Baron, J. & Jurney, J. (1993). Norms against voting for coerced
reform. Journal of Personality and Social Psychology, 64,
Norms against voting for coerced reform
Jonathan Baron and James Jurney
University of Pennsylvania1
Some reforms, such as prohibitive laws or binding agreements to solve social
dilemmas, involve coercion. In hypothetical cases, subjects sometimes said
that they would not support coerced reforms even though they acknowledged that
the reforms would improve matters. Subjects justified such resistance by
noting that the reform would harm some group (despite helping many others),
that a choice would be taken away that people ought to be able to make (i.e.,
that a right would be violated), or that costs or benefits would be distributed
unfairly. These results were found when subjects indicated whether each
justification was true or false, chose justifications from a list, or
provided open-ended responses. Subjects also exhibited a status-quo effect:
they were more likely to vote against a reform than to vote to repeal the same
reforms once it was passed.
1 Norms against voting for coerced reform
Reforms are social rules that improve matters on the whole. Examples of past
reforms are the institution of motor-vehicle laws, the regulation of drugs, the
granting of rights to women and repressed minorities, the recent deregulation
of markets in Communist countries, and, on a smaller scale, countless
minor changes in rules and traditions within smaller institutions such as
schools and businesses. Some reforms require little coercion: those reforms
that involve coordination (Luce and Raiffa, 1957; Ullmann-Margalit, 1977)
require only that the change be made salient. For example, no law or coercion
is required to insure that telephone books and dictionaries put their entries
in alphabetical order. In coordination, it is to the advantage of all to
follow the rule. Other reforms that require little coercion are those that
increase the options available to each person, as in the market reforms in the
Other reforms require some coercion. The classic cases are those that involve
social dilemmas (Dawes, 1980; Schelling, 1978 - also known as commons dilemmas
[Hardin, 1968] or n-person prisoner's dilemmas [Luce & Raiffa, 1957]). In
these cases, as we define them, each person is faced with a conflict between
options, one of which is better for the individual and one of which is
better for all of the members of the group in question. Examples of unsolved
social dilemmas are those that result from excessive childbearing in some
countries, cutting of trees for fuel, and production of gasses that cause
global warming. Many social dilemmas have been solved (sometimes only
partially), by rules or laws that effectively make the selfish response
unrewarding. Examples include laws against tax evasion, treaties
concerning the use of fisheries, rules that make sure that employees do their
jobs (instead of shirking), laws regulating medical drugs (and thus removing
the defections that used to occur in the form of `snake oil'), and compulsory
Social dilemmas are not the only cases in which coercion may be required.
Often, standard bargaining situations (Nash, 1950; Raiffa, 1982) lead to
breakdowns of cooperation (Elster, 1989). In these situations, each of several
agreements (rules or allocations) would be better than the status quo for all
parties, but the parties involved cannot agree on which to choose. Thus, state
coercion is sometimes required to resolve strikes, even though resolution would
be best for all. In still other cases, coercion is required to bring about an
improvement for many at the expense of a few, as when taxes are raised for the
A characteristic of many of these reforms requiring coercion is that it is in
the interest of most people to support the reform, even if it is not in their
interest to cooperate in the absence of coercion (Yamagishi, 1986, 1988). We
can think of the individual's utility for supporting a reform as a function of
the payoffs for cooperation or defection with and without the reform and the
effort involved in supporting the reform. Supporting the reform is often
worthwhile because the reform can increase the number of other cooperators,
which, in turn, increases the payoff for cooperation (and possibly even
the payoff for defection, despite the increased cost of defection relative to
cooperation). The gain to the individual from this increase is often greater
than the the loss from the cost of supporting the reform plus the loss from the
individual's switch from defecting to cooperating (if this switch is made) or
the loss from the penalty for defection (if the individual defects regardless
of the reform). For example, alcohol drinkers might benefit from supporting an
increased tax on alcohol, because of the reduction in drunk driving (by
others), improved government services, etc., whether or not they continued to
drink and pay the new tax. The cost of support is often very small
(Petit, 1990), involving only the expression of approval or disapproval. In
sum, self-interested individuals often have good reason to support compulsory
reforms even when they do not have sufficient reason to cooperate
spontaneously. When enough individuals act according to this reason, a reform
can be initiated and maintained.
So far, we have limited the discussion to self-interested reasons. Clearly,
some moral norms will strengthen the side of cooperation. The cost of
supporting a reform might, for example, be seen as no cost at all to a person
motivated by a desire to promote the general good. People motivated in this
way would support reforms even when the support is costly. Those who initiate
reforms are often in this category, as considerable effort is required to bring
a reform to the point at which it may be instituted with only small amounts of
effort on the part of most people (Elster, 1989).
Elster (1989) has also argued that social norms, including what we are calling
moral norms, can have deterimental effects. For example, a norm of retribution
or revenge can lead to escalating conflict (as argued as well by Frank, 1985).
