Data Analysis of Graduate Admission

Data is an important tool used to analyze trends, develop hypotheses and predict potential outcomes. It can also miss important factors or hide potential variables to help get to the root cause of an issue or point out a trend. An example of this would the below graph of the admissions for graduate programs at the University or California at Berkeley. When looking at the overall picture that is presented by the data, this shows more male students are being accepted in comparison to women. While this shows a current bias, this may not paint the whole picture.

When you take a step back to review the data as whole you can see some additional insights that may be the reason for the difference in the amount of accepted students. The below graph shows that more men in total have applied for the graduate programs. With a larger population there would be a potential for more students acccepted however the graph also shows the amount of male and female students is potentially equal for each department.

The data can also be further broken down to compare the actual programs and the amount of male and female applicants. When comparing the acceptance rates of all the departments, in the graph and tables below, the data shows the difference in acceptance between genders is almost equal. These additional data points provide a better picture that the bias is not a great as it was first believed.

We can break this down further to see if there is an actual bias in Department A. To see if there the graph below shows that this Department had 8 times more male applicants than female students. This could skew the data to show that the amount of women accepted is less and a bias present. However, when looking at the table you compare the amount of accepted to the total applied for each gender respectively you see that 82% of women were accepted and 62% of men were accepted. This is an example of Simpson’s Paradox. The first glance of the data shows the bias towards male students but when you break down and look at the relative frequency of acceptance the data actually shows more female students are being accepted, which is the exact opposite of the original shown data.

In summation there is no bias present to gender bias for acceptance.

**Citations

misc{towardsdatascience2023, author = {Vega, Santiago}, title = {Simpson’s Paradox and Interpreting Data}, year = {2023}, url = {https://towardsdatascience.com/simpsons-paradox-and-interpreting-data-6a0443516765}, note = {Accessed: 2024-09-10}

```