Quote:
Originally Posted by tame_deuces
Both socio-economic status and country of origin are included in the regression analysis and thus controlled for.
I would have liked to see more discussion of it. There's a bit of black-box-ness going on (which I acknowledge is not uncommon) because there are lots of ways to try to control for various variables.
The most common way of doing this sort of thing for numerical data is to just have a whole bunch of variables in a linear combination. So for this study it would be something like
Altruism = A * Age + R * Religiosity + S * SocioeconomicStatus
And then you run the linear regression on the data set and get the coefficients A, R, and S. And, at least in this case, they would have "controlled for socioeconomic status" because you have an equation where you can hold that variable fixed and look at how age and religiosity influence the shape of the graph.
But that isn't what they seem to have done, because they have a graph of just alturism, age, and religiosity, and there's not information about the specific socioeconomic level they're looking at. But they list N = 1170 on that graph. So where is that data? Did they do some sort of projection of the data onto the three dimensional Altruism/Age/Religiosity subspace, possibly by just ignoring the SocioeconomicStatus variable? Because that would be a very bad (and potentially misleading) way to present the data.
Quote:
Originally Posted by article
Results from a linear regression with number of stickers shared as the dependent variable and age (1-year bins), country of origin, socioeconomic status (SES), and religious identification of the household
(dummy coded) suggest that ...
Here, we have that they are dummy coding the categorical data. The charitable reading of this is that they've dummy coded both the religious identification and the country of origin. (I assume they did, because to code country of origin data without dummy coding would be really bad.)
But did they do any analysis of interaction effects between the two classes of categorical data? I didn't see anything in the paper about that. It seems to me that country of origin and religious identification are going to be variables that are related, and there might be something else there.
I'm not saying that they necessarily made mistakes in the analysis. Maybe they did it but didn't report that aspect of it. I do know that a lot of people use statistical tools in the wrong way, and that the more complicated the data they're working with the more ways there are of screwing it up, and the more careful they need to be.
And so that's why when I see N=18 Buddhism and N=510 Muslim, it gives me pause. I see a high potential for overrepresentation in one country, and it may have influenced the regression, especially because it's not entirely clear what their controlling mechanism was. There are just a whole lot of pitfalls here.
-----
Here's a link about the interaction of dummy coded variables. It may be too mathy for most, so I've just snipped it up. The data set has three variables: job prestige (numberical), marital status (categorical), and gender (categorical). And then it proceeds to do an analysis
http://www.theanalysisfactor.com/int...ar-regression/
Quote:
First question: Do married people have more prestigious jobs than non-married people?
The answer is yes. On average, married people have a job prestige score of 5.9 points higher than those not married.
...
Next question: Do men have jobs with higher prestige scores than women?
The answer is no. Our output tells us there is no difference between men and women as far as the prestige of their job.
...
But can we conclude that for all situations? Is it possible that the difference between job prestige scores for unmarried men and women is different than the difference between married men and women?
...
Here’s where the concept of interaction comes in. We need to use an interaction term to determine that. With the interaction we’ll generate predicted job prestige values for the following four groups: male-unmarried, female-unmarried, male-married and female-married.
...
If we had not used the interaction we would have concluded that there is no difference in the average job prestige score between men and women. By using the interaction, we found that there is a different relationship between gender and job prestige for unmarried compared to married people.
In other words, even though the original analysis showed no gender in job prestige, there is one. But you would have to parse the data by marital status in order to see it, and you can't see it if the two different categorical variables were left independent of each other.