[Paleopsych] Public Choice: Anyone for higher speed limits?

[Premise Checker]

Olof Johansson-Stenman and Peter Martinsson: Anyone for higher speed 
limits? - Self-interested and adaptive political preferences
Public Choice (2005) 122: 319-331 C
DOI: 10.1007/s11127-005-3901-x

[This is a nice article that gets at the issue of whether voters vote for 
what is in their own self-interest or what they think is in the overall 
public interest. I say vote for what's in your own interest, since you 
don't know what's in others' interest, certainly not better than they do 
themselves. It's the American way, that idea that government serve 
the people. One chief problem of altruism is that, if I am to serve other 
people, who are all these other people going to serve.

[One matter of surprise to me: "The result presented here is also 
consistent with the result of Hemenway and Solnick (1993) and Shinar, 
Schechtman and Compton (2001), who found that levels of education higher 
than high-school tended to increase the probability of speed violation."

[But be wary of the article, though, since the variables explained only 
20% of the variance. There's a real likelihood that, if someone will come 
up with other variables that will explain more of the variance, the 
coefficients on the ones used here could get drastically altered. One of 
the critics of _The Be** Cu**e_ complained that the authors often buried 
the R^2s in the back, and in fact the R^2s varied all over the place.

[Note that I said, "IF someone will come up with other variables.... 
Researchers use the variables they can get ahold of. What _The B*ll C*rv*_ 
did was to thoroughly mine a data set, that collected for the National 
Longitudinal Study of Youth, that almost uniquely had a measure of 
intelligence. The results are known: IQ correlated more with things like 
income and scholastic achievement than do the usual Socioeconomic status, 
parental income, and so on. I'd love to know whether having an IQ measure 
upset any conventional wisdom on these other factors.

[When I was in graduate school at UVa, self-interested voting was 
emphasized and expressive voting barely recognized. I was Virginia School. 
The Rochester School (William Riker and then others) came along later. It 
was in political science but used economics tools. It emphasised 
disinterested voting. These two Schools were never, I don't think, hostile 
toward one another, and this article shows that the issues are empirical.]

Department of Economics, SE 40530 Göteborg, Sweden; e-mail:
Olof.Johansson at economics.gu.se, Peter.Martinsson at economics.gu.se

Accepted 17 November 2003

Abstract. Swedish survey-evidence indicates that variables reflecting 
self-interest are important in explaining people's preferred speed limits, 
and that political preferences adapt to technological development. Drivers 
who believe they drive better than the average driver as well as drivers 
of cars that are newer (and hence safer), bigger, and with better high- 
speed characteristics, prefer higher speed limits. In contrast, elderly 
people prefer lower speed limits. Furthermore, people report that they 
themselves vote more sociotropically than they believe others vote on 
average, indicating that we may vote less sociotropically than we believe 
ourselves. One possible reason for such self-serving biases is that people 
desire to see themselves as socially responsible.

*We are grateful for very constructive comments from an anonymous referee. 
We have also received useful comments from Fredrik Carlsson and had 
fruitful discussions with Per Fredriksson and Douglas Hibbs. Financial 
support from the Swedish Agency for Innovation Systems (VINNOVA) is 
gratefully acknowledged.

1. Introduction

The purpose of this paper is twofold: i) to use survey evidence about what 
speed limits different people prefer on motorways, and what their own 
subjectively perceived and self-reported voting motives are, in order to 
provide new insight into the determinants of individual voting behavior, 
in particular the self-interested voting hypothesis; and ii) to identify 
adaptations in political preferences due to technological development, in 
our case changes in safety and high-speed features of cars. The analysis 
is based on two recent representative Swedish surveys: In the first one 
people were asked about their preferred speed limits on motorways. In the 
second they were asked about why they vote as they do, and about why they 
think other people vote as they do.

