Association between nominal and continuous variables

[hellofolks]

Suppose you have two random samples of the profit margins obtained by two
stock traders, Trader A and Trader B. The first data set consists of 18
data, and the second has 21.

I want to check if there is association between the variables
"type of trader" (values A or B) and "profit margin". In words,
I need to find out if the choice of trader influences the profit
margin obtained.

Since the observations are not paired and, after all,
the numbers of observations are distinct, I cannot use
the linear correlation coefficient formula.

I know there is a formula to verify if a nominal (discrete)
variable and a continuous varible are associated. I also
know it involves the means and standard deviations of
both data sets. I have been told, too, that it gives a result
ranging from 0 to 1 and that, the closer it is to 1, the stroger
is the association.

The trouble is I missed the class about the formula
and it has proven really hard to find it on the web, since
I don't know its name. So I would really appreciate if you guys
could write it down for me. I need it as soon as possible and
I can't ask my teacher right now; so I really need your help.

If the data sets are needed, I can provide them. But I don't think
they're necessary, now that it's only the formula what I need. I
already know how to solve the problem.

Thanks in advance,
hellofolks.

 

[Valkarie]

The formula you are looking for is called the Mann-Whitney U test. It is used to compare two independent samples to determine if there is a statistically significant difference between the two samples. The formula is: U = n1n2 + (n1(n1+1))/2 -∑Ri where n1 and n2 are the number of observations in each sample, and ∑Ri is the sum of the ranks from the first sample. The resulting U value can then be compared to a critical value from a Mann-Whitney table to determine if the difference between the two samples is statistically significant.