Testing people's honesty...

A sociological experiment was conducted in Illinois called the “Wallet Test”. You can read about it at http://www.wallettest.com/

The researchers deliberately dropped 100 identical wallets (containing money and ID) in public places to test honesty.

The test results were broken down by age, gender and race ----> http://www.wallettest.com/Lost_Wallet_Test/Results_Page.html


What I’d like to know is why did men steal over twice as much as women?


Because women are more interested to find out to whom the wallet belong to than to its immediate content.

I wonder what the results would be with the wallets men found if they had a hot chick in the ID photo.

Just a WAG, but maybe the women were more honest on the whole because of their ability to empathize- i.e. they find the wallet, think to themselves how devastated they’d be if they lost the wallet (and how happy they’d be if someone returned it), so they returned it.

The men, with a less empathetic view, don’t really consider the feelings of the person who lost the wallet and only with practical concerns like “Will I get caught?” etc.

Like I said, just a guess using very broad generalizations.

They didn’t, in actual fact the exact opposite is true: all other factors being equal males stole less than one quarter as often as females.

The apparent sex descrepancy has nothing to do with gender and is entirely attributable to age. A classic case of assuming statistical significance when even a cursory glance would tell you that none exists. Results are strongly skewed by age, and there are more almost twice as many Young Males as Young Females .

Age ratio (F:M) (Old VS Young) 39O:12Y vs 27O:22Y or 3.25F : 1.22M
Age honesty ratio (young vs not young) 19:15Y vs 55:11O or 1.3Y : 5O

So if age ratios were maintained independent of sex we’d predict an honesty ratio equivalent of 3.25F : 0.24M = 13.54

Observed sex honesty ratio (F:M) 86:14F vs 61:39M or 6.1F : 1.6M = 0.28

If all other factors were equal we would have expected less than 12 men to be honest simply because of the age discrepancy. Instead we find a whopping 39 men were honest.

The way people interpret this study and the way the researchers have presented their results would make a good subject for an introductory stats class. Beware your assumptions and control for variables.

Nonsense; absolute nonsense.

I would expect everyone to be honest. Pretending, on the basis of statistical reasoning that age should change that expectation and therefore “we would have expected fewer than 12 men to be honest simply because of the age discrepancy…” is hysterically funny to me.

But your post is a good example of statistically distorting common sense, and it’s a good example of why any good statistician can put any spin needed on what is otherwise obvious.

It’s a silly study, with uncontrolled variables, tiny numbers, etc etc. But one thing it does not show is that three times as many many as expected were honest unless it is a statistician creating the expectation by choosing, a priori, youth as the baseline standard to make the rest of a group look good.

You can expect penguins to rain from the sky if you wish. What you like is entirley irrelevant to what objectively is. The fact is that based on this study honesty increases with age. That’s not an opinion, it’s not an expectation, it’s a fact.

You are free to criticise or even reject the study in its entirely if you wish. What you are not free to do is spout ignorant nonsense.

Which simply confirms that you have absolutely no understanidng of probability, science or logic.

Yes, like Homer Simpsons said “Facts are meaningless, they can be used to prove anything that’s even remotely true.”

Right :rolleyes:

a priori? You should not use that word. I do not think it means what you think it means.

The original investigators stated that they never had any a priori knowledge of the effect of age. Nor could you possibly be implying a priori choice on my part since the decision to use age as a category was made by the original invetsigators, not myself.

I won’t even go into the ignorance and total lack of logic required to argue that an analysis shouldn’t choose a specific division because it is being applied to another division selected in exactly the same manner. The mind boggles that anyone could not see the double standard and lack of rigor in that criticism.

I repeat, the way people interpret this study (and the reaction to my back-of-the-envelope analysis) would make a good subject for an introductory stats class.

Let’s see here…hmmmm…

All comers: 86% of women and 61% of men returned the wallets.

Older group: 91% of women and 85% of men returned the wallets.
Middle-age group: 86% of women and 71% of men returned the wallets.
Younger group: 83% of women and 41% of men returned the wallets.

