Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

by Scott Y. H.

Kim, Renee Wilson, Raymond De Vries,


H. Myra Kim, Robert G. Holloway, and Karl Kieburtz

Could the High Prevalence of Therapeutic Misconception Partly


Be a Measurement Problem?

T
herapeutic misconception (TM) is said to occur but a polite request). Thus, what appear to be misun-
when an individual “inaccurately attributes a derstandings, misconceptions, or unrealistic beliefs of
primacy of therapeutic intent and individual- research subjects may at least partly reflect the limita-
ized care typically seen in ordinary clinical settings to tions of the way such phenomena are measured.10 For
research procedures”1 or “when individuals do not un- example, when a research participant uses a probability
derstand that the defining purpose of clinical research statement about a potential outcome (such as personal
is to produce generalizable knowledge.”2 It is reported benefit), evidence shows that he or she may be using it
that many research participants fail to understand the to express hope, faith, or wish for luck rather than stat-
purpose of research. For instance, some studies have ing his or her belief in a mathematical probability.11
found that clear majorities of oncology clinical trial We hypothesize that when people are asked about
participants failed to understand that the trial was the purpose of research in the context of considering
“mostly research,”3 and studies specifically designed participation in research, they could be answering a dif-
to detect TM reported that a majority of the research ferent question than the one intended by the researcher
participants showed evidence of TM.4 Although other attempting to detect TM. This hypothesis is based
studies have found lower rates of misunderstanding on two considerations. First, in our interviews with
of the purpose of research (e.g., 25%5), a recent study research participants, some answer questions about
found that nearly 40% of patients enrolled in early- the purpose of research incorrectly (e.g., stating that
phase oncology trials failed to understand that the pur- the primary purpose is to benefit them) but neverthe-
pose of their clinical trials was not primarily to benefit less show in other statements that they understood that
them.6 Indeed, the phenomenon of TM is sometimes the research was not tailored for their benefit. These
spoken of as ubiquitous.7 Is this conventional view patient-subjects seem to interpret the purpose of the re-
correct? Are research participants routinely making a search question as asking what their own purpose is in
fundamental error in understanding? participating in a study. Second, this anecdotal obser-
A small but growing body of literature is beginning vation is consistent with theories about how linguistic
to probe this issue and to challenge the conventional context can have a profound effect on how we interpret
view.8 These new studies draw on insights from the statements. The sentence “He overlooked the battery”
linguistic field of pragmatics—the study of meaning can have two very different meanings depending on
conferred by factors beyond the semantic meaning whether it is preceded by “The technician was incom-
of the words such as the context and function of the petent” or by “The judge was lenient.”12 If what we
communication.9 Ordinarily, mutual comprehension in currently measure as misunderstandings of the purpose
communication relies on the fact that context removes of research are, at least partly, not misunderstandings
ambiguities or even allows interpretations contrary of the purpose of research but rather reports of re-
to literal semantic meanings (as with “Can you pass search participants’ motivation for participation, then
the salt?,” which is not a question about an ability our current estimates of TM could be inflated. We do
not know to what extent, if any, people might interpret
Kim SYH, Wilson R, De Vries R, et al. Could the high prevalence of therapeu-
tic misconception partly be a measurement problem? IRB: Ethics & Human
TM survey questions in this unintended way.
Research 2015;37(4):11-18.

IRB: E t h i c s & H u m a n R e s e a rc h J u ly -A u g u st 2015


© 2015 The Hastings Center. Permission is required to reprint. 11
Figure 1.
Follow-up Question regarding Purpose-of-Research Question

Closed-ended follow-up to purpose question Open-ended follow-up to purpose question


Question asked “We would like to know how you interpreted the “Use the space below to tell us what was going
above question. Which of the following do you through your mind when you answered the last
think best captures how you understood question. In your opinion, what did you think
the question?” the question was asking? Why did you give the
answer you did?”