Such norms can be seen as crude rules that are usually beneficial but
are misapplied (or that have outlived their usefulness) because they are not
understood (Baron, in press b). In this paper, we ask about the existence of
norms that are recognized by their supporters to have such detrimental effects
on the initiation of reforms.
One such set of norms concern fairness in the distribution of the benefits
or costs of reform. People may reject a generally beneficial reform on the
grounds that it helps some more than others, or hurts some in order to help
others, in a way that is unfair (Elster, 1989). Sometimes this problem can
be remedied by modifying the proposed reform so that it is generally seen as
fair, e.g., by compensating workers displaced by a tariff reduction.
Sometimes, however, the `losers' are difficult to identify. More often,
different standards are available for fairness (Baron, in press a), and each
group tends to apply a standard that provides its members with a better outcome
(Cook & Yamagishi, 1983; Elster, 1989; van Avermaet, 1974, cited in Messick,
1985). For example, in deciding how to allocate the right to produce carbon
dioxide in an international treaty on global warming, the United States is
likely to favor a formula based on past behavior, while China is likely to
favor a formula based on population. In such cases, a majority may oppose
the reform even when all could benefit.
Some people hold a strong norm prohibiting helping one person through harming
another (Baron, in press b; Ritov & Baron, 1990; Spranca, Minsk, & Baron, 1991,
Experiment 3), even if the benefit outweighs the harm and even if unfairness is
not at issue (e.g., when those to be harmed are determined randomly). The
difference between this harm norm and the fairness norm may be in
the choice of reference point. Those who oppose reforms on grounds of fairness
may see people as being harmed relative to a reference point defined by the
ideally fair result. Those who oppose reform on simple grounds of harm use the
status quo as a reference point.
Related to the norm against harm is a norm favoring rights. In this
context, a right is an option to defect. The removal of this right might be
seen as a harm, even if, on other grounds, the person in question is clearly
better off when the option to defect is removed (because it is also removed for
All of these norms - fairness, harm, and rights - can involve a kind of framing
in which the status quo is given priority. One trouble with most reforms is
that they help some people and hurt others. For example, an increased tax on
gasoline in the U.S. may help the whole world by reducing CO2 emissions,
and it will help most Americans by reducing the budget deficit. But it will
hurt those few Americans who are highly dependent on gasoline, even when we
take the other benefits into account. Congress might try to craft some sort of
compensation for these people, but it is difficult to target them accurately.
If the reform were already in effect and the question were whether to undo it,
the same considerations could favor keeping the reform. Repeal would hurt some
people while helping others. Thus, as we discuss later, the norms in question
could constitute the psychological basis for a status-quo effect for
institutional changes, analogous to the effect found for individual
decisions (Samuelson & Zeckhauser, 1988; Ritov & Baron, in press).
In the present paper, we report three initial studies of the opposition to
reform. We use questionnaires about hypothetical situations, some based on
real situations. We seek cases in which subjects agree that a reform will
improve the situation but still oppose the reform. We question subjects about
reasons that could underlie these discrepancies, seeking evidence for norms
of fairness, harm, and rights.
2 Experiment 1
Fifty-one subjects, paid $5 per hour for completing this questionnaire and
others, were solicited by a sign on a prominent walkway of the University of
Pennsylvania. Almost all subjects were college students.
Each of the six cases in the questionnaire (available from the first author)
described a proposal for coerced reform, along with the rationale for the
1. TV ads. A law `would make it illegal for politicians to use television
commercials for campaign purposes.' This would save money and fundraising
time. The subject was asked to take the position of a member of congress.
2. Vaccine. A law would require vaccination of the entire U.S. population
against an epidemic flu. Vaccination would reduce the incidence of flu from
20% (caused by wild virus) to 10% (caused by the vaccine). Vaccinated
individuals cannot transmit flu to others, although nonvaccinated individuals
can do so even if they do not display symptoms.
3. Mouthwash. An epidemic of a flu-like bacterial disease hits a college
campus. Transmission of the disease is prevented by `gargling with a very
unpleasant-tasting mouthwash five times a day' before symptoms start. A rule
would require every student to have a blood test for infection, with those
infected required to use the mouthwash.
4. Auto. A no-fault automobile insurance law would eliminate the right to sue,
reduce insurance rates, and increase the reliability of damage settlements
(while limiting their amount). The subject was asked to take the role of a
5. Doctors. A law would make it impossible to sue an obstetrician.
Obstetricians would still be subject to disciplinary procedures. The law would
reduce medical costs, increase the number of obstetricians, and have no
effect (according to experts) on the quality of care. The subject was asked to
take the role of a prospective parent.
6. Gasoline. To combat global warming, a gasoline tax would double the price
of gasoline. There is no alternative: it is this tax or no tax.
