The results from my experiment

This page is a very brief summary of the major results of my experiment. The analysis of the questionnaires is only available in Danish. The description of how I distributed the questionnaire is also only availabel in Danish.

If you are not familiar with the questionnaire, on which this experiment is based, please take a couple of minutes to go through it.

Demographic results

4940 people have answered the questionnaires. 8 people have answered, that they never use a computer although they have answered the questionnaire on the Internet. 12 people are under 10 years old. These 20 people are sorted out and this brings the number of observations down to 4920.

Danish English French Spanish German
1918 2301 546 21 134
Number of observations from each questionnaire

The questionnaires have been answered by people from all over the world. There has been a contest and a sweepstake in connection to the survey/experiment. This might have influenced the answers.

H) Age
     


Fig. 1: The respondents' age
Missing value: 24.


Fig. 2: The respondents age divided into gender
Missing value: 36.


Fig. 3: Education level
Missing value: 38.

I) How many years have you attended school or other education/training (includes apprenticeship/traineeship)?
     


Fig. 4: Education level divided into gender
Missing value: 57.

Conclusion on demographic results: The sample is biased compared to the population (which I have defined to be all non-illiterated people who is older than 10 years), but not as much as one could expect, when sampling on the Internet and adding sweepstakes to the survey/experiment.

Are preferences transitive?

This analysis is based on questions number 1, 6, 9 and 12:

1) If you were to choose between the two following writing utensils, which would you choose?
     

 

6) If you were to choose between the following two writing utensils, which would you choose?
     

 

9) If you were to choose between the following two writing utensils, which would you choose?
     

 

12) If you were to choose between the following two writing utensils, which would you choose?
     

The philosophy is as follows: If you prefer a pencil rather than a ball point pen, and you prefer a ball point pen rather than a fauntain pen, then you can not prefer a fauntain pen rather than a pencil (you then have intransitive preferences).

The analysis is a bit more complicated than the above example, and the result is, that 114 people has intransitive preferences, which corresponds to 2.3% of the respondents.

The choice between a male and a female when hiring someone

This analysis is based on question number 2a and 2b.

2a) Imagine yourself as an employer. You have to hire a new worker. The choice lies between a woman or a man. They are both 25 years of age, work equally fast and have the same qualifications. The job can be done equally well by a woman or a man. They are both to receive the same salary and are both to be trained for a year.

Who would you hire?
     

2b) Why?
     

If you are strictly rational in an economic sense, you would choose the male in this situation, since the female may be pregnant. You have some costs of training, so you can't get the full utilization of a new employee if one of the old employees leaves the job for a while. Therefor you have to choose the male (that's economic theory: Cold and rough).

Only 4567 persons have answered the question, which indicates that many people didn't know who they should choose. 64.4% chose the female candidate. 62.5% of the male respondents chose the female candidate, and 66.82% of the female respondents chose the female candidate.

In the Danish analysis, I have divided the question 2a into 13 subgroups. The analysis contains only the first 3257 observations. This showed, that only 129 female and 273 male respondents gave the rational economic answer (pregnancy) to this question. This corresponds to 13.5% of the respondents. Most of the arguments concerned equal opportunity, where 312 female respondents chose a female candidate and 235 male respondents chose a female candidate based on this criteria.

The prisoners' dilemma

(questions number 3, 4 and 5)
3) Imagine yourself arrested for a crime you have committed with a friend of yours. The police only have little evidence against you. You are isolated from each other and questioned separately. You have the choice between confessing to or denying the crime. If you confess and your friend denies, you will get the shortest penalty because you have helped clearing up the crime. Your friend receives the same offer, if you deny and he confesses. If you both deny, you receive a penalty that is less than if you both confess. Your future domicile will be the prison and you must spend the following years there:

If you confess and your friend denies: You will get 1 year and your friend will get 7 years.
If you confess and your friend confesses: You will get 5 years and your friend will get 5 years.
If you deny and your friend denies: You will get 2 years and your friend will get 2 years.
If you deny and your friend confesses: You will get 7 years and your friend will get 1 year.

What do you do?
     

 

4) Your friend is a hardened criminal and is familiar with the above mentioned penalties. Therefore you agree before committing the crime, that both of you will deny. As before, you are questioned separately.

What do you do?
     