Response options A. “It was asking what my own intentions would Free text
(A-C presented be for participating (or not participating).
in random order) In other words, which of the intentions mentioned
best reflects why I would or would not participate
in the study.” [PERSONAL MOTIVATION]

B. “It was asking whether the research is being done


to create knowledge for knowledge’s sake or to help
actual people.” [KNOWLEDGE VS. PEOPLE]

C. “It was asking about the official purpose of the


research study. That is, the researchers’ overall main
goal in doing the study.” [OFFICIAL PURPOSE]

D. “I am still not sure what the question was asking.”

A key feature of our hypothesis is that what appears 1) whether the question is understood by people as
to be TM (at least in regard to understanding of the intended and 2) whether there is a relationship between
purpose of research) is not essentially tied to actually people’s answers to the purpose question and whether
being a patient-subject whose desire for benefit blinds they understood the purpose question as intended. A
him or her to the real purpose of research. Rather, such confirmatory follow-up study was also conducted.
a misunderstanding is hypothesized to be due to the lin-
guistic context in which the purpose question is asked. Study Methods
Thus, we envision two steps to test this idea. First,
if people incorrectly answer the purpose-of-research
question due to a linguistic effect (and not because
W e recruited 100 individuals through Amazon’s
Mechanical Turk (MTurk), one of Amazon’s
web services that allows users to post tasks, such as
of some special psychological state essentially tied to surveys, and that is widely used in experimental surveys
being a patient-subject), this should be detectable in in psychology, economics, and political science.14 Re-
competent users of English hypothetically discussing spondents can browse and complete tasks for monetary
research participation, regardless of whether they are compensation (we paid $0.50 per subject for this five-
actual patients considering real research participation. minute survey). All respondents were from the United
If such effect cannot be observed in nonpatients, then it States, and the data were collected anonymously, as
is unlikely that a linguistic effect could account for high Amazon handles the compensation.
prevalence of TM. Second, if the effect is observed by The survey began with a hypothetical description
merely creating the linguistic context even in nonpa- (Supplementary Figure A, available on the IRB: Ethics
tient-subjects, then the next step would be to repeat the & Human Research website) of an early-phase clinical
study in actual patient-subjects. This article reports on trial for amyotrophic lateral sclerosis (ALS), a usually
our work with nonpatients. terminal and debilitating neurological condition with
We used a purpose-of-research question (which we no effective treatment. Using a scale anchored at one
will refer to as the “purpose question”) from a study end by “would not consider at all” (score of 0) and at
by Pentz and colleagues13 to examine two questions: the other by “definitely would consider” (score of 10),

J u ly -A u g u st 2015 IRB: E t h i c s & H u m a n R e s earc h


12
respondents were asked whether they would consider analysis of respondents’ comments. First, two team
participating in the clinical trial if they had been diag- members read the comments and independently cre-
nosed with ALS. They were invited to provide com- ated coding categories. These two team members then
ments to their responses. compared their codes and merged them into one coding
The participation question was followed by a scheme; two additional team members reviewed the
purpose-of-research question from the study by Pentz coding scheme and made further revisions. All four
and colleagues: “Is the research study mostly intending coders tested the resulting coding scheme to code the
to help research and gain knowledge or mostly intend- comments. Any discrepancies and further modifications
ing to help you as a person?”15 The response choices to the codes were made by group discussion. We tested
consisted of “Mostly intending to help research and whether the responses to the purpose question were re-
gain knowledge” (the response that researchers would
ordinarily interpret as correct and not evidencing TM),
“Mostly intending to help you as a person” (the incor-
The phenomenon of TM is sometimes spoken of
rect response ordinarily taken to exhibit TM), and as ubiquitous, but is this accurate? Are research
“Don’t know/Not sure.” participants routinely making a fundamental error
We then randomized the participants to two arms
and tested two modes of asking a follow-up question
in understanding?
(closed-ended and open-ended) to elicit the respon-
dents’ interpretation of the purpose question (Figure 1). lated to the responses to the follow-up questions, using
Half of the respondents were randomized to a closed- the Fisher’s exact test.
ended question with three response options (plus a “not Approximately nine months after the above experi-
sure” option), with the order of presentation for the ment, we conducted a confirmatory survey using the
three options randomized. The other 50 respondents closed-ended format with a sample of 135 participants
in the second arm were presented with an open-ended recruited via Amazon MTurk in a similar manner as
question that they were asked to answer in free text. above.
The purpose of the two randomized arms was to allow
validation of the closed-ended question responses by a Study Results
qualitative analysis of responses to the questions in the
open-ended arm.
Descriptive data (frequencies and percentages) were
T he mean age of respondents was 32.8 (SD =
10.4). Eighty-six percent were Caucasian. Women
comprised 44% of respondents. Forty-two percent of
calculated for demographic variables and responses respondents had graduated from college. The income
to the survey questions. For the open-ended follow- distribution range included 14% earning less than
up question arm, we conducted a qualitative content $20,000 a year, 59% earning between $20,000 and