Why do people vote in the way they do and why do they vote at all? One 
reason for the latter is simply that we are heavily indoctrinated to do 
so; c.f. Tullock (2000). But is how we vote motivated solely by the 
instrumental outcome induced by our votes? Or are we perhaps, as proposed 
by Brennan and Lomasky (1993) and Brennan and Hamlin (1998, 2000), 
motivated largely by the expressive act of voting? If the expressive 
motive is important it becomes more likely that people are concerned with 
society as a whole when voting, rather than what is good solely for 
themselves.1 Indeed, as found by Brekke, Kverndokk and Nyborg (2003), most 
people seem to prefer a self-image that reflects social responsibility, 
rather than pure self-concern. The relative importance of purely 
self-interested voting, versus sociotropic voting, is still debated. This 
is partly because it is difficult to draw strong conclusions from general 
elections that are characterized by few political parties (or candidates) 
and many political issues and indicators; see e.g. Kinder and Kiewiet 
(1979), Kramer (1983) and Mitchell (1990). The reason for this is that 
some opinions of one party may favour a certain group while other opinions 
may favour other groups, and it is difficult to know the relative weights 
that different voters give to the different opinions of the parties. Thus, 
there are clear advantages to be gained from testing the self-interested 
voting model when the choice set is small and when there are few political 
issues, such as on a single-issue referendum or by using tailor-made 
surveys.

Smith (1975) analyzed the voting behavior from a referendum in Oregon 
concerning tax equalization between different districts, and concluded 
that self-interest does seem to play an important role. Sears, Lau, Tyler 
and Allen (1980), on the other hand, analyzed survey data on people's 
attitudes toward specific policies in the US, and concluded that 
self-interest plays a very minor role. However, their conclusions, based 
on their statistical results, can be questioned: for example, they found 
that the support for a national health insurance decreased with income and 
increased with age, and that the support for more resources to be given to 
law and order increased with income, but these findings were not 
interpreted to reflect self-interest. Nevertheless, there have also been 
other studies such as Gramlich and Rubinfeld (1982) and Shabman and 
Stephenson (1994) that have concluded that self-interest alone does a poor 
job of explaining the results. These findings are also consistent with 
much experimental evidence from public-good games; see e.g. Ledyard (1995) 
and Keser and van Winden (2000).

Much of the analysis here is based on the first survey about the preferred 
speed limits on motorways, which is an issue that has been frequently 
debated for a long time in Sweden. Besides being a single issue, it has 
the advantage of being fairly neutral from an ethical point of view, 
meaning that the opinion of good and responsible citizens is not 
straightforward to predict.2 Survey responses can otherwise be biased 
towards what is perceived to be the most ethical alternative, which is an 
argument that for example is put forward in the environmental valuation 
literature. A possible underlying reason for this bias, in turn, is that 
people typically attempt to present themselves in a positive manner to 
others, which implies that we sometimes deliberately conceal or colour our 
true opinions or preferences, i.e. what Kuran (1995) denotes "preference 
falsifications." An alternative reason is our desire to see ourselves as 
good people, and our tendency to bias our impressions of reality in 
various respects to maintain or improve this self-image (see e.g. 
Gilovich, 1991). Such tendencies may, for example, influence people to 
believe that they would be willing to pay more for a socially good cause 
than they would actually be willing to pay, which is denoted "purchase of 
moral satisfaction" by Kahneman and Knetsch (1992).

In order to broaden the insights on voting motives, we also performed a 
second survey where a representative sample of Swedes was asked about why 
they vote as they do, and why they think other people vote as they do. 
This allows us to compare the findings about the preferred speed limits 
with the perception that people have of their own and others' voting 
motives. The reason we also asked about the perception of others' voting 
motives is the one just mentioned, i.e. that we suspected that the 
responses may be biased since most people would presumably consider voting 
out of conviction to be ethically superior to voting solely for one's own 
good.

