What's bad is easy: Taboo values, affect, and cognition

Judgment and Decision Making, Vol. 2, No. 3, June 2007, pp. 169–188
What’s bad is easy: Taboo values, affect, and cognition
Sarah Lichtenstein and Robin Gregory
Julie Irwin?
Decision Research, Eugene, Oregon
McCombs School of Business
University of Texas
Abstract
Some decision situations are so objectionable or repugnant that people refuse to make a choice. This paper seeks to
better understand taboo responses, and to distinguish choices that are truly taboo from those that are merely dif?cult
or confusing. Using 22 scenarios that describe potentially taboo issues, Experiment 1 explores reasons for disapproval
of the scenarios. We measure a large number of possible reasons for disapproval and a variety of preference responses
(including willingness to accept), in order to test for subtleties in taboo responses. We also test cognitive and affective
responses to the scenarios. Experiment 2 further explores the interaction, found in Experiment 1, between affective and
cognitive factors. Taken as a whole, our results show that people are able to indicate their disapproval consistently across
a variety of preference elicitation methods, that their disapproval is better understood as an attitude measure than as an
economic valuation (even when the measure is in monetary terms), and that taboo responses are driven primarily by
affect.
Keywords: taboo, decisions, tradeoffs, values, risks, affect, cognition, refusals, willingness to accept (WTA).
1 Introduction
tionable decision context leads to such a strong negative
reaction that the individual may refuse to make a choice.
Decision making is characterized by the need to learn
There are several ways in which a person may indicate
about and balance the pros and cons of various options
that a choice is objectionable. One is simply to refuse
and then, with use of this information, make a choice.
to make it. Another is to make a decision but to do so
This process typically involves making tradeoffs across
only after indicating extreme reluctance. In the context
multiple dimensions of value and subjectively weighing
of public choice surveys, for example, this reluctance of-
the anticipated gains and losses. Most people make it
ten takes the form of a protest response to a question
through their day’s normal decisions without a crippling
(Mitchell & Carson, 1989), such as indicating a willing-
degree of effort: we spend money for groceries, we de-
ness to pay millions of dollars for a bene?cial action or
cide on commuting routes and vacation options, we vote
indicating that all the money in the world would not pro-
for preferred candidates in elections, and so on, without
vide acceptable compensation for a loss. Of course, there
much stress.
may be situations in which the risks or costs are so large
Choices that require picking the best option among
(e.g., a substantial decline in the health of one’s children)
several unattractive possibilities, however, typically in-
that most people would never seriously consider the pro-
duce a sizable degree of decision anxiety (Luce, Bettman,
posed tradeoff no matter how large are the bene?ts, but
& Payne, 1997). For instance, many people ?nd it worri-
this stance is especially notable when respondents con-
some to be faced with a choice among alternative cancer
tinue to refuse to make the choice even when the disad-
treatments, a vote on whether nuclear wastes should be
vantage or risk or cost is reduced to a very small level
stored above or below ground, or a decision about euth-
(see Baron, 2001). In still other cases, an individual who
anizing stray cats. Not only do these decisions include
feels that a choice is objectionable might respond but not
unfamiliar tradeoffs, but they may involve moral and eth-
use the desired metric; a response requested using dol-
ical concerns that people often are reluctant to think about
lars or a rating scale might instead be provided in terms
or consider taboo to balance against other issues (Hoga-
of an essay or a picture or a single word. For instance,
rth, 1987). In some situations, the reluctance associated
in surveys asking for dollar willingness to accept com-
with weighing the pros and cons associated with an objec-
pensation for environmental changes, some respondents
?
have written long diatribes in the margins of the survey,
This material is based upon work supported by the National Science
Foundation, Program in Decision, Risk and Management Science, un-
admonishing the experimenters for their insensitivity (Ir-
der Awards No. SES-0114924 and SES-0451259. The ideas expressed
win, 1994, Irwin & Baron, 2001).
in this article are those of the authors alone.