Table 1.
Relationship between Purpose-Question Response and Interpretation of Purpose Question,
Closed-Ended Arm (N = 50)

Purpose-question response
Help research Help you as a person Don’t know Total (%)
and gain knowledge
Interpretation
Personal motivation 5 5 0 10 (20%)
Knowledge vs. people 8 2 2 12 (24%)
Official purpose 26 0 2 28 (56%)
Total (%) 39 (78%) 7 (14%) 4 (8%) 50 (100%)

Fisher’s exact test, p = 0.001

IRB: E t h i c s & H u m a n R e s e a rc h J u ly -A u g u st 2015


13
Table 2.
Relationship between Response to Purpose Question and Interpretation of Purpose
Question, Open-Ended Arm (N = 49)

Purpose-question response
Help research and gain knowledge Help you as a person Don’t know. Total (%)

Interpretation
Personal motivation 8 5 2 15 (31%)
Knowledge vs. people and other 1 2 1 4 (8%)
Official purpose 27 0 3 30 (61%)
Total (%) 36 (73%) 7 (14%) 6 (12%) 49 (100%)

Fisher’s exact test, p = 0.001

$60,000, 16% earning $60,000 to $100,000, and 11% er’s exact test, p = 0.001). Thirty-three percent (13/39)
earning more than $100,000. When asked whether of those who correctly answered the purpose question
they would consider participating in the ALS trial (on in fact misinterpreted the question. Most notably, no
a scale where 1 = would not consider at all and 10 = respondent who interpreted the purpose question as
definitely would consider), most respondents said they intended gave an incorrect (i.e., TM) response to that
would consider participating, with a mean response of question.
9.1 (SD = 1.4). Sixty respondents gave 10 as an answer, One of the respondents in the arm that was given an
with only three responding on the negative side (i.e., open-ended follow-up question to the purpose ques-
5 or lower). All but one respondent provided written tion did not provide a response. Of the remaining 49
comments. Content analysis of their comments showed comments, 30 were coded with unanimous agreement
that 76% (76/99) made comments reflecting a strong among the four coders. Three of the four coders agreed
therapeutic motivation, such as “ALS is almost always about 13 comments, and for 6 comments a group
fatal so I would want a chance to live even if it is a discussion was required to reach a consensus. As can
small chance.” The written comments reflect a strong be seen in Supplementary Table C (see the IRB: Eth-
therapeutic orientation toward participation in a hypo- ics & Human Research website), although there were
thetical ALS study. numerous subthemes, the comments generally fell into
With regard to the purpose question, 75% gave the three broad themes corresponding to our closed-ended
“correct” answer, 15% gave the “incorrect” answer question response options. Sixty-one percent of respon-
(i.e., the answer that would be interpreted as exhibiting dents (30/49) interpreted the purpose question as some
TM), and the remaining 10% chose “don’t know/not variant of its intended meaning, 31% (15/49) interpret-
sure” (Supplementary Table A, available on the IRB: ed it as a form of a personal motivation question, and
Ethics & Human Research website). The rates were two respondents (4%) interpreted it as a “knowledge
similar in both arms (chi-square test, p = 0.75). vs. people” question. Cross-tabulating these results
With regard to how our respondents interpreted the against the responses to the purpose question reveals a
purpose question, in the closed-ended question arm, strong relationship (Fisher’s exact test, p = 0.001). As
56% (28/50) of respondents chose the intended mean- in the closed-ended arm, no one who interpreted the
ing of the purpose question, with the remainder split purpose question as intended gave an incorrect answer
between the “personal motivation” interpretation and to it. Twenty-five percent (9/27) of those who correctly
“knowledge vs. people” interpretation (Supplementary answered the purpose question in fact misinterpreted
Table B, available on the IRB: Ethics & Human Re- the question (Table 2).
search website). Cross-tabulation of these results with The results of our follow-up survey (identical to
the response to the purpose question (Table 1) shows the closed-ended arm, but with 135 participants) are
a strong relationship between the purpose question re- consistent with the results from the first survey (Table
sponses and the interpretations of that question (Fish- 3). Overall, 15.6% (21/135) chose the incorrect (i.e., in-

