and we saw there was a link between interpersonal trust and generalized trust. We looked at the concept of social capital and the ideas of bonding and bridging capital. In this video, we're going to see how social scientists have tried to measure trust in different societies. We'll look at some psychological experiments, and we'll explore the trust question and explain the concept of sampling to you. That's quite an agenda. Now if trust is central to the formation of society and if it's evident that there are different levels of trust in different societies, well, social scientists will want to measure it. Now one way is through social psychological experiments. Now in 1993, researchers at University of Minnesota devised a trust game and tested it with 32 pairs of students. All the students were given a fee of $10 for participating in the experiment. Half of the students were taken to separate rooms and told that they could keep their fee, but if they wanted, they could give some of it away. Whatever they gave away would be trebled and given to a stranger, who'll be told where it came from, and who could return some if he wanted to. Now these strangers were the second group of students, also in separate rooms. They also had had their fee. But they were now given extra from their unknown partner, and we're told they could keep some of it or return some of it. So let's have a look at the results. The students started with $10, and on average gave away $5.15. This meant that the receiving group got something over 15, $15.50. The average amount they returned, however, was only $4.66. Only 40% of the first group received a positive payback from their gift. The rest lost out. When the experiment was repeated, but this time telling all the participants the results of the previous run, the amounts gifted increased slightly. But the payback now swung
into positive territory,
climbing up to just under $6.50. The researchers suggested that the different outcomes could be explained by the fact that the first experiment relied purely on socialized norms of behavior, those inherent in the students. Where as in the second experiment, they'd actually learned and were able to make strategic calculations, and this led to an increase in mutual trust. Now, variants of this game has been played ever since. Usually beginning by asking participants if they trust other people and linking this to their observed behavior. But before leaving this experiment, let's have a look at a couple of points. The results differ if the sums involved were larger. The results were also different if the participants were drawn randomly from the population. The students knew that the other group were also students, and this probably influenced their behavior. And finally, the calculation need not always enhance trust. In one variant of the game, the sum given away was not returned to the individual, was put into a common pot from which they would all benefit, rather like voluntary form of taxation, which would be spread out afterwards. The game is then played in several rounds, and the results are announced after each round. Now as the game progresses, what's interesting is the amount being given decreases, and those that give absolutely nothing actually increases. In other words, they're avoiding paying taxes and allowing others to do so. Our second trust experiment is known as the dropped wallet test, devised by The Reader's Digest. In 1996, they dropped a dozen wallets in each of 20 European cities, to see how many would be returned. Each wallet contained $50 in local currency, an address card, a family photograph, as well as other personal items. It's subsequently repeated this experiment in different locations. Most recently in 2013, when it visited 15 different world cities. Let's look through the results. Helsinki came out best with 11
of the 12 wallets returned hm,
Scandinavian cities generally do well in these tests. Next came Mumbai, with nine, followed by Budapest, New York, with eight each. Amsterdam and Moscow followed with seven apiece. Thought, I often wonder how Amsterdam would have done if we used unlocked bicycles instead of wallets. Berlin and Ljubljana came next with six. And after that, you've got a less than 50/50 chance of seeing your wallet again. So let's run through the rest. London and Warsaw has five each, Bucharest, Rio and Zrich had four each, Prague had three, and at the bottom of the list, Madrid returned only two wallets. Since 1996, social scientists have dropped wallets all the place, which is quite expensive actually, bearing in mind that there's less than a 50/50 chance of ever seeing it again. Social scientists though have gotten around this problem by asking people whether they expect to have their wallets returned and measuring the results that way. Saves a lot of money. Now actually asking people is the more usual approach to ascertaining knowledge about trust. In 1956, a social scientist Morris Rosenberg devised what has become the standard trust question. It goes as follows. Generally speaking, would you say that most people can be trusted or you cannot be too careful when dealing with other people? Since 1981, this trust question has been included in what is known as a world values survey. These surveys ask about a thousand respondents in a whole host of countries a series of questions on their attitudes on a range of social issues. There've been six waves of this survey so far. The results of the most recent one were published in April 2014. Now the next question we need to ask is whether experiments with 12 wallets or 30 pairs of students or even a thousand respondents can actually say something about a country as a whole. And the answer is yes. And you get it by what we call sampling. First of all,
you have to define a population.
Now this is a statistical term meaning the entire entity. This case, it actually is the adult population. Now during sampling, it is any subgroup should show the same characteristics as the population as a whole. But the sample must be of sufficient size, and it must be random. Let's take an example. For sake of argument, assume that society is made up equally of men and women. Choose four people at random, and you should get two men and two women. But you could get three men and one woman. The sample is simply too small. But choose 100 people at random, and you should get close to 50 men and 50 women. You might get 49, 51 or even 52, 48. So there's always a margin of error or a confidence level in your results. So you can ask a population whether it trusts other people or whether it shouldn't be too careful when dealing with other people. To sum up, in this video, we looked at two trust experiments. We introduced the trust question, and we've discussed the idea of sampling. In the next video, we'll look at the results, and we'll see how reliable they are.