169

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
170
Refusals to make a choice due to the objectionable na-
tions), agent relativity (it matters who makes the trade-
ture of the required value tradeoff have captured the at-
off), and moral obligation (denoting a social, as well as
tention of researchers and writers from many ?elds, in-
personal, reference).
cluding economics, policy analysis, anthropology, the de-
Many choices involve both cognitive and emotional or
cision sciences, and political science as well as psychol-
affective1 dimensions of value, including the typically de-
ogy (Gregory, 2002). Yet there has been little research
sirable choices faced by consumers (e.g., selecting one of
on the important related question of separating what is
two snacks; see Shiv & Fedorikhin, 1999) as well as more
truly taboo from what is instead a choice that we protest
objectionable choices such as those involved in dif?cult
because it is dif?cult or confusing or it makes us angry.
decisions (e.g., selecting non-commodities such as pain;
From the individual’s perspective, however, it is impor-
see Beattie & Barlas, 2001). Luce, Payne, and Bettman
tant that others understand whether one’s apparent refusal
(2001) have argued that, when decision environments are
to make a choice re?ects a deeply held belief or ethical
both consequential and potentially threatening, individ-
standard, with the refusal therefore to be respected be-
uals not only need to trade off the accuracy of decision
cause it is in keeping with the person’s underlying val-
strategies against the required effort (Payne, Bettman, &
ues, or whether it instead re?ects dif?culty or confusion,
Johnson, 1993) but also need to add in the third objective
in which case help may be desired to facilitate engage-
of coping with negative emotions. In such cases, cog-
ment in a consequential personal or social decision. A
nitive activity may have the effect of focusing the deci-
similar distinction exists from society’s perspective, in
sion maker on the facts of the problem instead of on the
terms of whether those refusing to answer questions are
associated emotion; the (typically unstated) assumption
simply exercising their rights or, alternatively, whether
that this is both possible and bene?cial lies at the heart
a poorly phrased or unnecessarily complicated or dif?-
of many problem-solving approaches such as cost-bene?t
cult description or question instead has the (unintended)
analysis, negotiation analysis, and analytically-based per-
result of denying access to individuals who would other-
sonal therapy techniques. The presumption is that think-
wise want to provide input to a dif?cult personal choice
ing may calm the respondents enough that they are able
or a controversial policy debate.
to deal with the negative emotion and form a coherent
Several previous research efforts have laid a sound
response. Yet the negative response itself also may carry
foundation for research on these issues. Philip Tetlock
important information about how individuals perceive the
and his colleagues, for example, have identi?ed a num-
choice and why they would refuse to provide an answer
ber of taboo tradeoffs and value con?icts, referred to as
(or, at least, what is considered to be an acceptable an-
transactions that transgress the spheres of justice (Tet-
swer).
lock, Peterson, & Lerner, 1996; Fiske & Tetlock, 1997).
Although it has been recognized for some time that
Tetlock’s research has focused on the conditions under
there is a link between deliberative and affective capabili-
which people are likely to treat tradeoffs as taboo, typ-
ties (e.g., Wilson et al., 1989), the interaction has received
ically as the result of a request to express something of
only limited examination in the context of response re-
value (such as the concept of freedom) in terms of a fun-
fusals. In this paper we seek to identify the cognitive and
damentally unfamiliar metric (such as dollars), yielding
emotional reasons that could help to identify why some
what he terms a disparate relationship model. A sec-
choices are seen as taboo and others are characterized by
ond research effort, focusing on the concept of protected
mere protest or disapproval. We also seek to study the
values, is most closely linked to Jonathan Baron and his
reasons people give for rejecting a choice in light of the
colleagues (Baron & Spranca, 1997; Baron & Leshner,
interaction between cognition and affect. Our interest in
2000; Irwin & Baron 2001). Protected values are char-
this paper is not directly to provide advice to decision
acterized as those that resist tradeoffs with other values,
makers (although we think our ?ndings are relevant) but
particularly economic values, resulting in rules for de-
rather to understand more fully an individual’s reasons for
cision making that apply irrespective of predicted con-
being reluctant to make a dif?cult choice and, in turn, to
sequences (e.g., “do not eat foods containing modi?ed
learn more about how to separate those choices for which
genes” regardless of what the food is, its relative cost, or
people can be helped to make more informed tradeoffs
which genes were modi?ed). Although Baron’s experi-
from those that are truly taboo.
ments show that individuals may recognize the existence
of multiple dimensions, the concept of protected values
1We use the term affect as referring to people’s positive and negative
asserts that in some cases they may feel that one ele-
feelings toward an external stimulus, such as an activity or a proposed
ment is in?nitely more important than others (Baron &
plan of action. These feelings persist over time as enduring emotional
Spranca, 1997). Baron’s research suggests that protected
dispositions and, in contrast to mood shifts, are a relatively stable aspect
of one’s psychological response (Peters & Slovic, 2000). Cognitive or
values often are characterized by three related properties:
deliberative capabilities, in contrast, are related to the individual’s abil-
quantity insensitivity (one abortion is as bad as 100 abor-
ity to perceive information, hold it in memory, and process it.