J u ly -A u g u st 2015 IRB: E t h i c s & H u m a n R e s earc h


14
dicative of TM) response to the purpose question, and response to the purpose question (total of 35 respon-
only 46.7% (63/135) interpreted the purpose question dents in our two surveys), only three interpreted the
as intended. Only 4.8% (3/63) of those who interpreted purpose question as intended (8.6%). In other words,
the purpose question as intended answered it cor- understanding the purpose question as intended virtu-
rectly (Fisher’s exact test, p < 0.001). Forty-six percent ally precludes a TM response, and a TM response is a
(50/109) of those who gave the correct response to the potent indicator that the person has misinterpreted the
purpose question did not in fact interpret the purpose purpose question. The apparent failure of the survey
question as intended. respondents to understand the purpose of research by
providing a TM response is almost entirely a function
Discussion of how they interpreted the purpose question.

T he federal regulations governing research with hu-


mans specify that one of the general requirements
of informed consent is that research subjects be pro- Understanding the purpose question as intended
vided with a statement that the study involves research. virtually precludes a TM response, and a TM
If research subjects routinely fail to understand that response is a potent indicator that the person has
research is not primarily to benefit them but to gener-
misinterpreted the purpose question.
ate knowledge, then they are routinely enrolled without
understanding a cardinal requirement of informed
consent. But is it true that this kind of misunderstand- About half of the respondents (46% when studies
ing is widespread? In order to attribute such a misun- are combined) did not interpret the purpose question as
derstanding, one must first assume that the purpose intended. We suspect that had we used typical pretest-
question (“Is the research study mostly intending to ing procedures for survey research, such as cognitive
help research and gain knowledge or mostly intending interviewing,16 we would not have picked up the fact
to help you as a person?”) is interpreted as intended by that nearly half of the persons in our survey did not
those who apparently exhibit TM. interpret the question as intended. First, pre-test cogni-
Our study about a hypothetical ALS clinical trial tive interviews are not intended to provide quantitative
suggests that we should not rush to this assumption. estimates. For instance, even if we find only one incor-
Almost no one who interpreted the purpose question as rect interpretation in 10 cognitive interviews (a typical
intended provided a response indicative of TM (“most- number for pretesting),17 the 95% confidence interval
ly intending to help you as a person”). And by combin- will extend to as high as a 45% misinterpretation rate.
ing the responses from both surveys, we found that Second, a slight majority (56%, 32/57) of the respon-
only 2.5% (3/121) of respondents who interpreted the dents in our survey who interpreted the purpose ques-
purpose question as intended gave the TM response to tion as a personal motivation question expressed altru-
the purpose question. Among those who chose the TM istic motivation. It is likely that none of these would be

Table 3.
Confirmatory Follow-up Survey Results of Relationship between Response to Purpose Question
and Interpretation of Purpose Question, Closed-Ended Format (N = 135)

Purpose-question response
Help research Help you as a person Don’t know Total (%)
and gain knowledge
Interpretation
Personal motivation 19 12 1 32 (23.7%)
Knowledge vs. people 31 6 3 40 (29.6%)
Official purpose 59 3 1 63 (46.7%)
109 (80.7%) 21 (15.6%) 5 (3.7%)

