Table 3: Education and Work Performance
High School or Less(1) / Male[Dr. Brown1] / Female(22-32 points) – Excellent Job Performance / (0)0% / (5) 12%
(15-21points) – Good job performance / (30) 75% / (10) 25%
(8-14 points)– Poor job performance / (10) 25% / (5) 12%
(0-7 points) – Very poor job performance / (0) 0% / (20) 50%
TOTALS / (40) 100% / (40) 100%
Associates Degree(2) / Male / Female
(22-32 points) – Excellent Job Performance / (0) 0% / (0) 0%
(15-21points) – Good job performance / (30) 75% / (40) 80%
(8-14 points)– Poor job performance / (10) 25% / (0) 0%
(0-7 points) – Very poor job performance / (0) 0% / (10) 20%
TOTALS / (40) 100% / (50) 100%
Bachelor’s Degree(2) / Male / Female
(22-32 points) – Excellent Job Performance / (0) 0% / (5) 8%
(15-21points) – Good job performance / (50) 50% / (45) 75%
(8-14 points)– Poor job performance / (34) 34% / (0) 0%
(0-7 points) – Very poor job performance / (16) 16% / (10) 17%
TOTALS / (100) 100% / (60) 100%
Degree Beyond a Bachelor’s Degree(3) / Male / Female
(22-32 points) – Excellent Job Performance / (0) 0% / (5) 10%
(15-21points) – Good job performance / (20) 100% / (40) 80%
(8-14 points)– Poor job performance / (0) 0% / (0) 0%
(0-7 points) – Very poor job performance / (0) 0% / (5) 10%
TOTALS / (20) 100% / (50) 100%
Continuation of tables explanatio[Dr. Brown2]n.
[Dr. Brown1]The First thing you must do is figure out if education makes a difference on the dependent variable (level of job performance) for males. To do this you compare each category of the dependent variable for males for each level of education. Thus, excellent job performance at the level of high school or less for males with excellent job performance at the level of associates degree with males with excellent job performance at the level of bachelors degree for males, with excellent job performance at the level of degree beyond a bachelors degree for males. Notice that there are no males in the excellent job performance categories regardless of their level of education. Because of this, we don’t count the “excellent job performance category for males” (but we do for females since we do have some in the category at the bachelors and at the degree beyond a bachelor’s degree level.
Keep in mind that even though we have 16% in the cell “Very Poor Job Performance” for males with a bachelors’ degree, we have no other males in the very poor job performance at any other educational level. Since this is the case, we still count the category of “very poor job performance” as an active category, and we compare the 16% against the “0%”’s. Since there’s more than a 15% difference between the two, we highlight it.
You then do the same thing with the next category: good job performance at the level of high school or less for males with good job performance at the associates level for males, etc. You must do this for the remaining categories. In every case you take the highest and the lowest amount and highlight these two IF THEY ARE MORE THAN 15% IN DIFFERENCE. Be sure to always use different colors when you highlight the differences in your categories (e.g. like we did for females… we used the colors green and blue so there are two color differences).
Thus, you’ll notice that for the males, after doing all of the above, there are three sets of highlighting (yellow, red, and pink). Since there are three active categories of the dependent variable (level of job performance) we must have at least two sets of differences (in our case, we have more than two, we actually have three). Notice that we only have three active categories of the dependent variable for males because we didn’t have any males in the first category (excellent job performance) at any education level.
You then repeat the same steps above for females. In the females case we have four active categories of the dependent variable. Thus, we must have at least two sets of differences (in other words, two different highlighted colors
[Dr. Brown2]
Now, here’s what the colors tell us: Since the males have three colors (yellow, red & pink) and since 3 is at least 50% of 3 (there are three active categories of the dependent variable, level of job satisfaction, for males), this means that education DOES HAVE an effect for males.
Suppose, however, that the numbers were different. Suppose that the males only had one color (only yellow and not red or pink) instead of 2 or 3. Since 1 is not at least 50% of 3 (there are three active categories of the dependent variable, level of job satisfaction, for males), this would have meant that education DOES NOT HAVE an effect for males.
Since the females had 2 different colors (green and blue) and since 2 is at least 50% of 4 (there are four active categories of the dependent variable, level of job satisfaction, for females), then this means that education DOES HAVE an effect for females.
In cases like this where it does have an effect for both of the categories of your independent variable (in this case the males and females) then you reject the hypothesis up to that point.
In cases where it does not have an effect for one of the categories of your independent variable (for example, if the males only had yellow highlighted and not pink) and does have an effect for the other category (females) then you would partially except the hypothesis at that point.
So, the tables should be arranged in the following order:
- Tables that have no effect for either category of your independent variable (in this case, for males or females). This is evidence for a direct relationship and the acceptance of your hypothesis.
- Tables that have an effect for one category but not the other category of your independent variable (in this case, for males or females). This is evidence for partially accepting your hypothesis.
- Tables that have an effect for both categories of your independent variable (in this case, for males or females). This is evidence for spuriousness so you would reject your hypothesis.