 

5) What would your answer to question number 3 be, if you know that your friend is a member of a very violent gang. You are not a member of any gang, and you have a family, you care a lot about. You havn't made any agreement in anticipation.

What do you do?
     

Question number 5 is not a prisoners' dilemma-question, but a question about rational behavior. Both prisoners have incitement to choose 'confess' in question 4 and 5, because this strategy is dominating. In question number 5, you should choose 'deny', because this answer gives your partner the lowest penalty, given your answer. This way, your partner's gang won't hurt your family. In other words: If you think, that your partner will deny, then you

will get the lowest penalty if you confess. The same thing applies for you partner. You both get the lowest penalty, if the opponent denies, and you confess. Therefore you have strong incentives to confess, allthoug if you cooperate, you will both receive the lowest penalty in total.

I have added a little difference in the questionnaires: In the English version, you commit the crime with a 'friend', in the German, French and Spanish version, you commit the crime with a 'man', and in the Danish version you commit the crime with a 'sinister man'. This way, it's possible to analyze the altruism/egoism-subject.

Question 3:

The crime is commited with a ... The Respondent confesses The Respondent denies

Missing

values

Absolute Relative Absolute Relative
sinister man 1516 79.6% 388 20.4% 14
man 408 58.3% 292 41.7% 1
friend 1573 68.6% 719 31.4% 9

One should expect that relatively more people would confess, when they commit the crime with a man rather than a friend. It is peculiar that relatively more female respondents confess (75.36%) than male respondents (68.42%). This is not true for the version, where the crime is commited with a man: 56.34% of the female respondents confess.

Question 4:

The crime is commited with a ... The Respondent confesses The Respondent denies

Missing

values

Absolute Relative Absolute Relative
sinister man 1154 60.6% 750 39.4% 14
man 242 34.7% 456 65.3% 3
friend 1066 46.6% 1222 53.4% 13

The degree of altruism should not change from question 3 to 4. Maybe the respondents can't see through the way of presenting the problem.

Question 5:

The crime is commited with a ... The Respondent confesses The Respondent denies

Missing

values

Absolute Relative Absolute Relative
sinister man 1049 55.2% 851 44.8% 18
man 369 52.8% 330 47.2% 2
friend 1385 60.6% 900 39.4% 16

There should not be any difference in the answers from the three versions of the questionnaires. The fact that relatively more respondents from the English version chose to confess, maybe because they think that the friend also is a friend of the family, and therefore he would make sure, that the violent gang doesn't hurt the family.

Rationality in an economic sense

If the respondents were minimizing there own penalty and in question 5 maximizing the health of their family, they should choose the combination (confess, confess, deny). 559 respondents have chosen this combination, which corresponds to 12.7%.

Cooperative solutions

348 respondents who commit the crime with a sinister man has answered cooperatively in question number 3 and 4. This corresponds to 18.3%. 77.59% of these are male respondents.

262 respondents who commit the crime with a man has answered cooperatively in question number 3 and 4. This corresponds to 37.5%. 65.65% of these are male respondents.

642 respondents who commit the crime with a friend has answered cooperatively in question number 3 and 4. This corresponds to 28.1%. 49.53% of these are male respondents.

The wheel-of-fortune

Question 8 and 9 are used to test the von Neumann-Morgenstern utility-theory. It's a variant of the Allais' Paradox, the so-called common-ratio-problem.

7) A very eccentric millionaire has set up two stalls at the market place where you can win money, and you don't have to place any bets. At both stalls there is a wheel of fortune with 100 squares. You only have one opportunity to visit one of the stalls. At the first stall you can win 3,000$, no matter where the wheel stops, and at the other stall you can win 4,000$ if the wheel stops at squares number 1-80, and nothing, if the wheel stops at squares number 81-100.

Which stall do you go to?
     

 

8) The eccentric millionaire likes you and lets you visit one of his stalls again. This time you have the chance to win more: at the first stall you can win 3,000$ if the wheel stops at squares number 1-25, and nothing if the wheel stops at 26-100. At the other stall you can win 4,000$ if the wheel stops at squares number 1-20, and you win nothing if the wheel stops at 21-100.

Which stall do you go to?
     

The two questions are identical, except that in question 8 the chances of winning are divided by four. If the respondents choose stall one in the first questions and stall two in the second question, it's a violation of the above mentioned utility-theory.