Fisher’s exact test, p < 0.001

IRB: E t h i c s & H u m a n R e s e a rc h J u ly -A u g u st 2015


15
seen as misinterpreting the purpose question during a personal motivation question. Weinfurt points out that
cognitive interview for the following reason. Imagine a this discrepancy in perspective between the researcher
cognitive interview with one such subject. This respon- and the individual he is interacting with has a vener-
dent may not express confusion, appear hesitant, or ask able history, labeled by William James in 1890 as “the
for clarification since the respondent feels that he or she psychologist’s fallacy.”19
knows what the question is asking. Further, when he A strength of our study is that we complemented the
or she talks about “helping future patients” (reflecting closed-ended follow-up question arm with an open-
the respondent’s altruistic motivation) in discussing the ended question arm, allowing important validation.
purpose question, it will not be apparent to the cogni- Further, a confirmation of the results with a bigger and
tive interviewer—unless the interviewer specifically entirely new sample provides robust validation that
probes for it—that the respondent is talking about his is unusual in empirical studies about research ethics.
or her personal motivation. It would be natural for the Our procedures also allowed us to see that one of our
interviewer to assume that the purpose question is func- response options (“create knowledge for knowledge’s
tioning as intended in such cases. sake or to help actual people”) did not elicit a uniform
It is remarkable, but consistent with our linguistic response across the two arms (24% vs. 4%) or in the
hypothesis, that even nonpatients thinking about a hy- confirmatory study (8%). Although this could be due
pothetical scenario fail to understand the purpose ques- to small sample size, we suspect that the meaning of
tion as intended. Meaning is more than semantic. There the response option was not clear. In future research, it
are pragmatic considerations (meanings conferred by may be preferable to eliminate this option.
the context or function of communication) that are at It is important to emphasize that the purpose of our
play. According to relevance theory—the most empiri- surveys was not to show that TM is rare among actual
cally tested theory of pragmatics—people “follow a clinical research participants. One would need to study
path of least effort in constructing an interpretation of actual patient-subjects in order to prove such a claim.
the utterance” they hear until their expectation of the Instead, our goal was to test and delineate a potential
relevance of what they heard is satisfied.18 Whether in mechanism to explain TM that could then be applied in
a real study or in a hypothetical exercise, the context actual patient-subjects. If TM (at least the component
for the potential research subject is one of making a de- that has to do with answers to “purpose-of-research”
cision whether to enroll in a study. That is their overall questions) is partly the result of a linguistic effect, then
organizing focus because that is what the experimenter that linguistic effect should be inducible even in non-
(i.e., researcher attempting to detect TM) has up to that patients in hypothetical scenarios merely by engaging
point indicated to him or her as the focus of their com- them in that linguistic context. If such an effect cannot
municative interaction. Thus, even if individuals fully be shown with nonpatients, the linguistic hypothesis
understand the purpose of research, many will organize cannot be true. Showing that the linguistic effect does
their conversation with the researcher studying TM occur with nonpatients provides evidence that the hy-
from the point of view of their own decision making. In pothesis must be considered as a potential explanation
such a context, a question containing concepts such as for the very high incidence of TM on the part of re-
intention and purpose may naturally be interpreted by search participants that some researchers report. Thus,
some as a query about why they would participate in our results provide a compelling reason to conduct
research. Such an interpretation is efficient, consistent future studies on the linguistic effect in actual patient-
with the context of the conversation, and consistent subjects as an explanation for the high incidence of TM
with what relevance theory would predict. Note that phenomena.
this does not mean that individuals are only self-inter- It is unlikely that patient-subjects are in particular
ested; some will indeed be motivated by altruism to en- “immune” to the linguistic effect that we have shown
roll in a study. But the point is that their focus is not on in nonpatients. In fact, it seems likely that being a
demonstrating their understanding to the interviewer; patient would make one more, rather than less, prone
it is on the decision being considered. Thus, it should to the linguistic effect (i.e., more likely to interpret the
not be surprising that almost half of the respondents purpose question as a personal motivation question)
in our survey asked about participating in a hypotheti- since the decision-making context is so much more
cal study tended to interpret the purpose question as a real for them than for the nonpatients. If so, then this