Given that self-interest is important for the political preferences, one 
would from the general self-interested hypothesis expect people with more 
exclusive and safer cars, and with higher subjective driving skills, to 
prefer relatively high speed limits, and elderly and more vulnerable 
people to prefer lower speed limits. In addition, one would expect people 
who drive faster, and who break the speed limits more often to prefer 
higher speed limits. The results, reported in Section 2, are consistent 
with these hypotheses. One would also expect that these preferences would 
change with changing circumstances. Indeed, behavioural adaptations in 
response to perceived changes in the environment are among the most 
important insights that modern economics can contribute to the public 
debate. For example, a safety improvement in cars of, say, 10% may cause a 
much smaller net effect on safety, since safer cars may induce people to 
drive faster and less responsibly; see Peltzman (1975), Keeler (1994), 
Peterson, Hoffer and Millner (1995) and Merrel, Poitras and Sutter (1999) 
for theoretical analysis as well as empirical evidence. This paper will 
concentrate on another kind of adjustment, namely how political 
preferences with respect to preferred speed limits on motorways change 
with the rapid technological development of private cars. The data used 
here is not ideal in this respect, since the survey is purely 
cross-sectional. Nevertheless, it is still possible to see whether the 
results are consistent with the hypothesis of adjustments of political 
preferences. If people demand higher speed limits when their cars get 
safer and have better high-speed characteristics, one would expect from 
the empirical analysis that more people would be in favour of increasing 
speed limits rather than decreasing them, since these limits were decided 
upon many years ago,3 and also that individuals with newer cars would 
prefer higher speed limits. This is also found in our empirical analysis.

It is interesting to compare the obtained motives that can be inferred 
from people's choices, in reality or in surveys, from their own 
subjectively perceived voting motives. This is the reason we undertook the 
second survey about people's perceptions of their own and others' voting 
motives. The results indicate that most people believe that others vote 
largely for their own interests, whereas they, on average, consider 
themselves to be influenced roughly equally by their own interests and by 
those of society as a whole. The results further help to identify possible 
self-serving biases, i.e. that people may tend, unconsciously, to believe 
that what is in the interest of society happens to coincide with what is 
in their own private interest. If so, one would expect systematic 
differences between people's reported perception of their own motives and 
that of others' motives, and that people on average believe that they 
themselves vote more out of conviction, or sociotropically, than others 
do. And as reported in Section 3, this is indeed found to be the case.

2. Analysis of preferred speed limits

The main survey was mailed to 2500 randomly selected individuals aged 
between 18 and 65 years old in Sweden, during spring 2001. The response 
rate of the overall survey was 62%, and 1131 car drivers answered the 
speed- limit question. Each respondent was asked the following question: 
What speed limit do you think we should have on Swedish motorways? They 
were given five options, all of which have been discussed in the Swedish 
debate from time to time: 90, 100, 110 (the level today), 120, and 130 
km/h.

The descriptive result in Table 1 shows that very few would like to have 
decreased speed limits, and that more than half of the respondents would 
like to see increased speed limits. This may in itself be an indication 
that people have adapted their political preferences to the increased 
levels of vehicle safety, but to be able to say more on this issue we 
would need to know who wants increased speed limits, and who does not. 
This is the issue to which we turn to next.

In order to obtain information on the characteristics that affect the 
preferred speed limit, we ran an OLS-regression with the preferred speed 
limit as the dependent variable on a number of socio-economic 
characteristics and the characteristics of the car that they most 
frequently drive. Because of missing or incomplete responses, primarily on 
the income and voting variables the number of respondents included in the 
analysis is 974. The results from the estimations are presented in Table 2 
along with the mean sample value of each explanatory variable.

Table 1. Sample distribution of the preferred speed limit on Swedish 
motorways. (N = 974)

90 km/h 100 km/h 110 km/h 120 km/h 130 km/h (as of today)
2% 3% 41% 25% 29%

The results show that those who drive newer cars do prefer higher speed 
limits, as one would expect, given that people adapt their preferences to 
changing circumstances, in this case safer cars with better high-speed 
driving characteristics.4 Similarly, drivers of the prestige cars BMW, 
Mercedes and Porsche, which are also safer and/or have better high-speed 
driving characteristics, also prefer higher speed limits. The size of car 
also affects the preferred speed limit in the expected direction, since 
bigger cars are on average safer, and have better high-speed 
characteristics, but the differences are not significant at conventional 
levels. Jeeps and vans constitute the base case, and although these are 
big vehicles, they have typically bad high-speed characteristics.