There is a little difference in the questionnaires. The currency is different, so the $3,000 in the English version is around $450 in the Danish version, $1,800 in the German version, $505 in the French version and $21,300 in the Spanish version.

Questionnaire Stall number 1 is chosen Stall number 2 is chosen Missing

values

Absolute Relative Absolute Relative
Question number 7:
All 5 languages 4001 81.8% 890 18.2% 29
Danish 1447 76.1% 454 23.9% 17
English 1992 86.8% 303 13.2% 6
Spanish 13 65% 7 35% 1
French 438 81% 103 19% 5
German 111 82.8% 23 17.2% 0
Question number 8:
All 5 languages 1744 35.7% 3148 64.3% 28
Danish 545 28.6% 1360 71.4% 13
English 1001 43.7% 1289 56.3% 11
Spanish 3 14.3% 18 85.7% 0
French 162 29.7% 381 70.2% 3
German 33 24.8% 100 75.2% 1

There are four combinations of answers to question number 7 and 8, and in the following table the numbers respondents for each combinations are listed:

Stall number 1 in question 8 Stall number 2 in question 8
Stall number 1 in question 7 1500 2496
Stall number 2 in question 7 236 649

2149 respondents do not change stall from question 7 to question 8. This corresponds to 44%. 2732 respondents have a shift in their risk-preferences from question 7 to question 8. The fact that the combination (1,1) (stall number 1 in each question) is more popular than (2,2) indicates that the respondents are more risk-adverse than risk-lovers (risk-lovers maximize expected utility/payoff). Since there is a difference in the currency and thereby the value of the payoffs, let us divide the above table into languages:

Danish French German English Spanish
(1,1) 434 141 27 895 3
(1,2) 1013 295 83 1095 10
(2,1) 106 18 6 106 0
(2,2) 347 85 17 193 7


Fig. 5: The combinations of answers to question number 7 and 8, divided into languages

Question number 10

10) Imagine that you are in a competition where you have one other competitor. You both have to choose between two options. You do not have the option of communicating with your competitor, but both you and your competitor's choice influence how much you get out of the competition. You will not be informed about what your competitor chooses and your competitor will not be informed of your choice.

You have the choice between playing cards 1 and 2, and at the same time your competitor will have choice between the front or the reverse side of the cards. If, for example, you choose card 1 and your competitor chooses reverse side, then you get the card in the bottom left hand corner, which gives you 0$ each. Both you and your competitor are informed of the prizes, which are:

          Card 1           Card 2
Front side:
 

You get 0$

Your competitor
gets 0$

 

 

You get 0$

Your competitor
gets 0$

 

Reverse side:
 

You get 0$

Your competitor
gets 0$

 

 

You get 100$

Your competitor
gets 100$

 

Which card do you choose?
     

This question is used to introduce the matrix-notation. I also use it to test, whether the respondents seem to answer the questionnaires seriously, because the question has one obvious solution.

494 respondents chose card number 1 in this question. This corresponds to 10.1% of the respondents. Especially this question differs from the others, because the answer depends on the respondents' level of education. Relative more people with lower education have made the wrong choice in this question, but still: 18 people, who have been under education for more than 21 years made the wrong choice.

The people who answered card 1 in this question do not differ significantly in the other questions from the respondents who answer card 2.

Question number 11

11) Imagine that you again are taking part in a competition, like the one above. This time you and your competitor both get 4 options:

You can choose between cards 1 to 4 (your prizes can be read from the top down on the cards below).

Your competitor can choose between the top half of the front sides, the bottom half of the front sides, and the top half of the reverse sides and the bottom half of the reverse sides (your competitor's prizes can be read across on the cards below).

If, for example, you choose card 4, and your competitor chooses the bottom half of the front side, you will get the prize on the bottom half of the card in the top right hand corner, which is 0$. each. The prizes in this fictive competition are:

          Card 1           Card 2           Card 3           Card 4
Front side:
You get 0$

Your competitor
gets 0$


You get 0$

Your competitor
gets 0$

You get 0$

Your competitor
gets 0$


You get 100$

Your competitor
gets 100$

You get 99$

Your competitor
gets 99$


You get 0$

Your competitor
gets 0$

You get 0$

Your competitor
gets 0$


You get 0$

Your competitor
gets 0$

Reverse side:
You get 100$

Your competitor
gets 100$


You get 0$

Your competitor
gets 0$

You get 0$

Your competitor
gets 0$


You get 0$

Your competitor
gets 0$

You get 0$

Your competitor
gets 0$


You get 0$

Your competitor
gets 0$

You get 0$

Your competitor
gets 0$


You get 100$

Your competitor
gets 100$

Which card do you choose?
     