After each case, the following questions were asked (using case 1 as an
A. Would you vote for the proposed law?
B. If no, would you vote to repeal the law if it were already in effect?
C. Do you think that the law violates anyone's rights? If so, what rights?
D. Do you believe that congresspeople would be better off with the law?
E. Without the law, would you be better off by refusing to advertise on TV?
F. Do you think that the law would make some people worse off as soon as the
law went into effect? If so, whom?
G. Is the law fair? If not, why not?
These questions were suitably modified for each case (e.g., `rule' instead of
`law') as necessary. Question D determined whether subjects accepted our
(implicit) assumption that the law would improve the situation. Note that
subjects could answer this question negatively simply because the law
violated someone's rights, because it harmed someone, or because it was unfair,
even if they thought that the law would increase overall welfare or utility.
By using this question to select subjects who vote against the law despite the
fact that they think it improves the situation, we provide a conservative
measure of the number of such subjects.
Question B looked for an explicit status-quo effect, in which the subject would
admittedly favor the status quo whether the rule was in effect or not.
(Previous studies of the status-quo effect have used designs in which different
conditions were either given to different subjects or presented to the same
subjects with many intervening items, so we would not necessarily expect
a status-quo effect here.) Question E determined whether subjects saw the case
as a true social dilemma. Questions C, F, and G posed the basic questions at
issue: rights, harm, and fairness, respectively.
Table 1 shows the percent of affirmative answers to each question. In general,
subjects were moderately in favor of the proposals (question A), but a
substantial number voted negatively. For all six cases, more subjects thought
that the proposals would improve the situation (question D) than would vote for
the proposals. The differences between the percents answering affirmatively to
questions A and D ranged from 5% for Case 5 to 22% for Case 3. (The number of
cases is too small to carry out meaningful analyses of the determinants of
these differences among cases.) In 13.6% of all cases, D was answered yes and
A was answered no, and, in 2.7%, D was answered no and A was answered yes.
Although the difference between these proportions was significant only for
Cases 2 and 3, 56.9% of the subjects gave more yes answers to D than to A
(across the 6 cases), and 9.8% did the reverse. This difference
was significant (p=.000, sign test), and it remained significant (37.3% vs.
9.8%, p=.007) when only those items for which question E (spontaneous
cooperation) was answered negatively were counted. Thus, the difference cannot
be accounted for by the belief that rules are unnecessary because people will
Percent of subjects answering each question affirmatively (N=51). Question
B is based only on subjects who would vote against the law or rule in question
A B C D E F G
Vote Repeal Rights Better Coop. Harm Fair
1 TV ads 53 83 67 59 8 78 57
2 Vaccine 69 72 61 80 91 75 67
3 Mouthwash 66 60 49 88 73 41 65
4 Auto 66 65 45 74 12 80 61
5 Doctors 49 83 65 54 12 69 50
6 Gasoline 47 63 35 58 64 88 51
As shown in Table 1, some people who voted against the law or rule would not
vote to repeal it if it were already enacted, suggesting a possible status-quo
effect. This issues, among others, is addressed further in Experiments 2 and 3.
Table 2 shows phi coefficients for various predictors of the answer to question
A (Vote) in subjects who answered yes to question D (Better). Yes answers are
coded as 1, no answers as 0, so the hypothesized coefficients are negative
for all predictors except G (Fair), where the subject was asked if the proposal
was fair rather than unfair. The variable T is included to permit an
overall test of the role of harm, rights and fairness; T is the sum of the yes
answers to questions C (Rights) and F (Harm) and the no answers to G (Fair).
(A subject who answered yes to C and F but no to G would get the maximum score
of 3 on this variable.) Significance levels are from Fisher's exact test
(except for T, where they are from one-tailed Mann-Whitney U tests comparing
those who answered yes and no to A).
Phi-coefficients and significance levels (Fisher's exact test for C, F, G,
and E; one-tailed Mann-Whitney U test for T: *, <.05; **, <.025, ***, <.01;
****, <.001) for the association between question A and each of questions C, F,
G, and E, and derived score T, for the six cases. All data are from
subjects who answered yes to question D.
C F G E T
Rights Harm Fair Coop. (C,F,G)
1 TV ads -.45** -.27 .81**** .09 -.89****
2 Vaccine -.25 .01 .67**** .15 -.62***
3 Mouthwash -.57**** -.08 .68**** .21 -.69****
4 Auto -.20 .05 .53*** -.16 -.51*
5 Doctors .19 -.13 .34 -.53*** -.18
6 Gasoline -.37* -.23 .32 .16 -.54**
The summary variable T was a significant predictor for all but Case 5. Belief
in spontaneous cooperation (E) generally showed nonsignificant positive
correlations, except for Case 5. In general, harm (F) did not play a
significant role, but rights (C) and fairness (G) often did. The role of
rights in Cases 1 and 3 did not seem to depend on the correlation between
rights and fairness (Case 1, phi=-.53; Case 3, -.51): subjects who thought the
proposal was better (D) and fair (G) almost always voted for it; but, among
subjects who judged the proposal better but not fair, the correlation between C
(Rights) and A (Vote) was no less than that shown in Table 2 (Case 1, phi=-.64,
N.S.; Case 3, phi=-.83, p=.011). The correlations taken together support
the hypothesis that subjects sometimes vote against proposals they perceive as
beneficial because the proposals are unfair or violate rights (depending on the
case). The role of harm is unclear.