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
171
2 Experiment 1
The government is considering a plan to per-
mit logging of old-growth trees on one portion
The goals for Experiment 1 were: (1) to explore whether
of a National Park in order to harvest micro-
mere disapproval can be distinguished from protest and
scopic organisms that grow in their roots, be-
from taboo, (2) to study the reasons for disapproval, and
cause these organisms might lead to new phar-
(3) to study the relationship between reasons and disap-
maceutical products that would help to ?ght
proval.
MS (multiple sclerosis).
Experiment 1 was based on questionnaires that de-
scribed 22 brief scenarios, covering a wide range of po-
The second paragraph followed the ?rst 11 of the ac-
tentially taboo issues, including environmental protec-
ceptability/reasons scales and described in more detail
tion, safety, human and animal rights, and human cell
how the decision would bene?t the participant (unless this
cloning.
Four of the 22 scenarios were designed to
was speci?ed in the ?rst paragraph) and also asked for an
be neutral (non-taboo); these were included as controls,
evaluation of the scenario in terms of a willingness-to-
to test the participants’ ability to make discriminations
accept (WTA) question. For the old-growth trees ques-
among their evaluations. Approval or disapproval was
tion, this second paragraph was:
measured by six different scales, including overall ac-
ceptability and dollar-based scales. Sixteen reasons for
disapproval were presented for each scenario.
The plan will require that the logging company
pay the Park Service for the timber it logs. If
new drugs are successfully developed using the
2.1 Participants
root organisms, then additional payments will
be made to the Park Service by the pharma-
The participants were recruited through an advertisement
ceutical ?rms. These payments would lower
in the University of Oregon student newspaper. Trained
your taxes. In addition, successful treatment of
experimenters waited in a room on campus during the
MS would lower your medical insurance costs.
hours speci?ed in the ad; participants could come in at
What is the SMALLEST savings in taxes and
a time of their own choosing. Participants received an in-
insurance costs each year that you would re-
formation letter and a packet, the ?rst task of which was
quire to approve of this plan?
this experiment. The packet also included a demograph-
ics sheet and several unrelated tasks. Each participant
was paid $10 upon leaving.
Table 1 shows a summary of these 22 stimuli. (The
Of the 254 participants who completed this experi-
scenario shown here is called MSLog.) The Appendix
ment, six were excluded because they gave notably incon-
provides the full two-paragraph stimuli, starting with a
sistent answers at least ?ve times (out of 11) to two items
brief label and ending with a code indicating in which
which both asked for overall evaluations of the same sce-
packet, and in which order, the scenario appeared and a
nario. Notably inconsistent means that they were on the
dollar amount, which will be explained later. (The label,
extreme opposing ends of the scale for the two (func-
code, and dollar amount were not presented to the partic-
tionally equivalent) questions. The remaining 248 par-
ipants.)
ticipants were 126 males and 121 females, plus one who
Eighteen of the scenarios described potentially taboo
did not ?ll out the demographics sheet. Their ages ranged
actions, such as a utility company wanting to reduce ex-
from 18 to 64; the median age was 20.
penditures on pollution controls, a hospital wanting to
harvest organs from patients in a deep coma, and an au-
tomobile company deciding not to ?x a defect in their
2.2 Design
cars in order to keep prices low. The other four scenarios
(these all start with “N” in Table 1) paralleled four of the
2.2.1 Scenarios
bad scenarios but with the taboo aspect removed. For ex-
The stimuli were 22 mini-scenarios, each in two short
ample, an insurance company proposal to charge higher
paragraphs. The ?rst paragraph of each scenario was
rates to blacks than to whites was changed to charging
written in bold type at the start of a page, followed by
higher rates to smokers than to non-smokers.
the acceptability and reasons scales (described later). In
Each participant received 11 scenarios, one per page.
each case, this ?rst paragraph explained a proposal, plan,
Two neutral scenarios always appeared as the second and
or decision that might be viewed as controversial in that
?fth pages, and no participant received both a neutral sce-
it entailed some costs but also would result in some bene-
nario and its bad counterpart. The scenarios were pre-
?ts. For example, one scenario began with this paragraph:
sented in one of two random orders.

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
172
Table 1: Summaries of scenarios used in Experiment 1
BanDrugs
Pharmaceutical company proposes to increase its pro?ts by selling less-developed nations drugs that
are banned in all industrialized countries.
CarCost
Automobile company decides not to recall 1999-model cars because costs of repairs are predicted to
be high in relation to number of lives saved.
CarSuits
Automobile company decides not to recall 1999-model cars because costs of predicted lawsuits due
to injuries and deaths are lower than recall costs.