I’ll get right on that statistics course so that I understand how “…all other factors being equal males stole less than one quarter as often as females.”

In the meantime I must admit I will continue to hope (at least based on this study) that a woman and not a man finds my wallet.

But thanks for the chuckle.

I’m having a little trouble following your math. I agree that you can’t just look at the grand total honesty percentage in men and women. But couldn’t you just look at the breakdown by age and gender?

How would you explain the fact that when the results were broken down age and gender, women were slightly more honest than men in 2 of the age categories and a lot more honest than men in 1 of the age categories?

I had a similar thought. I agree with Blake’s general point, but I think there might be an error in his math somewhere.

By the way, given that the wallet contained only $2.10 in actual money, I wonder if its an honesty test or a laziness test.


The men, with a less empathetic view, don’t really consider the feelings of the person who lost the wallet and only with practical concerns like “Will I get caught?” etc.

Well you could have knocked me over with a feather when I discovered (or more likely was informed) that “empathetic” not only had a different meaning from “sympathetic,” but it is not even a word. Since being enlightened and learning the actual word is “empathic” I feel compelled to share. Sorry if you prefer not to know.

What I would like to know is why would anyone draw any conclusions at all from this. Honesty may have played no part at all. Myabe failure to return the wallet was a function of either laziness or forgetfulness or both. Maybe those results that can measure honesty were based on economics rather than gender or age, or perhaps culture, religion or national origin. There are so many possibilities and the sample is so small (as was the amount of cash) I wonder what the point really was.

That thought occurred to me too. $30 is a lot more money to a 15-year-old than a 35-year-old.

Here’s a thought on how to do the test:

Put a phoney $5000 bearer bond in the wallet that is supposedly redeemable at a particular bank. See who shows up at the bank trying to cash the bond.

I am bothered by the insistence that not returning the wallet is the same as stealing it. The wallet was found, not stolen.

For all we know, the people who did not return the wallet simply threw it in the trash.

The age variable is really an economic variable, IMO. I bet that men are less honest than women, but that primarily, people with less money are less likely to return a found wallet.

Also that site is full of gems like this:

“Age ratio (F:M) (Old VS Young) 39O:12Y vs 27O:22Y or 3.25F : 1.22M
Age honesty ratio (young vs not young) 19:15Y vs 55:11O or 1.3Y : 5O”
I think one problem here is you’ve done old vs young in the top line then young vs old in the second line.

I think you have to keep the order the same or you get problems? Also you seem to have carried over the 3.25 instead of carrying over 3.25/1.22.

My take is it should be 2.66/3.84, ie 0.69 vs observed of 0.28, making women more honest than expected.


Exactly! Anecdote alert but I found a wallet one night in a nightclub, actually someone else did but was all on for stealing the contents so I took it off him and was going to hand it in to the bar. I noticed when I looked in the wallet that the owner was an acquaintance of mine. I took the wallet home and brought it to her house the next day. She was happy to get it back because although it contained little or no cash it had her id cards and I think debit card etc. Going to her house required somewhat of detour from the direction I was going that day. However, it was a Sunday and I had no particular place to be at a certain time so calling by her house wasn’t that big an issue. I gave it to her, she thanked me offered no reward (nor did I expect one) and I was on my way.
If I had found the wallet at another time of the week I’m not sure how certain I am I would have returned it, not out of dishonesty but out of laziness. Knowing the girl I suppose I could have waited til I saw her again in the club, us both being regulars there.

I’m much more intrigued by the companion study, in which the original owners of 100 lost wallets were secretly observed to determine whether or not they wept.

There are significant problems with this analysis. The easiest disputation is to just look at the website where they break it down for you:

As some other posters mentioned, there are a few things that they did not account for:

  1. Effect of quantity of money.
  2. Effect of gender of wallet owner.
  3. Effect of attractiveness of wallet owner.

Although, I think a poor ugly guy is probably the safest way to go.