Answers to question D (Better) were highly correlated with answers to question
A (Vote) within each case (even when answers to question G [Fair] were held
constant: 10 of the 12 phi coefficients were significantly positive). Answers
to question D were themselves predicted significantly by the fairness
question (G) for all cases except Case 3, the rights question for Case 1 and
Case 5, and the harm question for Case 2. (The cooperation question, E, did
not predict D except for Case 5, where the correlation was positive.) Question
D might not have been answered in strictly utilitarian terms; rather, it might
have been affected by the same judgments that determined voting. Our results
might have been stronger if judgments of benefit were made in terms of total or
average utility alone.
These results are consistent with subjects' written answers in this experiment
and in a pilot study with somewhat different scenarios. For example: `I would
not support this law because it would give up my right to sue. I would
hate to think that I could be seriously injured and not be
recompensed suitably.' `I would not support a law that would take away
my right to sue an obstetrician who was proven to be negligent no matter how
much bad publicity or discipline that doctor received.' [From a scenario
concerning Chinese-style family limits in Kenya,] `It is my right as a human
being to have as many children as I please without being ostracized for doing
5 Experiment 2
In Experiment 2, we presented modified forms of the same cases. Some
modifications were designed to emphasize that the reforms in question were
improvements. Also, instead of asking subjects to answer various questions and
then correlating the answers with their support of the proposal, we asked
subjects directly about their reasons for opposing the proposal, when they did
oppose it. And we reworded the questions for clarity. The item
concerning whether the subject would be `better off' cooperating was changed to
emphasize the fact that some people might cooperate spontaneously, and expect
others to do so, even if they do not see themselves as better off (Yamagishi,
1986); this new item asked simply whether the law was needed to induce people
to change their behavior.
We ran three other conditions in order to test two additional hypotheses.
First, we wanted to determine whether resistance to coerced reform had the
property of a status-quo effect. In a `vote-after' condition, subjects were
asked to imagine that the reforms had already been enacted, and subjects were
asked whether they would vote to repeal the reforms. We hypothesized
that subjects would vote to repeal the reforms less often than subjects would
oppose the reforms in the original `vote-before' condition (taking into account
their judgments of whether the reforms were improvements).
The basis for this hypothesis was that norms used as arguments against coerced
reform could be construed as biased toward the status quo. These norms would
dictate opposition to the reform, but, if the reform were already instituted,
they might not dictate support for its repeal. In the case of the norm
against harm, someone will be harmed by the reform, but, if the reform were
already in effect, others would be hurt by its repeal. (This must be true if
the reform is beneficial overall.) In the case of the fairness norm, some
subjects might think of the changes as unfair rather than the resulting
distribution of goods. Indeed, the question was worded with this in
mind: `unfairly distribute the costs of change.' In such cases, reversing the
change, if the proposal were already in effect, might be seen as equally
unfair. Rights, too, could conceivably argue for the status-quo if subjects
thought that whether a given choice was one `that people ought to be able to
make' depended on people's expectations about what choices they were allowed
The second additional hypothesis was that the tendency to vote against reforms
seen as beneficial would also be found when the judgments of overall benefit
was made by different subjects than those who voted. We therefore asked two
other groups of subjects, the `judge-before' and `judge-after'
groups, respectively, to indicate only whether the reforms did more harm than
good. No voting was involved. The judge-before group assumed that the reforms
had not been enacted (as did the vote-before group), and the judge-after group
assumed that the reforms had been enacted (as did the vote-after group).
This comparison provided an additional test of the basic hypothesis that people
will vote against proposals that they themselves see as beneficial.
Forty-five subjects were first solicited as in Experiment 1 for the vote-before
questionnaire alone. Then an additional 101 subjects were solicited for the
four conditions. Subjects were assigned to conditions in approximate rotation
as the subjects arrived (depending on the availability of the
different conditions). Of these additional subjects, 24 did the vote-before
condition, 26 the vote-after condition, 23 did the judge-before condition, and
28 did the judge-after condition. The original 45 subjects in the vote-before
condition did not differ significantly on any measures from the new group of
24, so these two groups were combined for analysis, yielding 69 subjects in the
The cases were similar to those used in Experiment 1. As noted, they
emphasized the benefits of the proposals. After each case, subjects were asked
the following questions (using case 1 as an example):
Would you vote for this law?