CloneCell
Genetics company plans to clone human cells to aid serious burn patients.
*NCloneCell
Genetics company plans to grow arti?cial skin in Petri dish to aid serious burn patients.
Coma
Local hospital plans to harvest organs for transplants from patients in deep coma.
Dolphins
Commercial ?shers propose technique to create employment and reduce ?sh prices but would in-
crease dolphin deaths by about 35%.
DrugTest
Hospital decides not to halt a study of a new cancer drug and to continue giving half its patients a
placebo although early results show the new drug saves lives.
*NDrugTest
Hospital proposes giving half its patients a promising new cancer drug and the other half a different
promising cancer drug.
GenMod
World-wide conglomerate to sell genetically modi?ed wheat with added vitamins to African coun-
tries.
GenRsch
Food conglomerate makes proposal to National Science Foundation to do research on genetic modi-
?cation of wheat, leading to more drought-resistant strains.
Highway
State Dept. of Transportation decides not to improve dangerous highway for cost reasons.
*NHighway
State Dept. of Transportation decides not to improve dangerous highway because there have been no
serious accidents.
LifeIns
Insurance company sets different life insurance rates for whites and blacks.
*NLifeIns
Insurance company sets different life insurance rates for smokers and non-smokers.
Military
Congress considers plan to allow military personnel or their families to buy their way out of foreign
military service.
MSLog
Government considers logging old-growth forest to harvest microscopic organisms that might help
lead to new drugs for treating MS (multiple sclerosis).
Pollute
Local utility company seeks permission to save ratepayers money but would increase pollution and
deaths from childhood asthma.
Puppies
State legislature considers whether unwanted puppies and kittens should be sold to experimental labs.
RadioNuc
Congress proposes law to permit higher rates of radionuclide emissions from coal-?red power plants
to lower electricity costs and reduce brownouts.
TreeTrade
Timber company proposes giving up large second-growth forested area near urban center in return
for rights to harvest remote tract of virgin forest.
UnivMonk
University plans to breed monkeys for use in HIV/AIDS research despite lack of appropriate labora-
tory space.
Note. Scenarios shown here are one-sentence summaries of the full two-paragraph scenarios actually seen by partici-
pants, shown in the Appendix. The four neutral items immediately follow their non-neutral versions and are denoted
by asterisks. The names of the scenarios were not shown to participants and are included here to facilitate discussion
in the text.

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
173
2.2.2 Kinds of disapproval
In this task we are asking you for your opinions
about eleven different plans, decisions, or pro-
Fourteen statements or questions about the scenario ap-
posals, each typed at the top of a page. There
peared on the same page as the scenario. Six of these, two
are no right or wrong answers; we are inter-
presented to participants at the beginning (see Table 2,
ested in your opinions. Please take your time,
#1 and #2) and four at the end (#11–#14), concerned the
reading carefully and giving us your thoughtful
overall acceptability of the scenario: Do individuals agree
opinions.
with what is proposed? What are their feelings about it?
Is it right to think about an exchange of this type in terms
Some of these plans may seem somewhat
of dollar values? Except for items #11 and #12, the par-
yucky. Please don’t let your feeling for one
ticipant was asked in each case to circle a number from 1
of them affect your feelings for the next one.
(Disagree) to 5 (Agree).
“Wipe the slate clean” as you turn each page.
The disapproval statements (as shown in Table 2) were
the same for all scenarios and all packets, except that the
2.3 Results
wording of item #12, the WTA item, differed for each
scenario in order to make the question make sense. The
2.3.1 Order effects
exact wording is shown in the Appendix as the second
The effects of scenario order, reasons order, and their in-
paragraph for each item. The response scale for item #11
teraction were tested for each scenario and for each scale
(“All in all, my feelings about this proposal are . . . ”),
(except the WTA scale). In all, 1,716 tests were con-
shown in Table 2, instructed the participant to continue
ducted, of which 8.3% were signi?cant at the .05 level,
to the last three items (which ask about dollar-based val-
but of course these tests were not independent of each
uations) only if their response was negative or undecided
other. None of the order effects was theoretically instruc-
(i.e., “1”, “2” or “3” on the ?ve-point “highly negative”
tive.
to “highly positive” scale).