The preferred speed limit is higher for those who believe they are better 
than average drivers, which is also consistent with the self-interested 
hypothesis, since the risk of an accident, for a given speed, would then 
be lower.5 A long annual driving distance also increases preferred speed 
limit, which, however, is not obvious from the self-interested hypothesis. 
On the one hand, those who drive a lot will gain more time from increased 
speed levels, but on the other hand they will also face a larger reduction 
in safety. In our case, it seems that the former effect dominates the 
latter. This is also consistent with Rienstra and Rietveld (1996), who 
found that self-reported frequency of speed-transgressions on Dutch 
highways increases with annual driving distances. The effects of always 
using a seatbelt may seem to contradict the theory, since those without 
seatbelts would face the biggest risk-increase from increased speed 
levels. However, it seems likely that the results largely reflect 
preference heterogeneity, so that those who are more risk-averse, or 
generally more cautious, prefer both to use seatbelts and to have 
relatively low speed limits.

People living in the bigger cities of Sweden prefer somewhat higher speed 
limits, for which one explanation may be the higher pace, in general, of 
urban life, which translates into a higher value of time. The effect of 
education is quite small, and perhaps in the opposite direction to that in 
which one would have guessed, since safety awareness is often believed to 
follow from, or at least to be positively correlated with, education. 
However, hardly anyone in Sweden, irrespective of education, can be 
uninformed about the public campaign messages that safety decreases as 
speed increases. Further, the true relationship between speed and safety 
may not be as clear and strong as is typically presented, and maybe highly 
educated people are less easy to convince by public propaganda. Generally, 
most (but not all)6 analysts seem to agree that safety typically does 
decrease with increased speed limits, but there is less agreement about 
how large the effect is. Nevertheless, the result presented here is also 
consistent with the result of Hemenway and Solnick (1993) and Shinar, 
Schechtman and Compton (2001), who found that levels of education higher 
than high-school tended to increase the probability of speed violation.

Table 2. OLS-estimation of preferred speed limit on Swedish motorways. 
Dependent variable: Preferred speed limit on Swedish motorways in km/h. (N 
= 974)

Variable Coeff. P-value Mean value
Constant -111.552 0.276
Model-year of the car 0.112 0.030 1993.299

Drives either BMW, Mercedes or Porsche 2.771 0.029 0.050
Drives a small-sized car 1.082 0.505 0.071
Drives a medium-sized car 1.553 0.229 0.516
Drives a big car 2.173 0.100 0.362
Drives better than average (self-reported) 2.693 0.000 0.424
Drove more than 25000 km last year 1.524 0.025 0.213
Always wears seat-belt in front-seat -2.446 0.003 0.860
Lives in Stockholm, Gothenburg or Malmö 1.469 0.040 0.196
University-educated 1.314 0.103 0.322
A-level educated 0.907 0.213 0.449

Equivalence-scaled household income* 0.201 0.001 12.047
Aged above 57 -1.952 0.021 0.151
Male 4.233 0.000 0.532
Has at least one child 0.669 0.307 0.406
Right-wing political preferences 2.799 0.001 0.140
Left-wing political preferences -1.582 0.011 0.299

R2 = 0.204

RESET** p-value = 0.281

Mean VIF*** = 1.84 and highest VIF for a single variable is 5.64.

*In 1000 SEK/month and person. In order to compare income between 
households, we employ the equivalence scale used by the National Tax Board 
(RSV) in Sweden. The scale assigns the first adult the value of 0.95, the 
following adults are set at 0.7 and each child at 0.61 units.

**RESET type of test is a general specification test (see e.g. Godfrey, 
1988). In the test we rerun the regression including the squared, cubed 
and quadratic values of the estimated value of the dependent variable from 
the original model and test if coefficients of the included variables are 
jointly significant.

***We test for multicollinearity in our data set by calculating the 
variance of inflation factor (VIF) for each variable. The largest VIF is 
5.64 and the mean VIF is 1.84. The largest value is thus smaller than 10 
and the mean value is not considerably larger than 1, as required to be 
able to judge that there is no apparent indication of mutlicollinearity 
according to STATA (2003: 378).

Increased household income causes both higher value of time and a higher 
value of a statistical life, or more generally, the willingness to pay to 
avoid traffic risks; hence the theoretical prediction is ambiguous. As for 
driving distance, the time effect appears to dominate. These results are 
also consistent with Rienstra and Rietveld (1996) and Shinar, Schechtman 
and Compton (2001) who found that those with the highest incomes tend to 
break highway speed limits more often than others. Older people prefer 
lower speed limits, as predicted due to their increased vulnerability.