This question is about the focal principle. The philosophy is as follows: If you have to choose between x equal opportunities and 1 opportunity which differs from the other, and your payoff depends upon a competitor's choice in an equal situation, you will choose the opportunity that differs, although this opportunity gives a smaller pay-off. In this game, you can choose between three cards, which gives you a 25% chance of winning $100 or you can choose a card, which gives you a 25% chance of winning $99. The card, which gives you $99 is a focal point, since this differs from the other cards. You might think, that your competitor thinks, that this card differs, and that she thinks, that you think this card differs. If so, there might be a bigger chance of winning $99 than winning $100.

This is the focal principle. But what does my experiment show? Notice, that there are two Danish questionnaires, which differ in question 11. I did this differentiation in order to analyze, whether corner-solutions might exist. A corner-solution could be a focal point too, since this differs from the other solutions.

11)

        Kort 1         Kort 2         Kort 3         Kort 4
Forside:
Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 100 kr.

Din modstander
får 100 kr.

Du får 100 kr.

Din modstander
får 100 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Bagside:
Du får 0 kr.

Din modstander
får 0 kr.


Du får 100 kr.

Din modstander
får 100 kr.

Du får 99 kr.

Din modstander
får 99 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Hvilket kort vælger du?
     

11)

        Kort 1         Kort 2         Kort 3         Kort 4
Forside:
Du får 100 kr.

Din modstander
får 100 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 100 kr.

Din modstander
får 100 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Bagside:
Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 99 kr.

Din modstander
får 99 kr.


Du får 0 kr.

Din modstander
får 0 kr.

Du får 0 kr.

Din modstander
får 0 kr.


Du får 100 kr.

Din modstander
får 100 kr.

Hvilket kort vælger du?
     

The possible focal points are in bold in the table below:

Card # 1 Card # 2 Card # 3 Card # 4 Respondents
Danish, first version 36.0% 24.5% 12.1% 27.4% 339
Danish, second version 30.2% 15.4% 29.2% 25.1% 1560
English, German, French and Spanish version 25.4% 28.3% 15.9% 30.4% 2984

If we exclude the corner-solutions, we see, that there is no reason to believe, that people chose the focal point in this experiment.

Literature and references

Bourque, Linda B.; Fielder, Eve P.: How to conduct self-administered and mail surveys; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Bradburn, Norman M.; Sudman, Seymour: Improving Interview Method and Questionaire Design; Jossey-Bass Publishers, San Francisco, 1980.

Cooper, Russell; DeJong, Douglas V.; Forsythe, Robert; Ross, Thomas W.: Cooperation without Reputation. Working paper, University of Iowa, 1991.

Davis, Douglas D.; Holt, Charles A.: Experimental economics; Princeton University Press, New Jersey, 1992.

Fink, Arlene: How to analyze survey Data; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Fink, Arlene: How to ask Survey questions; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Fink, Arlene: How to design surveys; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Fink, Arlene: The Survey Handbook; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Gibbons, Robert: A primer in Game Theory; Harvester Weatsheaf, 1992.

Jehle, Geoffrey A.: Advanced Microeconomic Theory, Prentice-Hall International, Inc.

Kagel, John H.; Battalio, Raymond C.; Rachlin, Howard; Green, Leonard; Basmann, Robert L.; Klemm, W.R.: Experimental Studies of Consumer Behavior Using Laboratory Animals. Economic Inquiry, 13, 22-38.

Kahneman, D.; Tversky, A.: Prospect Theory: An Analysis of Decision Under Risk. Econometrica, 47, 263-291.

Kreps, David M.: A course in Microeconomic Theory. Harvester Wheatsheaf, 1990.

Litwin, Mark S.: How to measure survey reliability and validity; SAGE Publications, Thousand Oaks, London, New Delhi, 1995.

Reips, Ulf: http://www.psych.unizh.ch/genpsy/Ulf/Lab/WWWExpMethod.html

Varian, Hal R.: Microeconomic Analysis. W.W. Norton & Company, Inc. Third Edition, 1992.


Ulrik Dahl's homepage