[Note for bibliographic reference: Melberg, Hans O. (1996) Testing cultural
differences: Four experiments, http://www.oocities.org/hmelberg/papers/961209.htm]
Testing cultural differences
Four experiments
by Hans O. Melberg
I have previously showed that some social scientists use cultural variables to explain and
predict. For example, it is sometimes argued that the Russians will not develop a
democracy or a market economy because their culture is authoritarian and
anti-capitalistic. In the mentioned paper I questioned the reliability of these
explanations. This paper, however, takes a more constructive approach. In short, I want to
present five experiments which could be conducted to test the reliability of cultural
explanations.
1. Are the Russians more concerned about justice than other people?
Assume that a person says that he is going to give you and another person $10 on one
condition: That you agree on how to split the money. Moreover, the procedure is that only
you are allowed to make a suggestion (only once) on how to split the money, not the other
person. After you have made your suggestion the other person have to state whether he
accepts it or not. If he accepts, the money will be split according to your suggestion; If
he does not, both of you get nothing. Now, you might ask in what way can this reveal a
concern for justice?
Assume that you propose the following: I get $9, you get $1. The other person would
still win $1 by accepting your offer (and loose $1 by not), nevertheless empirical
research shows that people often turn the offer down. A plausible reason for this is that
they are so outraged by the unfair offer, that they would rather suffer the loss of $1
than be exploited. Real life experiments have showed that on average people do not accept
offers below $7-$3. The interesting question is then whether this number is differs from
country to country. If we found that the American average was $8-$2, while the Russian
average was $6-$4, this may indicate that the Russians really are more concerned about
justice than the Americans. In fact, we may develop an index of concern for justice ($10-
the difference) which would rise as concern for justice increased.
There are, of course, several problems with this procedure. For example, the value of
$1 is not the same for an American and a Russian. This problem can be reduced by using $1
in the USA and its equivalent in Russia, but this rises another problem since we do not
know the exact equivalent (taking the market exchange rate is helpful, but it would not
solve the problem). To reduce these problems once might develop some kind of measure of
how much the money is worth to the individual participant (by offering a choice between a
certain amount and a gamble of nothing with probability p and a larger sum with
probability 1-p. Alternatively one might try simple questions as how long would you
be willing to wait in a line to get this sum of money for free).
2. Is the Russian culture more risk-averse than other cultures?
One simple example to measure this, is to make people choose between the following two
options:
Alternative A: $100 for sure
Alternative B: $200 with the probability of 0.5 and $0 with the probability of 0.5
A risk-neutral person would be indifferent between the two options since the expected
payoff from each is $100. If we change the probabilities in alternative B to $200 with the
probability of 0.6 and $0 with the probability of 0.4, a risk-averse person would still
choose alternative A despite the fact that the expected payoff from alternative B is
higher. By varying the probabilities (and payoffs) and by presenting this kind of choice
to a number of people, one finds a reliable picture of the average risk attitude of a
group of people.
As always the measure is not perfect in that - once again - $1 in one country is not
the same as $1 in another. My reply would be the same as above: These problems can be
reduced and compensated by other indexes.
3. Do the Russians trust each other less than other nations?
Imagine that you and another person are playing a game in which you press either a red or
a green button. Assume that you go first, then the other person presses one of the
buttons. Each time you both press the green button, both of you get $5. This may go on for
maximum 100 rounds when the game automatically ends. If, however, one of you presses the
red button, that person gets $100 and the game ends right away. In what way does this game
measure trust?
Clearly, it would be best for the participants if they could trust each other to press
the green button for a long time since they would then earn $500 each. However, in round
99 the temptation to press the red button would be overwhelming since I would then get
$100 while if I continue to play green I can only get $5 more (since there is only one
round left of the game). But, knowing that I will play red in round 99, the other player
will himself be tempted to play red in round 98. After all, why leave $100 to you when I
can get the $100 myself. However, by the same reasoning I will play red in round 97 to
prevent you from doing to in round 98. As the reader will realize, this leads to a
regression in which the red button will be played in the very first round. However, one
again empirical research shows that in real life people do not press the red button before
round 85-93. One way of interpreting this is that they somehow trust each for some time.
Hence, if the average breakoff round is different from nation to nation we may speculate
whether this is because they have different levels of trust
Now, I am not certain that this experiment actually measures the level of trust. An
alternative interpretation is that the experiment measures the cleverness of the people
involved. The "smart" people are good at backward induction (I know that he
knows that I know etc.), while the less mathematically able usually do badly. This is one
problem which I will have to work on, though one might argue that it does not seem likely
that the people of one country is inherently smarter than another (though, of course, it
cannot be excluded a priori)
4. Is the Russian mind more bipolar than the minds of people in other countries?
It is sometimes argued (see Lotman and Uspenskij and Per-Arne Bodin) that the Russian have
more extreme value and belief systems than others: Either something is good or it is bad;
Either you believe something or you don't. They lack - so the argument goes - the neutral
category in which uncertainty and doubt is acknowledged. Moreover, the argument is also
that the Russians update their beliefs in a more extreme manner than other people. In
other words, that they go from one extreme to another when they change their mind. How
could one test this?
Assume that you have a box of 100 marbles, 50 blue and 50 yellow. Now, the distribution
(50-50) is unknown to the people participating in the experiment. You the proceed to take
five random marbles out of the box, showing the colour of each to the participants. You
then ask the participants to estimate the number of blue and yellow marbles based on the
information from the five they have seen. Moreover, you ask them to say how certain they
are that their estimate is correct (for example by giving a 95% confidence interval). One
may then pick five more marbles and ask the participants to re-evaluate their estimates
and the certainty of their estimates.
The information revealed by the above experiment can be used to see whether the
Russians are more certain than other peoples that their estimate is correct (i.e. more
extreme) and whether when they change their mind they do so in a more extreme fashion
(i.e. by overestimating in the other direction), than other nations.
As always, the experiment is not perfect. First of all, it should not be done on with
statisticians who could calculate the objectively true answers. Moreover, it is an
experimental test and we cannot directly infer from these results that the Russians really
update their beliefs in an extreme manner in the real world and about less statistical
topics. To reduce these problems we may conduct more experiments with a different context
(for example, whether they thought a defendant was guilty given a certain amount of
information and how they changed their opinions when given new information).
In Conclusion
I do not believe that the above experiments can provide conclusive proofs about the
Russian culture. However, I do believe that they may provide some indication as to the
validity of the cultural explanations mentioned in the introduction. Moreover, I do
believe that these experiments provide more information than opinion surveys since we
measure actual behaviour, not what people say they will do or (even less reliable) their
opinions grand concepts such as democracy and market-economy. In any case, there is no
inherent conflict both kind of experiments may be conducted.
Informal note on references
The first experiment (justice) is from a case study in the book A Critical Introduction
to the Theory of Choice by R. Sugden and others. The second is inspired by the Allais
paradox. I first heard about the third experiment from B. Williams (Economics tutor at
Jesus College, University of Oxford), but its origin is J. von Neumann and Expected
utility theory. The fourth experiment is inspired by (but not equivalent to) the Ellesberg
paradox. I am also inspired by Hillel and Yaari, Sugden, Elster, Binmore and many others.
[More experiments: See whether outside options matters in a bargaining situation;
Examine how changes in the bargaining strength affects the solutions; See how changes in
the market conditions of a commodity affects the perception of what is just and unjust -
example: if it is acceptable to raise the price of flowers after a frost.]
Suggestions are welcome.
[Note for bibliographic reference: Melberg, Hans O. (1996) Testing cultural
differences: Four experiments, http://www.oocities.org/hmelberg/papers/961209.htm]