The relatively large male coefficient, corresponding to more than 4 km/h, 
can possibly be explained by observed higher risk aversion among women 
(e.g. Jianakoplos and Bernasek 1998, Hartog, Ferrer-i-Carbonell and Jonker 
2002), but it might also reflect a taste difference concerning how fun 
fast driving is perceived to be, or some kind of macho image.

The influence of political voting is also in the expected direction, since 
political parties to the left have typically proposed, and been associated 
with, a more restrictive speed policy, and vice versa. These parameters 
too may reflect direct instrumental self-interest, if people choose 
political party partly due to the politically proposed speed limits. 
Still, it seems reasonable that these parameters rather reflect 
ideological conviction and expressive concern. This does not necessarily 
mean that they represent sociotropic concern, however, since people may 
have different kinds of values and opinions that they want to express; see 
e.g. Brennan and Lomasky (1993) and Brennan and Hamlin (2000). There is 
also a large part of the variation left unexplained, and we do not know 
how large a share of this part can be explained by non- included variables 
that reflect self-interest, such as how fun it is considered to be to 
drive fast.

3. Perceptions of voting motives

This second survey was mailed to 1500 randomly selected individuals aged 
between 18 and 65 years old in Sweden, during spring 2002 (i.e. a year 
after the first survey), and the response rate of the overall survey was 
58%. To compare actual voting motives with the perception people have of 
voting motives, we simply asked another representative sample of Swedes 
about why they thought other people vote as they do, followed by a 
question about why they themselves vote as they do. Before the questions, 
they were given the following information: One can vote for a political 
party for different reasons. One can vote for a party because one is 
favored oneself, or one can do it out of conviction that it is the best 
for society as a whole.

As can be seen from Tables 3 and 4, most people believe that others vote 
largely for their own interests, whereas they, on average, consider 
themselves to be influenced roughly equally by their own interests and by 
those of society as a whole. To test whether the observed differences are 
statistically significant, i.e. whether there is a statistical difference 
between people's perception of the degree to which they themselves vote 
sociotropically, and the degree to which others vote sociotropically, we 
used a simple ordered probit model; the motives are ordered from "Mostly 
because it benefits me (them)" to "Mostly out of conviction." This is an 
appropriate econometric specification since the empirical analysis focuses 
on an ordered discrete variable. The approach is based on the idea of a 
latent unobservable variable, Socio*, representing, in our case, 
individuals' perception of the degree of sociotropic voting with the 
following structure: 7

Socio*= d DOthers + e,

where DOthers is a dummy variable indicating that the responses are given 
to the framing on how others vote, and d is the associated parameter to be 
estimated; e is assumed to be a normally distributed error term with zero 
mean and constant variance. The results in Table 5 show that the between- 
sample difference is indeed highly significant as reflected by a 
significant d-parameter (at less than 0.1% level).

One possible reason for this systematic bias is that people want to have a 
good self-image, or identity, and that they therefore engage in a degree 
of self- deception so that they believe that they would vote more for the 
common good than they would actually do in reality. Indeed, there is much 
psychological evidence for systematic self-deception that enhances 
people's perception of their own abilities in many respects; see e.g. 
Gilovich (1991) and Taylor and Brown (1994). An alternative, slightly more 
sophisticated version of this argument, is that people answer truthfully 
and without bias concerning their own motives. However, since most of us 
want to see ourselves as good and responsible people, and at the same time 
to do what is best for ourselves, we may unconsciously try to reduce the 
cognitive dissonance (cf. Akerlof and Dickens, 1982) by adapting our 
perceptions of what is best for society as a whole so that it more or less 
coincides with what is best for ourselves. Hence, when we honestly try to 
judge different alternatives as objectively as possible on behalf of 
society, we will still unconsciously bias our judgment in favour of what 
is best for ourselves; see Babcock and Loewenstein (1997) and references 
therein for much evidence of such self-serving biases.

