails, this increases the likelihood of heads in future tosses. At first, this respondent said, “I have no idea. … And I’m a gambler.” But then he went on to explain:

Usually I know the odds are 50/50. I don’t know how many have been operated on. I don’t know how many of those have gotten the real thing. I would have a better idea of my chances if I knew that. (S23)

This fallacy, however, was compatible with a correct appreciation of his own probability for receiving the gene transfer, since he had no special knowledge of the actual results of intervention allocation.

The second domain of comments made by respondents in group 1 seemed to address a more subjective take on the appreciation question (group 1, comment types e, f, and g). The following response to the appreciation question by S9 exemplifies how someone may have a correct statistical appreciation and yet express something more subjective: “Well, hopefully it’s better than 50/50, but statistically it isn’t.” He expresses a hope about what he would like to receive in addition to, or contrasted with, his appreciation of the actual probabilities.

This tendency to express one’s personal feelings regarding a future outcome in addition to the correct quantitative answer to the appreciation question was not uncommon, and it took other forms besides “hope.” For example, respondent S10 said, “If I don’t get one [i.e., gene transfer], I’m going to be surprised.” This comment may seem at odds with this respondent’s correct answer to the appreciation question, but the comment is preceded by this rationale: “I’ve got a lot of people praying for me, a lot of people believing, and I think that I’m going to go in there hoping with a positive attitude that I’m getting one.”

Three respondents (S20, S23, and S27) included a comment to the effect that they did not know the answer when asked the appreciation question—even as they gave the correct answer to it. For example, S20 said “50/50” in response to the appreciation question but then went on to say, “I don’t know. I gave up gambling a long time ago. … I could say one thing, ‘I don’t know.’” These responses were not an admission of lack of knowledge of the probabilities. This is clear from the comments by S27:

Interviewer: What do you think about your chances [of receiving the gene transfer intervention] going into it?

S27: 50/50. Who knows?

Interviewer: Same?

S27: Yeah. I can’t change those odds. If it’s 50/50, then it’s 50/50.

These respondents felt a need to say “I don’t know” or “who knows” even while clearly articulating the correct probabilities to the appreciation question. Thus, they seemed to be communicating something more than their knowledge: an expression of how they feel about what will actually happen to them.

In group 2, four respondents (S4, S19, S21, and S22) made comments of type a and b that were quite similar to those made by group 1 respondents who were explaining their quantitative answers. Although not explicitly providing quantitative answers to the appreciation question, these group 2 respondents made comments that practically implied such an answer. For example, S22 said, “I think that I’m no different than anyone else. I think in surgery they’re going to say, ‘Well, this guy is here. We’ll draw the card today and that’s it.’ And S19 said, “I have no way of controlling it or no way of knowing, no way of putting any input in.” It is notable that three of these four respondents (S4, S19, and S21) also made a comment to the effect of “I don’t know,” just as some of those in group 1 did, as an expression of uncertainty about what will actually happen, even though they provided correct explanations about how randomization worked (“Heads or tails,” “I have no way of controlling it,” and “flip a coin”).

The remaining six respondents in group 2 made comments similar to those made by respondents in group 1 regarding the role of influences such as prayer and hope (group 2 comment types c–g made by S11, S12, S13, S15, S25, and S28). Two respondents made comments about the role of prayer and faith that were essentially the same as S10’s comment quoted above: “Well, because I’ve got my whole church group praying for me” (S13) and “I’ve got about 400 people praying I get the good stuff, and so I think it is in my favor (laughter)” (S15). Respondent S25 mentioned hope, saying he “was hoping that my will to get the real thing was going to count for something.”

In addition to prayer and hope, these group 2 respondents mentioned two other potential influences: superstitious beliefs and lucky feelings. S28 was “afraid that thinking about it will jinx it.” Respondent S12, after stating that half the subjects would receive the placebo, said:

S12: I’m going to be the other half.

Interviewer: You’re going to be in the other half?

S12: Yeah.

Interviewer: You feel pretty sure about that?

S12: The real thing. I’m pretty sure.

Yet, later in the interview, he explained the basis for this confidence: “My wife is convinced that I’m going to get the real thing. She says, ‘I have a feeling.’ I’m going with that. That sounds good to me.” S11 also talked in a similar vein, connecting his hope with belief in luck:

S11: I’m good. I’m a gambler. I like my odds … I’m a good guy, too, so, you know … I figure, I’m lucky.

Interviewer: You feel in general like you’re lucky, and it might be the same way here?

S11: Yes, I feel I’m pretty lucky.

Interviewer: Do you think you have a better than 50% chance of getting the gene transfer?

S11: I hope so. Like I said, I feel lucky.

