| Page Summary This page …. |
Related Pages |
One dimensional data
Pie Charts
Multiple Data Series
One Point Something Dimensional Data
Frequency distributions
Two Dimensional Data
Suppose we have a simple 2x2 table with the dimensions SEX and EDUCATIONAL ATTAINMENT.
Our general rule with tables like this is PERCENTAGE DOWN, COMPARE ACROSS. This assumes that the INDEPENDENT (CAUSE) variable is in the columns and the DEPENDENT (EFFECT) is in the rows. We ALWAYS want to compare the rate of the dependent variable between the categories of the independent variable.
Look slowly at the above table and make sure you see what the percentages are OF.
We would describe this table as being EDUCATION by SEX. We are looking at the outcome variable EDUCATION broken down by SEX. The pattern is DV by IV. The comparison we want to do is the percentage of the boys who have only high school education to the percentage of the girls who have only a high school education.
STOP AND THINK: Can you see why the second comparison is somewhat moot?
So, which of these charts allows us to make this comparison?
 |
 |
The right hand chart is the one that let's us make the comparisons we want by looking at bars side by side. Note that the legend describes the "inner most" comparison and the horizontal (category) axis labels describe the "outermost" comparison.
| More generally, we have |
 |
 |
If we reverse the table, like this:
we would name it "Sex by Education." Now the comparison we are making is "of the kids in high school, what percentage are boys, compared to college," or, to put it a different way, "how does the sex ratio among those with just high school compare to the sex ratio of those with college degree."