Table 3. Self-reported perceptions of own voting motives. (N = 751)

Why do you vote as you do?
Reason Fraction

Mostly because it benefits me 10%
Because it benefits me, but also to a certain degree out of conviction 23%
Equally because it benefits me and out of conviction 27%
Out of conviction, but also to a certain degree because it benefits me 22%
Mostly out of conviction 18%

Table 4. Self-reported perceptions of others' voting motives. (N = 762)

Why do you, on average, believe that people vote as they do?

Reason Fraction
Mostly because it benefits them 20%
Because it benefits them, but also to a certain degree out of conviction 
39%
Equally because it benefits them and out of conviction 19%
Out of conviction, but also to a certain degree because it benefits them 
17%
Mostly out of conviction 5%

Table 5. Ordered probit regression to estimate the differences between the 
respondents' perceived degree to which they themselves and others vote 
sociotropically. (N = 1513)

Variable Coeff. P-value

Dummy variable reflecting the additional degree that others (compared to 
oneself) vote sociotropically -0.553 0.000

Cut-off 1 -1.346
Cut-off 2 -0.377
Cut-off 3 0.246
Cut-off 4 0.968

The dependent variable is the perceived degree of sociotropic voting coded 
as follows: 1 = Mostly because it benefits me (them); 2 = Because it 
benefits me (them), but also to a certain degree out of conviction; 3 = 
Equally because it benefits me (them) and out of conviction; 4 = Out of 
conviction, but also to a certain degree because it benefits me (them); 
and 5 = Mostly out of conviction.

When we observe others, however, we just know roughly how they vote and 
their other circumstances. Hence, we can only crudely observe the 
correspondence between how others vote and their personal interests. But 
since we do not take into account the fact that others too adapt their 
perceptions of what is in the interest of society, through self-serving 
biases, the perception of the degree to which others vote sociotropically 
may be biased downwards.

4. Conclusion

Most results from our survey indicate that self-interest is an important 
determinant of the preferred speed limit; for example, those who have a 
newer car (and hence one that is typically safer and more comfortable at 
high speeds) that is bigger and faster, prefer higher speed limits. This 
is also true for those who believe they are better than the average 
driver, whereas older people prefer lower speed limits. Furthermore, the 
results are also consistent with the existence of political offsetting 
behaviour, so that when cars become safer due to technological 
developments, people adapt their political preferences in favour of higher 
speed limits, which reduces road safety overall.

However, the results from people's self-reported subjective voting motives 
are not consistent with purely instrumental pocketbook voting. Rather, it 
seems that the expressive motive is important, as argued thoroughly by 
Brennan and Lomasky (1993) and Brennan and Hamlin (1998, 2000),8 and it 
seems in particular that people want to express that they are socially 
responsible people who care about the overall welfare of society. This is 
also strengthened by the observed fact that people tend to believe that 
others vote more in their own interests, on average. Still, despite such 
biases, we also find that most people answer that they vote both for their 
own interest and for the interest of society. Hence, the hypothesis that 
most people solely or primarily vote sociotropically appears to be 
incorrect too.

Answering a survey, such as ours on preferred speed limits, is in some 
respects quite similar to voting. Since the respondents were informed that 
the survey was sent out to a large random sample of Swedes by a 
university, and was a part of a research project, they could hardly 
believe that their single response would influence actual policy in a 
non-negligible way. Furthermore, the financial incentive of answering was 
zero, and it took probably almost half an hour to answer the whole survey 
on average. The response rates (62% and 58% respectively) were also 
similar to electoral participation rates in many countries.9 Presumably, 
most of the respondents answered based on a sense of civic duty, or due to 
the disutility associated with not answering which would break what they 
perceive to be a social (or personal) norm. But given that expressive 
voting, and expressive answering of surveys such as ours, is the main 
explanation behind observed behaviour, how can we explain the fairly 
strong correlation with their own self-interest? Although it is perceived 
as socially admirable to vote, it is hardly perceived to be admirable to 
vote solely for your own best interests. Rather, we are socialized to 
focus on the collective good when wearing our "political hats" (Sears, 
Lau, Tyler and Allen, 1980; Sears and Funk, 1990). One possible 
explanation to this paradox is provided by the idea of self-serving bias. 
As expressed by Elster (1999, 333): "Most people do not like to think of 
themselves as motivated only be self-interest.They will, therefore, 
gravitate spontaneously towards a world-view that suggests a coincidence 
between their special interest and the public interest." (italics in 
original.) In this way we can vote for improvements for ourselves without 
feeling guilty that this would, overall, be bad for society, and we are 
hence not plagued by any cognitive dissonance. After all, it is much more 
pleasant to think that what is good for you is also good for society, 
isn't it?