The two respondents in group 3 (S14 and S24) gave incorrect answers to the understanding question but their appreciation question responses were the same as for the understanding question, with little or no comment. However, one of these respondents (S14) at a later point in the interview stated he believed he would get the real intervention because “my patient number is 007” and that he thought that was a “good sign.”

One respondent in group 4 felt his chance of receiving the real intervention was “very, very good”(S8). This respondent was interviewed after his surgery. He felt that since he was the first participant at his trial site to undergo surgery in the study, “more than likely they wouldn’t do the placebo on me” because there was a visiting supervising surgeon. His rationale was that the site surgeon “wouldn’t have learned anything” from the visiting supervising surgeon if only a placebo surgery were done.

Discussion

The therapeutic orientation of the participants in our study was quite evident: they preferred to be in the gene transfer arm of the GAD 2 trial. This should not be taken as evidence of a misunderstanding. Most individuals enroll in clinical trials with a desire for therapeutic benefit, especially when there are no curative treatments for them. Thus, it would be surprising if research participants did not want to receive the experimental intervention. But the desire for therapeutic benefit as a motivation for participating in a clinical trial is compatible with an understanding of the purpose of research.¹⁷ The key question is whether, as a matter of empirical fact, such motivation impedes understanding and appreciation of the purpose of the trial and of randomization. The clear therapeutic orientation of participants in our study provided an important opportunity to examine whether therapeutic orientation clouded or impaired their ability to understand randomization probabilities of the GAD 2 trial design and whether they were able to apply that understanding to their own situation (i.e., appreciation) as a research participant.

We found that most of the respondents (83%) showed intellectual comprehension or, in our use of the term, understanding of the randomization probability. Most of these respondents, in turn, also gave a correct answer to the appreciation question (12 of 22, or 55%) while the remainder of the respondents did not give a quantitative answer to the appreciation question. Thus, the key question of our study is whether respondents in this latter group (group 2 in Tables 2 and 3, comprising 45% of those who answered the understanding question correctly) lacked appreciation despite intellectual comprehension. Several considerations indicate that their comments do not imply a failure to appreciate this key design element of research.

Notably, no participant in our study actually provided a quantitative response that was inconsistent with his or her response to the understanding question. Group 1 participants all provided correct quantitative responses to both questions, and group 3 participants gave incorrect understanding responses but were consistent in responding to the appreciation question. Respondent S16 in group 4 was unsure about his response to the understanding question but in fact gave a correct response to the appreciation question. Respondent S8 in group 4, who was interviewed after surgery, did overestimate the likelihood of having received the gene transfer (“very, very good”), on a basis whose rationality is relatively difficult to assess. Recall also that some of the respondents in group 2 all but gave quantitative answers (S4, S19, S21, and S22), providing comments indicating that they were addressing the statistical question to themselves. Indeed, in two cases (S4 and S22) our team considered categorizing them as group 1 because both clearly agreed that the probability was 50%, but we refrained from doing so because it was the interviewer who first mentioned the quantitative answer in probing their comments.

What, then, of those group 2 respondents not in this subgroup whose comments could be taken, at least on the surface, as being in tension with (or going beyond) the actual statistical probabilities? Recall that even some participants in group 1 (who gave correct answers to both questions) went beyond a purely probabilistic realm in answering the appreciation question. Several expressed their subjective feelings about what the actual outcome might be. What were they communicating by such statements as “Who knows?” or by appealing to hope or to prayer? From a purely statistical point of view, statements about the likelihood of a future event can only be a mathematical probability statement. But conversational conventions, which have their own rules of rationality,¹⁸ often allow and even require that the speakers make statements about the likelihood of a future outcome that are not expressions of mathematical probability. Thus, when an underdog soccer team’s captain is asked about prospects for a victory against, say, the Brazilian national team, it would be conversationally jarring if that captain simply said, “Well, there is a 95% chance that Brazil will win and we will lose.” But if he said, “Our team’s been practicing well; I think we’ve got a really good chance at an upset,” that would make perfect sense. Taken only as a statistical statement, “good chance at an upset” is tantamount to saying something quite nonsensical: “there is a high probability that a low probability event will happen.” But we do understand “good chance at an upset” as a rational statement in the context of sports interviews. Likewise, some of our respondents said “50/50” and yet added, “Who knows?” or expressed confidence in an outcome due to prayer or hope.

Still, for respondents in group 1, one might say we can be reassured that they appreciated the randomization probabilities, whereas for those in group 2 who made similar comments about hope, prayer, luck, or superstition, we cannot be reassured of this. Here again, we wonder whether we are ignoring the context of the conversation from the point of view of the respondents. Although we as researchers saw the significance of asking the randomization probability question in the third person (understanding) and then in first person (appreciation), this distinction was likely not understood by some respondents as intended. It is reasonable to conjecture that for some respondents, the distinction between understanding and appreciation as we have framed it is a foreign one. Perhaps in answering the understanding question, some felt that the answer they gave already applied to them as a matter of course. As one respondent put it with apparent surprise:

S1: It would be 50/50.