J u ly -A u g u st 2015 IRB: E t h i c s & H u m a n R e s earc h


16
could explain the higher prevalence of the TM response University of Rochester and the University of Michigan
among patient-subjects than was found in our study of institutional review boards.
nonpatients.
Disclaimer
Our results should not be taken to suggest that TM
does not exist. We are convinced that it does because The opinions in the article are the authors’ own and
do not represent the views of the National Institutes of
we have observed it.20 At times, people do truly fail to
Health, the Department of Health and Human Services, or
understand that the purpose of research is not primar-
the federal government.
ily to benefit them. But when TM appears to be very
n Scott Y.H. Kim, MD, PhD, is senior investigator in the Department
common even after it has been explained to prospective of Bioethics at the National Institutes of Health; Renee Wilson, MA,
participants that they are being asked to be in a re- is senior project manager at the Center for Human Experimental
Therapeutics at the University of Rochester; Raymond De Vries, PhD,
search study, the first impulse should be to assess how is professor at the Center for Bioethics and Social Sciences in Medi-
we are measuring their understanding of the study. cine and the Department of Medical Education at the University of
Michigan; H. Myra Kim, ScD, is adjunct professor in the Department
That TM is ubiquitous is a long-standing view in the
of Biostatistics at the University of Michigan; Robert G. Holloway,
world of research ethics. Therefore, a claim that TM is MD, MPH, is professor and chair of the Department of Neurology at
not as prevalent must explain why it appears to be so the University of Rochester; Karl Kieburtz, MD, MPH, is the Robert
J. Joynt Professor in Neurology and professor in the Department
prevalent. Based on careful qualitative analyses of inter- of Environmental Medicine and the director of the University of
views with actual patient-subjects and borrowing from Rochester Clinical and Translational Science Institute and Center for
Human Experimental Therapeutics at the University of Rochester.
insights from pragmatics (especially relevance theory),
we have developed a linguistic hypothesis regarding at
References
least one core aspect of TM: people appear to misun-
1. Lidz CW, Appelbaum PS. The therapeutic misconception: Prob-
derstand the purpose of research because the questions lems and solutions. Medical Care 2002;49(9):V55-V63, at V57.
used to measure TM are not understood as intended. 2. Henderson GE, Churchill LR, Davis AM, et al. Clinical trials
and medical care: Defining the therapeutic misconception. PLoS
Instead, the context allows some people to interpret the Medicine 2007;4(11):1735-1738, at 1736.
purpose question as a question about their own motiva- 3. Schaeffer MH, Krantz DS, Wichman A, et al. The impact of
disease severity on the informed consent process in clinical research.
tions. As a linguistic phenomenon, this misinterpreta-
American Journal of Medicine 1996;100(3):261-268; Daugherty C,
tion of the purpose question should be measureable in Ratain MJ, Grochowski E, et al. Perceptions of cancer patients and
nonpatients who are competent English users, and our their physicians involved in phase I trials. Journal of Clinical Oncol-
ogy 1995;13(5):1062-1072.
results robustly demonstrate this. The next step is to 4. Appelbaum PS, Lidz CW, Grisso T. Therapeutic misconception
test whether among patient-subjects who understand in clinical research: Frequency and risk factors. IRB: Ethics & Hu-
man Research 2004;26(2):1-8; Appelbaum PS, Anatchkova M, Albert
the purpose-of-research question as intended, a high
K, et al. Therapeutic misconception in research subjects: Develop-
proportion believe that research is primarily intended ment and validation of a measure. Clinical Trials 2012;9:748-761.
to benefit them. 5. Joffe S, Cook EF, Cleary PD, et al. Quality of informed
consent in cancer clinical trials: A cross-sectional survey. Lancet
2001;358:1772-1777.
Supplementary Figure and Tables 6. Pentz RD, White M, Harvey RD, et al. Therapeutic misconcep-
Supplementary Figure A and Tables A, B, and C are tion, misestimation, and optimism in participants enrolled in phase 1
trials. Cancer 2012;118(18):4571-4578.
available on the IRB: Ethics & Human Research website. 7. Dresser R. The ubiquity and utility of the therapeutic miscon-
ception. Social Philosophy & Policy 2002;19(2):271-294.
Acknowledgments 8. Weinfurt KP. Discursive versus information-processing per-
spective on a bioethical problem: The case of ‘unrealistic’ patient
This work was supported in part by the National expectations. Theory and Psychology 2004;14(2):191-203; Kim S, de
Institute for Neurological Disorders and Stroke (R01- Vries R, Wilson R, et al. Research participants’ ‘irrational’ expecta-
NS062770) and a Clinical and Translational Science tions: Common or commonly mismeasured? IRB: Ethics & Human
Award program award (UL1 RR024160) from the Research 2013;35(1):1-9; Kim SYH, Schrock L, Wilson RM, et al. An
National Center for Research Resources and the National approach to evaluating the therapeutic misconception. IRB: Ethics &
Human Research 2009;31(5):7-14; Sulmasy DP, Astrow AB, He MK,
Center for Advancing Translational Sciences of the Na- et al. The culture of faith and hope. Cancer 2010;116(15):3702-
tional Institutes of Health. 3711; Singer E, Vonthurn DR, Miller ER. Confidentiality assurances
The authors wish to thank David DeGrazia, Frank and response—A quantitative review of the experimental literature.
Miller, and Luke Gelinas for helpful comments on previ- Public Opinion Quarterly 1995;59(1):66-77; Weinfurt KP. Under-
ous drafts. We also thank Sonali Parnami and Kerry Ryan standing what participants in empirical bioethical studies mean:
Historical cautions from William James and Ludwig Wittgenstein.
for assistance in the development and implementation of
AJOB Primary Research 2013;4(3):49-54.
the survey. 9. Grice P. Logic and Conversation. Studies in the Way of Words.
The studies discussed in this manuscript were reviewed Cambridge, MA: Harvard University Press, 1989, pp. 22-40; Lee CJ.
by and deemed exempt from federal regulations by the Gricean Charity: The Gricean turn in psychology. Philosophy of the

