Which states have taken the greatest advantage of the rapid growth of AP Computer Science Principles to attract students from a broader demographic to CS?
Background
The introduction of AP CS Principles by the College Board in 2016-17 represented a significant “cultural” shift in K12 computer science education. AP Computer Science Principles is an introductory course in computer science designed to attract students from a broad demographic with little previous CS experience. Until AP CS Principles, students could only take AP Computer Science A – a challenging object-oriented programming course in Java – which tended to attract students who already had a strong background in CS often gained outside of the classroom.
Since its introduction, the number of students taking AP CS P has grown quickly while AP CS A has remained flat. By the 2017-18 school year, the number of students taking the AP CS P exam had already surpassed those taking AP CS A, and in 2020, students took 65,000 AP CSA exams and 114,188 AP CSP exams.
AP CS P has attracted students from under-represented populations to CS as intended. The below chart from Page 9 of AP® Computer Science Principles and the STEM and Computer Science Pipelines published in Dec 2020 by the College Board shows differences in the demographics of AP CS A and AP CS P.
About the data
There are many ways to increase the number of students – particularly those from underserved populations – learning computer science in high school. AP Computer Science is just one of them. In an earlier post, we learned that 22.5% of students taking a computer science course took an AP Computer Science exam. AP Computer Science is just the apex of the CAPE pyramid and is not the only way to define the strength and diversity of a state’s computer science education program. However, the number of students taking a computer science course is a new metric still not available for all states. AP CS data has been around since at least 1998 and is available for all states. Further, while “everyone” knows what “AP” is, the concept of a “foundational” computer science course that contains at least 20 hours of programming that underlie the statistics in the State of Computer Science Education reports is little known.
Unfortunately, even this AP data is at risk. The data used for the study below comes from data stored on Dropbox and available from Barbara Ericson’s Computing for Everyone blog. Dr. Ericson compiled her data from detailed reports provided by the College Board on test results broken down by state and demographics for each year. However, the College Board did not provide this detailed data for the 2021 exams and even removed detailed data from previous years from their website. Mark Guzidal also lamented this loss of data in his blog post on May 17. Dr. Ericson’s compiled data is the best and perhaps only public source of this detailed AP data at present and is only available through 2020 exams
The start of the COVID pandemic affected the 2020 AP exams in very different ways depending on the state. I could neither use all 2019 data nor all 2020 data. For this study, I decided to use the data for whatever year had the most exam takers for that exam for that state. For example, if a state had more AP CSA students in 2019 than in 2020 but had more AP CSP students in 2020 than in 2019, I used the CSA data from 2020 and the CSP data from 2019. Because of how I decided to combine data from exams from 2019 and 2020, some of the following results may be somewhat different from those reported elsewhere.
Pass Rates
My previous blog post on AP CS test-taking rates stated that the overall pass rate for AP CS was 70.5%, and the leading states were Montana (78.8%), New Jersey (78.1%), and South Dakota (78.0%). Breaking down this overall pass rate into AP CSA and AP CSP gives the following:
| AP CSA | AP CSA | |
|---|---|---|
| National Average | 69.2% | 71.3% |
| Leading States | Montana (87.5%) | South Dakota (87.5%) |
| Wyoming (86.4%) | New Jersey (80.6%) | |
| Illinois (77.0%) | Connecticut (79.5%) |
The following scatterplot diagram plots AP CSA and AP CS Principles pass rates for each state. The correlation between these two pass rates is 0.385, indicating some – but not a strong – correlation between the two pass rates.
States in which one pass rate was significantly different from the other are as follows:
| District of Columbia | 19.8% |
| Wyoming | 14.0% |
| Idaho | 13.3% |
| South Dakota | 43.1% |
| Oklahoma | 30.2% |
| New Mexico | 24.3% |
Pass Rate Relative Strength Metrics
The chart at the top of this post from the College Board showed the different demographics of AP CS A and AP CS P. Previous studies have noted how much better under-represented populations have performed in CSP relative to CSA.
Previous posts in this series provided relative strength metrics comparing the overall AP CS pass rates of Women/Men, Black/White, and Hispanic-Latinx/White. The following table breaks these relative strength metrics by AP CSA and AP CSP and shows that underrepresented groups pass the AP CSP exam at a much greater rate than with AP CSA:
| Metric | Overall | AP CSA | AP CSP |
|---|---|---|---|
| Women/Men AP Pass Rate Relative Strength |
103.1% | 97.1% | 105.7% |
| Black/White AP Pass Rate Relative Strength |
64.1% | 56.1% | 66.0% |
| Hispanic-Latinx/White AP Pass Rate Relative Strength |
77.3% | 67.5% | 80.2% |
The “P-Ratio”
Which states have taken the greatest advantage of the launch of AP Computer Science Principles to attract students from a broader demographic to AP CS? The following histogram shows the “P Ratio” – the percentage of students taking AP CSP divided by the number of students taking either AP CSA or AP CSP for each state:
The national P Ratio of P / (A + P) is 63.4%., and the states with the highest and lowest values are as follows
| Mississippi | 91.7% |
| Alabama | 85.9% |
| Nevada | 85.3% |
| Washington | 36.6% |
| Virginia | 40.1% |
| Vermont | 41.4% |
It is relatively easy for states with relatively small percentages of AP CS A students to have high P Ratios. The following scatterplot charts the P Ratio against the total percentage of all students taking any AP CS exam metric seen in the previous post. The states in the top right quadrant have both high overall AP CS rates and high P Ratios:
The following states have overall percentage of students taking AP CS >= 1.2% and a P ratio >= 65%.
| State | % Students Taking AP CS | P Ratio |
|---|---|---|
| Maryland | 2.8% | 71.8% |
| Rhode Island | 1.7% | 73.9% |
| Florida | 1.5% | 78.9% |
| Nevada | 1.2% | 85.3% |
| Hawaii | 1.3% | 79.7% |
| Georgia | 1.2% | 66.1% |