Notes

1. However, as argued by Brennan and Lomasky (1993) as well as Brennan and 
Hamlin (2000), expressive voting per se does not necessarily imply 
sociotropic voting.

2. If anything, it may be considered somewhat more ethical to vote for 
lower speed limits. Nevertheless, despite a possible bias in this 
direction, very few respondents (5%) prefer a lower speed limit than the 
current one, as can be seen from Table 1.

3. Highway speed limits have increased rapidly in many states in the USA 
during the last 15 years (Greenstone, 2002), and also in other countries 
such as Italy, while there are on-going discussions in many other 
countries.

4. However, it is possible that people who drive newer cars do so due to 
stronger preferences for safety. For this reason, those who have new cars 
would then prefer lower speed limits than others would. Given that the 
empirical result presents the net effect, the isolated effect of a newer 
car on the preferred speed limit would then be larger than the effect that 
is presented here.

5. This does not necessarily mean that actual safety increases with 
self-reported subjective driving ability, however, since over-optimism 
regarding one's own driving ability is likely to be positively correlated 
with subjective driving ability. Still, what matters for the preferred 
speed limit it the subjective risk, which is independent of such biases.

6. Indeed, some analysts have even questioned the sign of the 
relationship: Lave and Elias (1997) argued that the accident increase on 
rural interstate USA roads resulting from increasing the speed limits to 
65 mph in 1987 were more than off-set by the decline of accidents on other 
roads due to compensatory reallocations of drivers and state police; see 
also Greenstone (2002), who, however, questioned the conclusion by Lave 
and Elias.

7. In our case five ordered categories are possible. The respondents are 
assumed to choose the alternative closest to their own perception, where 
we observe Socio = 1, i.e. "mostly because it benefits me (them)," if 
Socio*= a1;Socio = 2, i.e. "because it benefits me (them), but also to a 
certain degree out of conviction," if a1 < Socio*= a2 etc.; until Socio = 
5, i.e. "mostly out of conviction," if a4 = Socio*;where a1 to a4 are 
cut-off points to be estimated simultaneously with the coefficient.

8. See also Copeland and Laband (2002) for recent empirical support.

9. In the 2002 General Election in Sweden 80.1% of the eligible population 
voted (SCB, 2002).

References

Akerlof, G.A. and Dickens, T.W. (1982). The economic consequences of 
cognitive dissonance. American Economic Review 72: 307-319.

Babcock, L. and Loewenstein, G. (1997). Explaining bargaining impasse: The 
role of self- serving biases. Journal of Economic Perspectives 11: 
109-126.

Brekke, K.A., Kverndokk, S. and Nyborg, K. (2003). An economic model of 
moral motivation. Journal of Public Economics 87: 1967-1983.

Brennan, G. and Hamlin, A. (1998). Expressive voting and electoral 
equilibrium. Public Choice 95: 149-175.

Brennan, G. and Hamlin, A. (2000). Democratic devides and desires. 
Cambridge: Cambridge University Press.

Brennan, G. and Lomasky, L. (1993). Democracy and decision: The pure 
theory of electoral preference. Cambridge: Cambridge University Press.

Copeland, C. and Laband, D.N. (2002). Expressiveness and voting. Public 
Choice 110: 351- 363.

Elster, J. (1999). Alchemies of the mind: Rationality and the emotions. 
Cambridge: Cambridge University Press.

Gilovich, T. (1991). Why we know what isn't so. New York: Free Press.

Godfrey, L. (1988). Misspecification tests in econometrics: The Lagrange 
multiplier principle and other approaches. Econometric Society Monographs 
No. 16. Cambridge: Cambridge University Press.

Gramlich, E.M. and Rubinfeld, D.L. (1982). Voting on spending. Journal of 
Policy Analysis and Management 1: 516-533.