If you would not vote for the law, indicate why not. Write the letters of all
the reasons that apply.
A. The law would do no more good than harm.
B. The law is not needed. Once people understand the situation, they will
change their behavior without a law.
C. The law would make some group worse off than they were before the law.
D. The law would take away a choice that people ought to be able to make.
E. The law would unfairly distribute the costs of the change. That is, some
people would suffer more than they should, relative to other people.
F. Any other reason not listed (explain).
Most cases in which the subject gave an additional reason (F) were recoded so
that they accorded with other responses, according to their spirit. In many
cases, subjects apparently desired simply to say things in their own words.
Responses indicating that the reform might make the situation worse were
coded as yes responses to A (e.g., `There may be a decrease [in quality of
care] with no liability'), and responses indicating that someone would be
unfairly rewarded were coded as E (e.g., `Drunk drivers would clearly be at
fault and should be responsible for damages incurred due to their
recklessness'). The few cases that could not be recoded were comments that
did not bear on the reasons for the subject's response (e.g., `The current
situation [concerning TV advertising] is only generating economic activity in
the advertising industry') or (in 5 cases) denials of the assumption
(`Pennsylvania, like California, can finally stand up to insurance companies
and impose ceilings on rates,' which denies that the proposal given is the only
chance for change). Over both the vote-before and vote-after conditions, 33
answers to F were recoded as A responses (a conservative coding, given our
hypothesis). The remaining were: B, 2; C, 6; D, 11; E, 17. (Coding was done
by J.B., whose reliability on such coding will be reported in Experiment 3.)
The vote-after condition was modified so that the law or rule had already been
put into effect, and the question was whether to repeal it. The descriptions
and questions, including the questions about reasons, were modified
accordingly. For example, the first item read, in part:
It is 1995. Until 1992, every election year, candidates for congress spent
most of their campaign money on television advertising ...
A law was passed in 1992 that made it illegal for politicians to use television
commercials for campaign purposes. If this law is repealed, the situation
would go back to the way it was before 1992, and no other change will be made
in the foreseeable future.
You are a candidate for congress.
A. Would you vote to repeal the law?
B. If you would vote to repeal the law, indicate why. Write the letters of all
the reasons that apply.
a. The law did no more good than harm.
The two judgment conditions, judge-before and judge-after, used the same
wording as the respective voting conditions, but the subjects were asked only
whether the law or rule `will improve matters overall' (in the judge-before
condition) or whether it `improved matters overall' (in the judge-after
Table 3 shows the frequencies of each kind of response, in percent. The first
column shows the percent of subjects who would vote for the proposal. The
second shows the percent of opponents who thought that the proposal would not
improve matters. The columns labeled B through E show the proportion
of subjects giving each reason as a percent of opponents who thought that it
would improve matters (as indicated by negative answers to question A). The
number of these subjects is shown in the rightmost column (N).
Percent of subjects answering each question affirmatively in Experiment 2.
For Question A, the reference sample is the number who would oppose the law.
For questions B-E, the reference sample is the number who would not vote for
the law and who answered question A negatively; the number of subjects (N) in
the reference sample is shown at the right.
A B C D E
Enact Bad Coop. Harm Rights Fair (N)
1 TV ads 67 39 21 57 64 57 (14)
2 Vaccine 42 25 33 27 90 30 (30)
3 Mouthwash 70 48 54 27 82 36 (11)
4 Auto 58 39 6 59 71 59 (17)
5 Doctors 58 52 15 53 69 46 (13)
6 Gasoline 39 52 30 75 35 85 (20)
Not A B C D E
repeal Bad Coop. Harm Rights Fair (N)
1 TV ads 80 40 33 0 100 67 (3)
2 Vaccine 65 22 57 57 86 43 (7)
3 Mouthwash 81 20 50 0 50 75 (4)
4 Auto 58 27 37 62 75 37 (8)
5 Doctors 64 56 0 25 75 0 (4)
6 Gasoline 74 33 0 75 0 100 (4)
'Good` for judge-before and judge after
1 TV ads 78 79
2 Vaccine 96 86
3 Mouthwash 87 82
4 Auto 68 72
5 Doctors 83 61
6 Gasoline 24 46
The A answers come nowhere close to 100%. Subjects often oppose proposals
despite the belief that they would improve matters. Forty-five out of 60
subjects in the vote-before condition would not vote for at least one proposal
that they considered beneficial. All three hypothesized explanations of
this opposition - rights (D), unfairness (E), and harm (C) (in contrast to
Experiment 1) - were chosen as justifications for opposition to a reform. The
belief that coercion is not needed (item B) also plays some role. (Items
should not be compared directly to those in Experiment 1, since their wording
To test for a status-quo effect, we examined the effect of status-quo
(vote-before vs. vote-after) on the proportion of supporting votes, with
subject as the unit of analysis, partialling out the proportion of cases in
which the proposal was judged better (assuming that subjects would judge the
proposal better if they voted for it - recall that they were not asked
for judgments in this case). The effect of status-quo was significant
(t(93)=2.11, p=.038; adjusted mean proportions of votes for the proposal: .576
[s.e.=.022] in vote-before, .667 [.037] in vote-after), although the effect was
The hypothesis of a group difference between willingness to vote and judgment
of a proposal's value was also supported. In particular, the proportion of
votes for proposals in the vote-before condition was less than the proportion
of judgments that the proposal would help in the judge-before condition (t=3.38
across subjects, p<.001). The judge-before and judge-after conditions did not
differ. The vote-after condition did not differ from either judge-before or
judge-after; although this result suggests that the reluctance to support
coercion is limited to putting proposals into effect, the failure to find
an effect here could be due to the insensitivity of between-subject comparison.