IRB: E t h i c s & H u m a n R e s e a rc h J u ly -A u g u st 2015


17
Social Sciences 2006;36(2):193-218; Sperber D, Wilson D. Pragmat- chanical Turk samples. Journal of Behavioral Decision Making
ics. In: Jackson F, Smith M, eds. Oxford Handbook of Philosophy 2013;26(3):213-224; Berinsky AJ, Huber GA, Lenz GS. Evaluating
of Language. New York: Oxford University Press, 2006; Politzer G. online labor markets for experimental research: Amazon.com’s Me-
Reasoning, judgement and pragmatics. In: Noveck IA, Sperber D, chanical Turk. Political Analysis 2012;20(3):351-368.
eds. Experimental Pragmatics. Basingstoke, Hampshire; New York: 15. See ref. 6, Pentz et al. 2012.
Palgrave McMillan, 2004, pp. 94-116. 16. Willis G. Cognitive interviewing as a tool for improving the
10. See ref. 8. informed consent process. Journal of Empirical Research on Human
11. See ref. 8, Kim et al. 2013; Sulmasy et al. 2010; Weinfurt et Research Ethics 2006;1(1):9-24.
al. 2013. 17. See ref. 16, Willis et al. 2006.
12. Vu H, Kellas G, Metcalf K, Herman R. The influence of 18. Noveck IA, Sperber D. Introduction. In: Noveck IA, Sperber
global discourse on lexical ambiguity resolution. Memory & Cogni- D, eds. Experimental Pragmatics. Basingstoke, Hampshire, UK; New
tion 2000;28:236-252. York: Palgrave Macmillan, 2004, pp. 1-22.
13. See ref. 6, Pentz et al. 2012. 19. See ref. 8, Weinfurt et al. 2013.
14. Paolacci G, Chandler J, Ipeirotis PG. Running experiments 20. Kim SYH, Schrock L, Wilson RM, et al. An approach to
on Amazon Mechanical Turk. Judgment and Decision Making evaluating the therapeutic misconception. IRB: Ethics & Human
2010;5(5):411-419; Goodman JK, Cryder CE, Cheema A. Data Research 2009;31(5):7-14.
collection in a flat world: The strengths and weaknesses of Me-

J u ly -A u g u st 2015 IRB: E t h i c s & H u m a n R e s earc h


18

You might also like