This study investigated whether finding a coin influenced a person's likelihood of mailing a sealed but addressed letter that appeared to have been accidentally left in a conspicuous place. Several variables were collected during the experiment, including two randomized variables of whether there was a coin to be found and whether the letter already had a stamp on it.
Format
A data frame with 42 observations on the following 4 variables.
- stamped
a factor with levels
noyes- found_coin
a factor with levels
coinno_coin- gender
a factor with levels
femalemale- mailed_letter
a factor with levels
noyes
Source
Levin PF, Isen AM. 1975. Studies on the Effect of Feeling Good on Helping. Sociometry 31(1), p141-147.
Details
The precise context was in a phone booth (this study is from the 1970s!), where a person who entered a phone booth would find a dime in the phone tray, which would be sufficient to pay for their phone call. There was also a letter next to the phone, which sometimes had a stamp on it.
Examples
table(mail_me)
#> , , gender = female, mailed_letter = no
#>
#> found_coin
#> stamped coin no_coin
#> no 0 3
#> yes 0 5
#>
#> , , gender = male, mailed_letter = no
#>
#> found_coin
#> stamped coin no_coin
#> no 1 6
#> yes 1 4
#>
#> , , gender = female, mailed_letter = yes
#>
#> found_coin
#> stamped coin no_coin
#> no 3 1
#> yes 4 0
#>
#> , , gender = male, mailed_letter = yes
#>
#> found_coin
#> stamped coin no_coin
#> no 4 0
#> yes 6 4
#>
(x <- table(mail_me[, c("mailed_letter", "found_coin")]))
#> found_coin
#> mailed_letter coin no_coin
#> no 2 18
#> yes 17 5
chisq.test(x)
#>
#> Pearson's Chi-squared test with Yates' continuity correction
#>
#> data: x
#> X-squared = 16.519, df = 1, p-value = 0.00004817
#>
(x <- table(mail_me[, c("mailed_letter", "stamped")]))
#> stamped
#> mailed_letter no yes
#> no 10 10
#> yes 8 14
chisq.test(x)
#>
#> Pearson's Chi-squared test with Yates' continuity correction
#>
#> data: x
#> X-squared = 0.33608, df = 1, p-value = 0.5621
#>
m <- glm(mailed_letter ~ stamped + found_coin + gender,
data = mail_me,
family = binomial
)
summary(m)
#>
#> Call:
#> glm(formula = mailed_letter ~ stamped + found_coin + gender,
#> family = binomial, data = mail_me)
#>
#> Coefficients:
#> Estimate Std. Error z value Pr(>|z|)
#> (Intercept) 1.5626 0.9868 1.583 0.113312
#> stampedyes 1.0018 0.9133 1.097 0.272661
#> found_coinno_coin -3.5752 0.9609 -3.721 0.000199 ***
#> gendermale 0.1544 0.8781 0.176 0.860387
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
#> (Dispersion parameter for binomial family taken to be 1)
#>
#> Null deviance: 58.129 on 41 degrees of freedom
#> Residual deviance: 35.538 on 38 degrees of freedom
#> AIC: 43.538
#>
#> Number of Fisher Scoring iterations: 5
#>