2.3.2 Kinds of disapproval
2.2.3 Reasons
General acceptability: Three scales, #1, #2, and #11
The middle items on each page were eight (of 16) state-
(see Table 2), re?ected the overall acceptability of each
ments designed to elicit the reasons that participants
scenario and were answered by all participants. There
might ?nd it dif?cult, or be unwilling, to approve of the
was a large range of means for these items across sce-
scenario. Quite simply, these 16 were all the reasons we
narios (#1: 1.52 for CarCost to 4.32 for NCloneCell; #2:
could think of to explain why a person might reject a sce-
1.67 for NCloneCell to 3.98 for CarCost; #11: 1.64 for
nario. The reasons are shown in Table 3. The 16 reasons
CarCost to 4.04 for NCloneCell, all on a scale from 1
were divided into two sets (I and II) such that each set
to 5); note that the best and worst scenarios were the
contained at least one affective reason (e.g., “This pro-
same for all three acceptability measures. Thus, there was
posal disgusts or repulses me.”) and at least one cogni-
discrimination across the scenarios and strong agreement
tive reason (e.g., “I think this proposal is complex, with
among the participants as to which scenarios were good
many aspects to consider, so it’s hard to evaluate.”).2
and bad.
Each reason set appeared in two orders; one order had
Further, we expected that the items constructed to be
the cognitive reasons ?rst and the affective reasons last,
taboo would be rated as less acceptable than the neutral
whereas the other order started with the affective reasons
items. This was also the case: The average acceptability
and ended with the cognitive reasons.
(as measured by the mean of scale 1, scale 11, and the re-
verse of scale 2) of the neutral scenarios was 3.95 versus
2.54 for the taboo scenarios, t = 25.8, p < .0001, for both
2.2.4 Instructions
sets of scenarios combined.
The factorial combination of two sets of scenarios by two
Appropriateness of valuing monetarily: The two ?-
orders of scenarios by two sets of reasons by two orders
nal scales (#13 and #14) concerned the appropriateness
of reasons produced 16 different versions. Each partici-
of getting paid to accept these scenarios. Both used 5-
pant received one version of eleven pages plus an instruc-
point scales to determine participants’ agreement with the
tional cover page. The instructions said:
statements “I don’t think it’s right to put a dollar value
on something like this” (#13) and “You couldn’t pay me
2Tables 2 and 3 show the word “proposal” in several items. This
word was changed as appropriate to “plan” or “decision” to match the
enough to approve of this proposal” (#14). The range of
wording of the scenario.
means across scenarios was 2.69 for NHighway to 4.47

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
174
Table 2: Acceptability Items
Agree
Disagree
1.
I agree with this proposal (circle one number)
1
2
3
4
5
2.
I would not approve of this proposal no matter how high the bene?ts
1
2
3
4
5
·
·
·
·
·
·
·
·
·
11. All in all, my feelings about this proposal are:
Highly Negative
Undecided
Highly Positive
1
2
3
4
5
If you circled “1,” “2,” or “3,” continue to ?ll out the rest of
If you circled “4” or “5,” go directly
this page before going on to the next page.
to the next page.
12. What is the SMALLEST reduction in your yearly cost for electricity that you would require to approve of
this proposal to reduce expenditures on pollution controls?
My saving would have to be at least $_________ per year to approve of this proposal.
13. I don’t think it’s right to put a dollar value on something like this.
1
2
3
4
5
14. You couldn’t pay me enough to approve of this proposal.
1
2
3
4
5
Note. The wording in item #12 is for the Pollute scenario. The lead of this item was different for every scenario; see
the Appendix.
for Military (for #13), and 2.21 for NlifeInsur to 4.18
car A”; another participant left a blank for WTA for Dol-
for LifeInsur (for #14). The mean of 4.47 on Military
phins and wrote, “You could not pay me to increase dol-
for scale #13 indicates virtual unanimity among the dis-
phin deaths”). In addition, one of the experimenters asked
approving or undecided participants, agreeing that they
several participants, after they had completed and turned
“don’t think it’s right to put a dollar value” on the item
in their questionnaires, why they didn’t ?ll in item #12.
under consideration.
Almost always, they replied to the effect that no amount
of money was appropriate. Other participants wrote in
Willingness to accept: Scale #12 was the ?ll-in-the-
ridiculously high amounts of money, up to ten billion dol-
blank WTA question. The instructions made clear that
lars, in response to the WTA question.
this scale (and scales #13 and #14) should not be an-
swered if participants were positive or highly positive in
2.3.4 Construction of the Protest measure
their feelings about the scenario (see Table 2). Across all
data, 31% of all responses fell in this category. (For this
We constructed a Protest measure for WTA to include
and all other analyses, when participants ?lled in the last
all the responses that seemed to be refusals to give a
3 items contrary to instructions, their responses were dis-
WTA amount. This measure included all of the “0” re-
carded.) An additional 23% of the responses were blank
sponses (because it makes no sense to say, for example,
(even though item #11 was scored lower than 4, indicat-
that you have negative feelings about the CarSuits sce-
ing negative or undecided feelings), 7% were “0,” and
nario yet you are willing to buy Car A for the same cost
39% were a positive dollar amount.