Greenstone, M. (2002). A reexamination of resource allocation responses to 
the 65-MPH speed limit. Economic Inquiry 40: 271-278.

Hartog, J., Ferrer-i-Carbonell, A. and Jonker, N. (2002). Linking measured 
risk aversion to individual characteristics. Kyklos 55: 3-26.

Hemenway, D. and Solnick, S. (1993). Fuzzy dice, drean cars, and indecent 
gestures: Correlates of drivers behavior. Accident Analysis and Prevention 
25: 161-170.

Jianakoplos, N.A. and Bernasek, A. (1998). Are women more risk averse? 
Economic Inquiry 36: 620-630.

Kahneman, D. and Knetsch, J.L. (1992). Valuing public goods: The purchase 
of moral satisfaction. Journal of Environmental Economics and Management 
22: 57-70.

Keeler, T.E. (1994). Highway safety, economic behavior, and driving 
environment. American Economic Review 84: 684-693.

Keser, C. and van Winden, F. (2000). Conditional cooperation and voluntary 
contributions to public goods. Scandinavian Journal of Economics 102: 
23-39.

Kinder, D.R. and Kiewiet, D.R. (1979). Economic discontent and political 
behavior: The role of personal grievances and collective economic 
judgments in congressional voting. American Political Science Review 23: 
495-517.

Kramer, G.H. (1983). The ecological fallacy revisited: Aggregate-versus 
individual-level findings on economics and elections, and sociotropic 
voting. American Political Science Review 77: 92-111.

Kuran, T. (1995). Private truths, public lies: The social consequences of 
preference falsification. Cambridge Mass: Harvard University Press.

Lave, C. and Elias, P. (1997). Resource allocation in public policy: The 
effects of the 65-MPH speed limit. Economic Inquiry 35: 614-620.

Ledyard, J.O. (1995). Public goods: A survey of experimental research. In 
J.H. Kagel and A.E. Roth (Eds.), Handbook of experimental economics, 
111-194. Princeton: Princeton University Press.

Merrell, D., Poitras, M. and Sutter, D. (1999). The effectiveness of 
vehicle safety inspections: An analysis using panel data. Southern 
Economic Journal 65: 571-583.

Mitchell, W.C. (1990). Ambiguity, contradictions, and frustrations at the 
ballot box: A public choice perspective. Policy Studies Review 9: 517-525.

Peltzman, S. (1975). The effects of automobile safety regulation. Journal 
of Political Economy 83: 677-725.

Peterson, S., Hoffer, G. and Millner, E. (1995). Are drivers of 
air-bag-equipped cars more aggressive? A test of the offsetting behavior 
hypothesis. Journal of Law and Economics 38: 251-264.

Rienstra, S.A. and Rietveld, P. (1996). Speed behaviour of car drivers: A 
statistical analysis of acceptance of changes in speed policies in the 
Netherlands. Transportation Research: Part D: Transport and Environment 1: 
97-110.

SCB (2002). http://www.scb.se/statistik/me0101/me0101_tab511.xls

Sears, D., Lau, R., Tyler, T. and Allen, H. (1980). Self-interest vs. 
symbolic politics in policy attitudes and presidential voting. American 
Political Science Review 74: 670-684.

Sears, D.O. and Funk, C.L. (1990). Self-interest in Americans' political 
opinions. In J. Mansbridge (Ed.), Beyond self-interest, 147-170. Chicago: 
University of Chicago Press.

Shabman, L. and Stephenson, K. (1994). A critique of the self-interested 
voter model: The case of a local single issue referendum. Journal of 
Economic Issues 28: 1173-1186.

Shinar D, Schechtman, E. and Compton, R. (2001). Self-reports of safe 
driving behaviors in relationship to sex, age, education and income in the 
US adult driving population. Accident Analysis and Prevention 33: 111-116.

Smith, J.H. (1975), A clear test of rational voting. Public Choice 23: 
55-67.

Stata (2003). Reference N-R. College Station Texas: Stata Press 
Publication.

Taylor, S.E. and Brown, J.D. (1994). Positive illusions and well-being 
revisited: Separating fact from fiction. Psychological Bulletin 116: 
21-27.

Tullock, G. (2000). Some further thoughts on voting. Public Choice 104: 
181-182.