Experiment 3 will address this issue again.
It seems that the status-quo effect concerns voting rather than judgment.
Judgment is, it seems, more driven by comparison of outcomes, as opposed to the
means by which outcomes were achieved (action vs. inaction). Judgment and
choice may differ in matters of public policy as they do matters of individual
decision making. In fact, some of the examples used by Tversky, Sattath, &
Slovic (1988) to demonstrate the inconsistency between judgment and choice
concern public policy, although we doubt that the present results can be
explained in terms of a prominence effect (in which choice, but not judgment,
is largely determined by the most `prominent' dimension).
8 Experiment 3
The categories used for multiple-choice responses up to now were based on pilot
studies and on our own analysis. Experiment 3 asked subject directly for their
reasons for discrepancies between voting and value judgments, without prompting
them with reasons that they might not have given (or thought of) on their own.
In addition, we counterbalanced the order of these two questions, and
re-examined the status-quo effect (vote-before vs. vote-after). The status-quo
effect in Experiment 2 could result from the subjects being influenced by the
fact that the proposal had already been approved by others in the vote-after
condition (although clearly it was still controversial enough that a repeal was
being considered). Social comparison theory, for example, would suggest that
people use the opinions of others as cues to their own opinion (Festinger,
1954). To reduce such influence, the vote-after condition stated that the vote
had been close the first time and that the proposal had not gone into effect
yet. Both conditions stated that the upcoming vote (to pass or repeal the
proposal) was expected to be close.
The cases were the same as in Experiment 2, except that: dates were specified
to indicate that the proposal never took effect in the vote-after condition;
and the descriptions specified that all votes were `close.' For example, the
vote-after form of the first case read: `It is 1993. ... A law was passed in
1992, by a very close vote, that made it illegal for politicians to
use television commercials for campaign purposes. The law is scheduled to take
effect in 1994. Opponents of the law now want to repeal it, and a second vote
has been scheduled. It is clear that the vote will be close again. If this
law is repealed, no other change will be made in the foreseeable future.'
The corresponding description for the vote-before condition read: `It is 1993.
... A law is proposed that makes it illegal for politicians to use television
commercials for campaign purposes. The law is scheduled to take effect in
1994. It is clear that the vote will be very close. If this law is not
passed, no other change will be made in the foreseeable future.'
In the first of four conditions, subjects were asked: `A. Would you vote for
the law?'; `B. Would letting such a law take effect do more good than harm on
the whole?; and, 'C. If you gave different answers to A and B, please explain
your answers.` The four conditions differed in the order of questions A and B
and (orthogonally) in whether question A was a vote to initiate (vote-before)
or repeal (vote-after) the law or rule.
Subjects were solicited as in Experiment 2, except that they were paid $6 per
hour. The number of subjects in each of the four conditions was: 62,
vote-before, vote first; 62, vote-after, vote first; 63 vote-before, judge
first; 62, vote-after, judge (total=249). Thirteen additional subjects were
omitted for failure to follow instructions or apparent misunderstanding.
Table 4 shows the percent of all subjects who supported each proposal, either
voting in favor of it or not voting to repeal it, and the percent who thought
it was better on the whole. There were more cases in which a subject supported
a proposal but voted against it (10.0% overall) than the reverse (2.2%; p=.01
or better for each of the six cases by a sign test, p=.000 overall by sign test
across subjects). The order of asking the questions did not affect this
difference (t(247)=0.69, across subjects). The general finding of reluctance
to vote for proposals judged beneficial is therefore supported.
Percent of all subjects supporting each item through voting and judging it
better, and the percent (of those who oppose a proposal that they judge better)
who give each justification for voting against it in Experiment 3.