as Car B), some of the blanks (depending on the context
for these responses), and dollar responses that were un-
realistically high. Interpretation of the upper limit for an
2.3.3 Additional participant responses
acceptable dollar amount of compensation (above which
The participants were not asked to write comments and
the response is coded as a Protest) necessarily re?ected
there was not much space to do so. Nevertheless, some
judgments of the experimenters, because these amounts
participants did write comments. Almost all of the com-
differed across scenarios. For example, it is at least re-
ments were about WTA and most of them indicated a re-
motely possible that you could save $12,500 on the cost
fusal to answer the question (e.g., to CarSuits one par-
of a $25,000 car but absurd to suppose that your home
ticipant answered “0” and then wrote “I would not buy
electrical bill could be cut by $12,500 per year. The par-

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
175
Table 3: Reasons statements.
Set I:
3/10
Complex: I think this proposal is complex, with many aspects to consider, so it’s hard to evaluate.
4/9
Phony: Asking me to approve or reject this proposal is a phony choice; there is a better way to deal
with this situation.
5/6
Not Me: I don’t feel competent to judge this proposal; somebody else should.
6/8
Future: I worry that there will be bad consequences from this proposal in the future (1, 5, or 50 years
from now).
7/7
Not Trust: I don’t trust the managers of this situation.
8/4
Moral: The proposal is morally or ethically wrong.
9/5
Norms: This proposal violates the norms of our society/culture.
10/3
Disgust: This proposal disgusts or repulses me.
Set II:
3/10
Con?ict: It is hard to evaluate this proposal because its good and bad elements are in con?ict.
4/9
Not Enuf Info: You haven’t given me enough information to make a careful evaluation of this pro-
posal.
6/8
Slippery: This is a slippery slope. Accepting this now will encourage new, worse proposals in the
future. We’ve got to draw the line here.
7/7
Send Message: An important reason for reversing this proposal is to send a message to the people
who proposed it.
5/6
Not Party To: It might be okay for this to happen elsewhere but I don’t want to be a party to it by
agreeing it.
9/5
Uncomfort: It makes me uncomfortable to think about this topic.
8/4
Religious: I have religious objections to this proposal.
10/3
Anger: This proposal makes me angry.
Note. The names of the reasons were not given to the participants; they are included to facilitate discussion in the
text. The numbering shows the two orders in which the reasons were presented. The word “proposal” was changed
to “plan” or “decision” as appropriate. All appeared with a 5-point rating scale (1 = Disagree; 5 = Agree).
enthetical dollar amounts shown at the end of each sce-
Accordingly, the coding rules for construction of the
nario in the Appendix give the upper limits for WTA an-
Protest measure were:
swers; above these limits the response was considered a
Protest. In all cases, these limits were intentionally set
0 if the response to item #11 was positive (4 or
quite high, ranging from $450 (Coma) to $12,500 (Car-
5).
Cost and CarSuits).
1 if the WTA was greater than zero and equal to
Leaving the WTA item blank is an ambiguous re-
or smaller than the upper limit shown in the Ap-
sponse; it might indicate that the participant couldn’t ?g-
pendix, or if WTA was blank and both items
ure out what number to give (e.g., for BanDrugs, one par-
#13 and #14 were 3 or less.
ticipant left it blank and wrote “Need to know more about
the likelihood of litigation arising from this proposal”) or
2 if WTA was greater than the upper limit
it might be a refusal to answer (e.g., for Dolphins one
shown in the Appendix, or if WTA was equal
participant left WTA blank and wrote “I would never ap-
to zero , or if WTA was blank and either #13
prove of this proposal”). The responses to scales #13
or #14 was greater than 3.
(“not right to put dollar values on this”) and #14 (“you
couldn’t pay me enough to approve”) can clarify this am-
Table 4 shows the frequencies and percents for all re-
biguity; agreement with either of them shows a rejection
sponses to the WTA item, for neutral and non-neutral sce-
of the WTA question.
narios separately.

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
176
Table 4: Frequencies and percentages for WTA responses, Experiment 1.