Interviewer: 50/50? So, like for everybody else?

S1: Isn’t that what it’s supposed to be? Yeah. That’s what it’s supposed to be, isn’t it?

Some respondents obviously did “get” the very distinction we had in mind (e.g., comments like “I’m no different from anyone else”). But many addressed an additional (in the case of group 1 respondents) or a different (in the case of group 2 respondents) dimension or interpretation to the appreciation question: their subjective feelings regarding what the actual outcome might be. We assumed that respondents’ comments would be communications of knowledge (as applied to themselves), but they apparently had other goals in responding to our queries. It may be that respondents in group 2 did not give a quantitative response to the appreciation question because, having addressed that issue in their response to the understanding question, they did not feel a need to repeat it. Then, when they heard the appreciation question, they not unreasonably surmised that the question must not be about statistics, but about something more personal, something that in their minds is quite rational to focus on, given the context: their feelings about whether they thought they would receive “the real thing”—a natural context to apply their beliefs about faith, luck, and optimism.

This interpretation relies on a distinction between what is statistically rational and what is conversationally or socially rational. Our worry is that by attempting to examine individuals’ statistical rationality, researchers may end up with misleading conclusions because they fail to take account of social and conversational rationality. This worry is not new. The importance of considering conversational pragmatics (i.e., study of meaning conveyed by context rather than simply by the literal meaning of words) is increasingly recognized in psychological research. As Hertwig and Gigerenzer point out, even in carefully designed and celebrated psychological experiments, research participants may appear to be falling into various “fallacies,” but their “irrationality” may in fact be rational in an important sense. They note that “adhering to social norms, here conversational maxims, is rational, although it conflicts with classical rationality, defined by many researchers … as adherence to the laws of probability theory and logic.”¹⁹ Ignoring the social and conversational context will ensure that the researcher will fail to see that the interpretation of the task by the participant may not coincide with what is intended by the experimenter.²⁰ In a careful philosophical treatment of this issue, Lee points out that Kahneman and Tversky, whose celebrated research has shown a variety of reasoning fallacies in people, specifically note that “conversational aspect of judgment studies deserves more careful consideration than it has received in past research, our own included.”²¹

As applied to bioethics, Weinfurt has pointed out that studies that have found “false beliefs” in the expectations of research participants (usually about benefits of research) tend to use an informational processing model of psychology that treats communication only as a vehicle for expressing the participants’ knowledge, rather than as additionally conforming to inherent conversational and social expectations.²² Other philosophical analyses support this need to go beyond an information transfer model of informed consent.²³ Indeed, there is emerging evidence from studies using both qualitative²⁴ and quantitative methods²⁵ that approaching informed consent as merely an informational transfer endeavor, ignoring its social and conversational context, can be quite misleading, both to participants and to researchers attempting to measure its quality.

Our study has limitations. The sample size was small. Although we took pains to ensure validity and reliability in our coding, they remain interpretations that can be disputed. Further, because respondents were recruited from the informed consent process for a highly specialized study with a highly invasive experimental protocol, one should be careful in generalizing the results to studies of lower risk or where effective treatments do exist. Finally, we agree with other writers that the essential informed consent issue for randomization is that individuals understand that assignment to a study arm is not based on individual therapeutic considerations, and not necessarily whether they have a correct appreciation and understanding of statistical probabilities.²⁶ However, to the extent that a disjunction between understanding and appreciation is proposed as a mechanism for misconceptions that threaten informed consent, the results of our study will be relevant.

We close with two main implications of our study. First, perhaps we need to be more hesitant about labeling research participants as “irrational,” “unrealistic,” and harboring “misconceptions,” unless such conclusions are drawn from carefully designed studies that take into account the relevant social and conversational contexts of their responses to questions posed to them. Such studies may take the form of careful conversational analyses or experiments that test specific hypotheses.²⁷ There is no doubt that some individuals’ expectations about participating in research are indeed misconceptions; but researchers assessing expectations cannot simply assume that individuals are merely vehicles for knowledge and information transfer.²⁸ We risk stigmatizing prospective and actual research participants not because of their vulnerabilities, but because we measure their behavior and speech in inadequate ways.

Second, improving the informed consent process may prove difficult if we misunderstand what individuals say about studies they are recruited to join. Misunderstandings due to lack of attention to conversational contexts may lead to interventions—such as out of context reassurances or information—that could promote confusion or distrust rather than clarity.²⁹ In the context of clinical treatment research, the correct approach to improving the informed consent process may involve some form of working with, rather than working against, the therapeutic orientation that individuals bring to that process.