Justifications for oppose/better
Support Better Harm Rights Fair Self Deny Unclear
1 TV ads 66 72 4 20 0 36 4 36
2 Vaccine 77 82 0 38 6 13 6 38
3 Mouthwash 75 82 0 57 5 0 5 33
4 Auto 54 61 8 69 8 0 0 15
5 Doctors 59 69 7 44 7 7 4 30
6 Gasoline 60 69 19 0 22 9 16 34
To test for a status-quo effect, we examined the effect of status-quo
(vote-before vs. vote after) on the proportion of supporting votes, partialling
out the proportion of cases in which the proposal was judged better. The
effect of status-quo was again significant (t(246)=2.14, p=.033; adjusted
mean proportions of votes for the proposal: .629 [s.e.=.014] in vote-before,
.672 [.014] in vote-after), although the effect was small. The reluctance to
vote for proposals judged beneficial was still highly significant (p=.000, sign
test comparing voting and judgment) in the vote-after condition, so this
reluctance does not appear to be limited to bringing proposals into effect.
To examine subjects spontaneous justifications, the first author (J.B.) read
the answers, devised additional scoring categories as needed, and assigned each
answer to a category. To check reliability, a second scorer (J.S.) was asked
to categorize all responses. The definition of each category is followed here
by an example (not given to the scorer). Each justification was assigned to
only a single category, representing the dominant justification, although a few
responses might have included more than one response category.
HARM: The reason against the law is that it causes harm to someone. Explicit
mention of the fact that the benefit to some does not justify the harm to
others is acceptable but not required for this category. - 'Though it has
overall good effects, this law would leave many people helplessly stranded
if suddenly they could not afford to buy gas, or buses were forced to stop
running because of this new tax.`
RIGHTS: The law would take away someone's right. Or the authority in question
does not HAVE the right to impose such a law. ... A violation of rights is not
seen simply as harm but rather as taking away a choice that someone should be
able to make. ... References to freedom and conflict with religion go here. -
'Those who are deserving of a larger settlement would not even have the
opportunity to get it. We can't compromise the rights of the minority.`
FAIRNESS: The distribution of costs and benefits is unfair. It is not just
that some are being harmed but also that the harm is unfairly distributed.
References to justice and deservingness are included here. - 'I wouldn't vote
for it because it favors the affluent, who would not mind the tax increase as
much as the penurious.`
SELF-INTEREST: The subject votes on the basis of self-interest, which is
perceived as being different from the interest of others. This is not the same
as someone who puts themselves in the position of someone who might be harmed
... - 'I would vote no because I would want to be able to advertise my
DENIAL: The subject denies the assumption that the law in question is the only
choice. - 'Rather than banning ads altogether, ... I would impose a maximum
air-time for each candidate.`
UNCLEAR: Subject does not answer question, or gives an answer on the other
side (e.g., justifying the law, when the subject voted against it), or says
that the law would not really do more good after all. - 'I tend not to think of
the long range effects of something. I would probably only see the
The reliability of scoring, estimated by Cohen's kappa, was .70 overall.
Reliability for the individual categories was: Harm, .45; Rights, .83;
Fairness, .61; Self-interest, .84; Denial, .53; Unclear, .67. The only
systematic discrepancy between scorers was that five cases classified by J.B.
as harm were classified by J.S. as fairness. Examination of these cases
suggested that the distinction is truly difficult to make, because it
requires deciding whether a harm is bad for its own sake or because of the way
in which harms are distributed, e.g., 'Sure [a fuel tax] would encourage
development of something else, but what would middle-class people do until
then?` J.S.'s ratings were used because they were all made after the
categories were defined.
Table 4 shows the number of subject giving each justification. The results are
in general agreement with Experiment 2: rights predominated for all cases
except for Gasoline, where harm and fairness predominated (putting aside the
last three response categories). No subject spontaneously said that people had
a right to burn fossil fuels, but some subjects thought that the fuel tax would
be harmful or unfair to some people. The larger number of Self-interest
responses to the TV-ads item reflected subjects' belief that they would use TV
better than their opponent. The larger number of denial responses to Gasoline
may reflect the ease of imagining alternative solutions to the problem. Note
that responses that referred to spontaneous cooperation were simply not found:
no subject said that the proposal was unnecessary because people would
We have found a partial dissociation between willingness to vote for a proposed
reform and judgment of whether the reform is truly for the better. In
declining to vote for a reform that they judge to be beneficial, subjects
appeal to moral norms of fairness, harm avoidance, and the individual's right
to choose. By the subjects' own judgment of the public good, these norms stand
in the way of maximizing that good. Our findings are therefore consistent with
recent theorizing on the possible detrimental effects of norms (Elster, 1989).
The effects we found are small, but we might have underestimated them (as
noted earlier) if judgments of benefit were also affected by the same norms.
Moreover, even small effects could affect support for public policies.