Neutral Scenarios
Non-neutral scenarios
Totals
Subtotals
Totals
Subtotals
f
%
f
%
f
%
f
%
Item #11 > 3
332
67
513
23
Blank
38
8
584
26
Blank protest
24* 5
510* 23
Blank non-protest
14
3
74
3
Zero
18*
4
169* 8
Positive WTA
108
22
966
43
WTA w/in range
92
19
621
28
WTA too high
16* 3
345* 15
Total
496
2232
* Protest = 2
58
12
1024 46
Our primary dependent measure:
Disapproval.
With the ?ve acceptability scales (#1, #2, #11, #13, and
Table 5: Disapproval scores for the 22 scenarios in Ex-
#14) and the Protest measure from WTA, we had six mea-
periment I.
sures of the overall acceptability (or approval) of each
Scenario
Disapproval Scenario
Disapproval
scenario. These six measures were highly intercorrelated.
Using the mean across participants for each scenario, the
CarCost
18.51 UnivMonk
14.46
absolute values of the intercorrelations across scenarios
LifeIns
18.41 TreeTrade
13.18
for the six measures ranged from .86 to .99, with a me-
Pollute
18.21 MSLog
11.20
dian of .95. With such high correlations, it seemed appro-
BanDrugs
17.99 DrugTest
10.82
priate to combine these measures into one Disapproval
Military
17.94 GenRsch
9.26
measure, the simple sum of the other six measures, added
or subtracted as appropriate:
CarSuits
17.76 GenMod
8.73
Dolphins
17.18 NDrugTest
8.17
Disapproval = ?#1 + #2 ? #11 + #13 +
Puppies
15.98 CloneCell
5.66
#14 + P rotest + 9.
RadioNuc
15.68 NLifeIns
4.44
Coma
15.06 NHighway
4.30
We added a constant 9 to the Disapproval measure to
Highway
14.52 NCloneCell
3.85
avoid a meaningless zero point. In calculating the Disap-
Mean
12.80
proval measure, if items #13 or #14 were blank because
the response to item #11 was positive (4 or 5), the par-
ticipant was given a value of 0 on those items. Thus,
the Disapproval measure has a potential range of 0 (most
Table 5 also shows that the four scenarios rewrit-
approving) to +24 (most disapproving). As shown in Ta-
ten to be neutral (NCloneCell, NHighway, NLifeInsur,
ble 5, the actual range of mean Disapproval scores across
and NDrugTest) did have the lowest Disapproval scores,
scenarios is 3.85 to 18.51, with a mean of 12.80 and me-
along with the CloneCell scenario.
dian of 14.48. Only one of the scenarios (which involved
cloning human cells to aid burn patients) was rated as ac-
Analysis of large WTA: As previously noted, many
ceptable by participants; 9 of the 18 taboo scenarios were
participants gave very large WTA responses. Does a re-
rated as 16 (rounded) or higher, in the upper one-third of
sponse of, say, one million dollars indicate greater dis-
the range for the Disapproval measure. Standard devia-
approval than an equally absurd response of $100,000?
tions, not shown, ranged from 5.4 to 8.5. The n’s ranged
Table 6 shows that the answer to this question is, surpris-
from 114 to 124.
ingly, yes. For this analysis large WTA responses were

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
177
Table 6: Mini-disapproval and rated WTA for large dollar-based WTA.
Experiment 1
Experiment 2
Mini-disapprovala
Rated WTAb
$WTA
n
Mean
Median
n
Mean
Median
$1,000
126
–1.12
–2
163
4.63
4
$1,001–4,999
98
–0.84
–1
115
4.50
4
$5,000–9,999
99
–0.11
0
94
3.78
3
5 digits
135
1.10
2
84
3.56
3
6 digits
40
1.03
2
11
3.27
2
7 digits
30
1.43
2
23
1.65
1
8 or more digits
37
1.68
3
15
1.40
1
Totals
565
(53% of positive $WTA)
505
(37% of positive $WTA)
a Mini-disapproval was – #1 + #2 – #11; range –9 for least disapproval to +3 for most disapproval.
b A 1–10 scale was used for Rated WTA, with 1 = Never and 10 = De?nitely willing to accept.
compared with a mini-disapproval score excluding the
items concerned with WTA:
Mini-Disapproval = ?#1 (agree)
+#2 (do not approve)
?#11 (positive feelings)
As shown in Table 6, this mini-disapproval measure
is systematically related to the size of the WTA responses
(the correlation between log WTA and this score for WTA
? $1,000 is .30; p < .0001).
Reasons. We measured participants’ reasons for disap-
proval, with an expectation that the reasons would capture
both cognitive and affective reasons for disapproving of
the scenario.
