Row or Column Percentages

[Yawo Kokuvi]

This is probably a basic question, but I thought I would ask for further clarification from here. We’ve been thought through grad school to compute percentages over categories of independent variables.

For example, assuming we want to find gender (IV in rows) differences in levels of victimization(DV in columns), then we can compute row percentages. This will allow us to make claims about the proportion of men or we who are victimized.

But under what circumstances would you calculate and use column percentages for comparison? What kind of research questions would column percentages answer?

thanks again, and I appreciate your responses in advance?

cY

[Igor Paploski]

Don't get too attached to a percentage being the "row" or "column" one. For me, it's just a matter of flipping the table. Both percentages can be of interest of the researcher, depends on the context and on the question. Consider the 2x2 table below:

...............Disease
...............+........ -
Sex M.....a........ b
........F.....c........ d

If your question is in what frequency people of different sexes have the disease, then you would calculate the row percentage (a / a+b and c / c+d). If your question is what is the proportion of males among the disease and non-disease groups, then you would calculate the column percentages (a / a+c and b / b+d). You can flip the table very easily. In Stata, instead of tab sex disease you can very easily issue tab disease sex. As long as you are careful with the interpretation, it doesn't change anything. That's why I suggest not getting attached to the percentage being the "row" or "column" one - look at the table, reconstruct in your head what it is displaying and then interpret it.

[Yawo Kokuvi]

Thanks very much - The explanation is clear ...

best wishes - CY