The norms in question are associated more strongly with action than with
inaction, although they act to encourage repeal of proposals as well as to
discourage their passage. We suggest three explanations of this kind of
omission bias: First, some people may think of the norms in question as
prima facie prohibitions against bringing about harm, unfairness, or
rights violations through action, and not so much as an injunction to eliminate
unfairness or harm or to promote rights. Second, the status-quo (here, the
result of inaction) could affect the way in which harm and unfairness are
evaluated: a loss could be considered more harmful than a forgone gain; and
inequities could be seen in changes from the status-quo rather than in
ultimate levels (Baron, in press b). Third, people may think of unfairness not
as a consequence of action (or inaction) but rather as a kind of action, as in
the view of justice as 'procedural` (Tyler, 1988).
A simple account of our results would suggest that people are utilitarians when
they judge whether one social situation is better than another but they are
deontologists when it comes to their own decisions to vote. This is surely too
simple. Voting and judgments often agree. People may base their voting on
their judgment of the total good. And their judgment of the total good need
not be based entirely on their judgment of total utility: they may well take
fairness, rights, and changes of state into account as well. None the less,
the simple account appears to explain the difference between voting and
judgment that we have found. Of course, the differences would be reduced if
people voted according to the same standard.
Baron, J. (in press a). Postscript. In B. A. Mellers and J. Baron (Eds.),
Psychological perspectives on justice: Theory and applications. New York:
Cambridge University Press.
Baron, J. (in press b). Heuristics and biases in equity judgments: a
utilitarian approach. In B. A. Mellers and J. Baron (Eds.), Psychological
perspectives on justice: Theory and applications. New York: Cambridge
Cook, K. S., & Yamagishi, T. (1983). Social determinants of equity judgments:
The problem of multidimensional input. In D. M. Messick & K. S. Cook (Eds.),
Equity theory: Psychological and sociological perspectives, pp. 95-126.
New York: Praeger.
Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31,
Elster J. (1989). The cement of society. Cambridge: Cambridge University
Festinger, L. (1954). A theory of social comparison processes. Human
Relations, 7, 117-140.
Frank, R. F. (1985). Passions within reason: The strategic role of the
emotions. New York: Norton.
Hardin, G.R. (1968) The tragedy of the commons. Science,ÿ162, 1243-1248.
Luce, R. D., & Raiffa, H. (1957). Games and decisions. New York: Wiley.
Messick, D. M. (1985). Social interdependence and decision making. In G.
Wright (Ed.), Behavioral decision making (pp. 87-109). New York: Plenum.
Nash, J. F. (1950). The bargaining problem. Econometrica, 18, 155-162.
Petit, P. (1990). Virtus normativa: Rational choice perspectives.
Ethics, 100, 725-755.
Raiffa, H. (1982). The art and science of negotiation. Cambridge, MA:
Harvard University Press.
Ritov, I., & Baron, J. (1990). Reluctance to vaccinate: omission bias and
ambiguity. Journal of Behavioral Decision Making, 3, 263-277.
Ritov, I., & Baron, J. (in press). Status-quo and omission bias. Journal
of Risk and Uncertainty.
Samuelson, W., & Zeckhauser, R. (1988). Status-quo bias in decision making.
Journal of Risk and Uncertainty, 1, 7-59.
Schelling, T. C. (1978). Micromotives and macrobehavior. New York:
Spranca, M., Minsk, E., & Baron, J. (1991). Omission and commission in
judgment and choice. Journal of Experimental Social Psychology, 27,
Tversky, A., Sattath, S., & Slovic, P. (1988). Contingent weighting in
judgment and choice. Psychological Review, 95, 371-384.
Tyler, T. R. (1988). What is procedural justice?: Criteria used by citizens to
assess the fairness of legal procedures. Law and Society Review. 22,
Ullmann-Margalit, E. (1977). The emergence of norms. Oxford: Clarendon
Van Avermaet, E. (1974). Equity: A theoretical and empirical analysis.
Unpublished doctoral dissertation, University of California, Santa Barbara.
Yamagishi, T. (1986). The provision of a sanctioning system as a public good.
Journal of Personality and Social Psychology, 51, 110-116
Yamagishi, T. (1988). Seriousness of social dilemmas and the provision of a
sanctioning system. Social Psychology, 51, 32-42.
This work was supported by grants from the National Science Foundation
(SES-8809299 and SES-9109763). We thank Judy Schlossberg for assistance in
scoring, and Colin Camerer, John C. Hershey, Eric Johnson, Howard Kunreuther,
Nicholas Maxwell, John Orbell, Joanne Schwartz, Mark Spranca, the editor,
and (especially) Paul Allison for comments, ideas, and discussion. Address
correspondence to J. Baron, Department of Psychology, 3815 Walnut St.,
University of Pennsylvania, Philadelphia, PA 19104-6196; email
File translated from
On 20 Sep 2003, 13:40.