I am Laura Belloni, and this is my thesis, conducted with professors Giovanna Guerrini and Nicoletta Noceti. The aim is to analyze gender disparities at the University of Genoa for the bachelor's degree in Computer Science, which I attended.

The data is official and comes from CEDIA: Centro Dati, Informatica di Ateneo, covering the years 2010 to 2024. It refers only to passed exams, with at least 50 grades recorded. By gender I will refer to the gender indicated by the students.

These are some of the analyses conducted, which represent the summary of the results. Let's take a look at them.

The first interesting result from this analysis is that the percentage of women in the bachelor's degree in Computer Science is 14.44%.

This percentage remains consistent when analyzing the exam grades of women compared to men. The average grades do not differ significantly by gender, with a difference of only 0.5.

We observe that the grade distribution resembles a normal distribution centered at 24, with peaks at the highest grades.

In the same graph, but broken down by gender, we see a higher percentage of women with lower grades and a higher percentage of men with higher grades. In particular, we notice differences in the honors grades, i.e., the 31s (honors).

Grouping the results into low, medium, and high grades, we see that the distribution is symmetrical, with the colors reversed. This is the result I obtained most often, but below, I have reported the more interesting results (those that deviate from this).

The graduation rate is 37%. In 12 years, there have been 547 graduates. The numbers are consistent; the same percentages recur: 14% and 37%. The percentage of female graduates, 14%, is consistent with the percentage of women in the program, and the percentage of female graduates, 36%, is consistent with the overall graduation rate. It is interesting to note that in 12 years, only 74 women have graduated, which means approximately 6 women and 39 men graduate each year.

The average graduation grades for women are slightly lower, but not significantly, with a difference of 1.25.

However, the distribution is significant, particularly in low grades and honors.

Grouping the grades into low, medium, and high categories, we observe that low grades are equally distributed between genders, while medium and high grades resemble a staircase. The percentage of women is higher in the medium grades, and the percentage of men is higher in the high grades. It's important to remember that the sample size is small, as there are only 74 female graduates, which may not be sufficient to draw conclusions.

Comparing high school graduation grades with university graduation grades, I noticed that, on average, women enter with a high school grade 4.5 points higher than men, but they graduate with a grade 1 point lower than men. This is based on a single sample, the graduates, which, as we recall, includes a limited number of women.

It's further interesting to note that the distribution of high school grades does not follow the same pattern we've seen so far in exam results. In fact, we see almost the opposite result: men have a higher percentage of low grades compared to women, and women have a higher percentage of high grades compared to men.

As for honors, we see that the probability of graduating with honors is 19%, but for women, it is 16%. Over 12 years, only 12 women graduated with honors, meaning 1 per year. In exams, the probability of receiving honors is 10%, but for women, it is 7%. In 12 years of exams, there were 118 honors awarded to women, meaning 10 honors per year. This is particularly significant in my opinion because this analysis includes both non-graduates and elective course exams. Additionally, we must remember that a student must pass at least 20 exams in their degree program to graduate.

Comparing written and oral exams by gender, I did not detect a significant difference. However, using t-tests, I was able to verify that the difference between written and oral exams is more significant. This is important to remember when analyzing future results.

We see that the distribution of oral exams is skewed towards higher grades, while written exams are skewed towards lower grades. Honors remain often inequitable.

Grouping grades into low, medium, and high categories, we again find the usual symmetrical distribution, but with different trends in the distributions.

I highlight three distributions of specific exams that show peculiar and differing results from one another. The first, on the left, has a distribution consistent with the general results, similar to a normal distribution centered on 24, with peaks in the highest grades: 30 and honors. The second distribution is similar but has peaks in the lower grades: 18 and 19. The last distribution on the right is completely skewed towards low grades; it might be the tail of the distribution centered on failing grades.

Looking at these graphs by gender, we notice that in the first exam, the grades are approximately equal for low grades. For high grades, women have a higher percentage, except for honors, where there is always a lower percentage of women. We specify that this exam is from the first year, so the sample size is rather large. In the second exam, we cannot assume much as the distribution is incomplete due to a sample of only 33 women. Notably, in 8 years of exams, no woman received honors. In the third exam, we again face a small sample of 41 women. It's important to note, particularly in this case, the scales of the graphs, as the heights differ and therefore the percentages on the scale differ as well. In the first graph, we have a scale with a maximum of 13%, in the second 18%, and in the third exam, a maximum of 50%. Observing this last graph, we also see that 45% of the women who passed this exam received an 18. It would still be concerning that 20% of the men received an 18, but it’s half as much. We also observe that in 8 years, none of these 41 women ever received a 28, a 29, or honors.

Comparing grades by the gender of the professors, I excluded this last case just observed, as it would skew the statistics. I consider it an exception. The grades given by female professors are evenly distributed, except for honors. It seems that 30s were awarded instead of 31s to women or vice versa to men. Male professors, on the other hand, have the same symmetrical distribution, consistent with the other data obtained.

An additional analysis I was able to conduct was thanks to the dataset from Professor Guerrini, which allowed me to verify whether the exam format affected the grades by gender. In particular, I compared two formats for written exams: quizzes and exercises on the same subject. Both were conducted on the same day, with the quiz taken on the computer and the exercise done on paper. The average grades were not particularly significant; however, the distribution in the graphs is interesting.

In the quiz, we see that the grades are symmetrical, like the usual results. In 3 years, women, who number 69, did not receive a single honor. This result is very important because the quiz is conducted on a computer with automatic grading. This may highlight a systemic issue rather than just human judgment. For the exercise grades, we observe one of the few cases where women have a higher percentage of high grades than men, and vice versa. In particular, we see that honors are higher for women, although the scale on which we are analyzing is not particularly significant, with a difference of 2 or 3%. Notice that both tests are on the same subject, and both include theory and practice in different ways.

I conclude by reporting the statistical test, a two-sample t-test, with a confidence level of 95%.

By obtaining a very low p-value, I was able to reject the null hypothesis in favor of the alternative. I can therefore affirm that there IS a statistically significant difference between the average grades of women and men.

Thank you for your attention.