Introduction

How valuable is Zybooks?

The question stands to reason, how well does a class like zybooks truly help us students learn and how well does it teach for a class of beginner computer scientists?

We can visualize this through statistical analysis of our data. We chose a sample size of one simple randomly chosen class, in this case CS010

That being said, we can go ahead and separate each data into meaningful statistics.

Our mission statement is to see that students are getting the optimal learning environment and usage of these statistics to help the creators and teachers who have taken their time to make this material possible.

Credit

The student’s performance is an important statistic in an optimal learning environment. As the students devotes more time into their learning, they should start to pick up on the subject material and attain a higher credit value score as they put work into the resource.

A note:~from this data we can see that some students spent significantly more time doing zybooks submissions. However, the credit they recieved appears to vary on quite a large scale. The zybooks in this particular class, CS10, is mandatory and is considered a homework assignment. In the case of this graph, the average score is around 275 based on the graph on the lower left. However, we can see that the total possible questions accumulate to 447.

Looking at the data by resource type, we can see that the custom questions has taken students the longest amount of time, as it makes up 201 of the 447 total questions. Custom questions have the most practical value to students and professors due to their coding questions that cannot be guessed. These questions have the most value to us since they demonstrate the students' understanding at a higher level.

Taking the count of the content resource id done by student basically returns our credit recieved per student. Pairing that with the time statistic, we can again gain an accurate representation on how the students is doing in the class. The benefit of the flat graph gives us the ability to see where the focus point, or the average is of all of the scores. In this case, we can see the mean is approximately 250 credits out of 447, which is a score of 55% total.

Students Completed

Each activity in zybooks has its own distinct ID number. In this figure, we have plotted average students completion of each activity distinctly to see how many of the students are completing each activity. We can see that the average completion in a class of 105 is about 90, so this is about an 85% participation rate on the majority of the questions. Note that some questions are required and some are mandatory for students.

Submissions and Resource Type

As we saw previously in the credit, the custom answer questions make up a large proportion. These questions require quite a bit of coding, so we can see here that they make up a fair amount of the time that students spend on the Zybooks platform.

Another point to note is that the proportion of the custom questions make up a larger number of the submissions in the platform. Which indicates that the instructors believe that CS010 in particular is more about the application of the coding rather than the theoretical questions in quantity.

ANOVA Significance Testing

ANOVA (analysis of variance) is a comparison of the means between the groups we are interested in i.e. custom, short answer and multiple choice. This test determines whether there is any statistically significant difference from one another. We can do this by implementing a null hypothesis Then we have an alternative hypothesis Ha where at least one mean is different Given our graphs, we can clearly see that our means are not the same

mean_sa = 500

mean_mc = 800

mean_custom = 1050

In this test, it would be seen as a low significance level. Our P value would likely be less than 0.05, which is considered statistically significant. In that case, we would have to reject the null hypothesis and that our alternative would be true. In this case, we are referring to attempts. We can clearly say there is a difference in attempts when it comes to short answer as compared to custom answer, as we can see in the clustering of the points when it comes to the comparison.

Similar to the previous ANOVA we used for attempts, this graph is comparing the means taken. However, the means seem to be very similar so we have an inherent representation that the means are satisfactory in terms of significance values in that they do not differ too much. For this reason the null hypothesis holds true.

Correlations

The way we were able to find the meaningful relationships in our data was through tried and tested correlations. Plotting anything X versus Y can show a meaningful chi-squared or just show plain regression. Some of our comparisons here are insignificant yet we chose to include them for the sake of showing the process.

We can see in the last row of graphs that the avg time taken clearly has a positive correlation. The reason I had not included it in the previous sections was because I had already proven that the custom attempts always took the most time. However, an important statistic that may have been glossed over is the lack of time spent on MC or SA questions. It leads us to believe that there is not enough time being spent on the theoretical portion of the learning resource.

These graphs above show regression again, however it is important to note that they are important to our study of relational models to find some conclusions. (Plotting X by Y involved most of the time spent in this project)

Scheduling

The last portion of data shows the number of questions answered per month in April, May, and June. We can see here that there were more submissions as the class started out, and as the class progressed less students saw the need to do the zybooks. This can be accounted for by many factors - perhaps the students were well on their way to learning the material over time or that the instructor simply did not assign as many tasks in April as opposed to in June.

About Us

Our team consists of Brennan Hall, Vanessa Le and Tarun Motwani.