The means of the ratings (1 = Disagree; 5 = Agree)
for the 16 reasons across the 22 scenarios ranged from
Figure 1: Plot of 22 scenario factor scores on 2 factors,
1.26 (Religious on NLifeIns) to 4.44 (Future on Ban-
Experiment 1.
Drugs and Future and Moral on CarCost). Two of the
reasons scales best discriminated among the scenarios:
Anger (the means ranged from 1.49 for NHighway to
accounts for correlations among the reasons. Each corre-
3.95 for CarCost) and Disgust (from 1.62 for NHighway
lation used the available data from all subjects and scenar-
to 4.20 for CarCost). There were ?ve reasons scales that
ios.) After ?rst examining the data averaged by scenario
the participants did not agree with for any scenario (i.e.,
to see if there were any patterns, we averaged the factor
the largest mean across the scenarios was less than 3.0,
data by participant. This yielded, for each participant, a
the mid-point of the scale): Not Party To, Religious, Un-
mean factor score for each factor, which was then used to
comfortable, Phony, and Not Me.
predict mean scenario disapproval scores. This analysis
allowed us to test whether particular classes of reasons
2.3.5 Factor analysis
best predicted taboo responses.
We conducted a factor analysis on the full matrix of rea-
A principal components factor analysis using an or-
sons ratings for each scenario. (In essence, this analysis
thogonal (Varimax) rotation revealed two factors with

---

Judgment and Decision Making, Vol. 2, No. 3, June 2007
Taboo values
178
eigenvalues greater than 1.0. The analysis produced a
clean solution, accounting for 82.4% of the variance in
the items. Factor 1 included many of the reasons, with
Anger and Disgust leading the list, followed by Not Trust,
Send Message, and Uncomfortable: it is clearly an af-
fective factor. Factor 2 was primarily characterized by
the Con?ict, Not Me, and Not Enuf Info scales; it is
apparently a cognitive factor. One reason scale, Phony,
loaded on both factors; it loaded most heavily on Factor
2. Nine of the 11 reasons loading on Factor 1 and four of
the ?ve reasons loading on Factor 2 have loadings of .80
or greater, showing high intercorrelations of these items
with the factor.
Affective and cognitive factors by scenario. Figure 1
shows the 22 scenarios in the factor space. Note the U-
shape in the ?gure: Scenarios that are judged to be high-
est in affect tend also to be judged easiest to think about.
Scenarios that are judged dif?cult to think about are in-
termediate in affect. Scenarios that are judged lowest in
Figure 2: Slopes showing the interaction between Affect
affect are also judged easy to think about.
and Cognition, Experiment 1.
Regressing disapproval on the cognitive and affective
factors. From the factor analysis, factor scores can be
computed for each individual and each scenario, based on
the responses given to the reason items. Averaging these
factor scores across the scenarios yields, for each indi-
vidual, mean factor scores. Likewise, each participant
can be characterized with a mean Disapproval, the mean
across scenarios. We regressed the mean factor scores on
the mean Disapproval scores, across individuals, to test
whether Disapproval was related to Affect, Cognition, or
both. As Table 7 shows, both Affect and Cognition were
related to Disapproval, in opposite directions: more Af-
fect led to more disapproval, but more Cognition led to
less disapproval.
There was also, for both Reasons sets, a signi?cant in-
teration, as shown in Figure 2. This ?gure shows that
when Affect is low, Cognition has little or no effect on
Disapproval. But for those individuals who showed both
Figure 3: Disapproval versus cognition factor scores
high Affect and low (easy to think about) Cognition, Dis-
across the 22 scenarios, Experiment 1.
approval was especially high — higher than would be
predicted from the main effects of Affect and Cognition.
scenarios are the easiest to think about. For the 13 sce-
This interaction can be interpreted in two ways. One
narios with Disapproval scores above the mean, the cor-
possibility is that Cognition exerts a moderating in?uence
relation between Disapproval and Cognition is ?.80; for
on Disapproval; that is, if participants think hard about a
the 9 low-Disapproval scenarios this correlation is +.82.
strong-affect scenario, they will ?nd it less awful.
The other possibility is that when a scenario strikes a
participant as being really bad, it therefore seems quite
3 Experiment 2
simple; no need to delve into the details because the pro-
posal is just wrong. We will call this the bad-easy hypoth-
We had three goals for Experiment 2: (1) Replicate the
esis. This second interpretation draws some support from
interaction between Affect and Cognition found in Ex-
the relationship between disapproval and cognition across
periment 1. (2) Test whether we can manipulate the mod-
the 22 scenarios, as shown in Figure 3. The best and worst
erating in?uence of cognition, and (3) Remedy a defect

---