data

You are currently browsing articles tagged data.

Last August, I wrote about the differing outcomes of the two Intensive Elementary French sections I taught: one remote synchronous, the other remote asynchronous. My write-up included some descriptive statistics, and ultimately concluded that learners in the synchronous course achieved higher scores than those in the asynchronous course. At the time I wrote the post, my statistical literacy was less than it is now, so the purpose of the present post is to determine whether these findings were statistically significant.

I. Do final course grades differ between the two modalities?

I took the two sections, asynchronous (n = 19) and synchronous (n = 10), and conducted a two-sample t-test to determine whether there is a meaningful, measurable difference in the average final course grade for each modality. The data from each section were normally distributed according to visual testing and the Shapiro-Wilk test (asynchronous p = 0.12, synchronous p = 0.26). An F-test revealed that the data were not of equal variance (F = 13.66, p < 0.001, 95% CI [3.69, 40.01]). The Welch two-sample t-test revealed a statistically significant, large difference between the asynchronous and synchronous sections (t(22.49) = 3.87, p < 0.001, d = 1.29, 95% CI [-2.04, -0.53]). The observed difference in means was -26.46, with a 95% CI [-40.63, -12.29]. A post-hoc test revealed an actual power of 0.80.

Boxplot comparison of the two groups' final course grades. Left boxplot tracks the asynchronous scores, right boxplot tracks the synchronous scores.
Plotted using ggplot2 with R.

As you can see from the visualization, the learners in the asynchronous course had a much wider spread of final grades, compared to the learners in the synchronous course. The huge, statistically significant difference between the modalities can be seen most glaringly: nearly 75% of the synchronous scores fall within just the fourth quartile of asynchronous scores.

Read the rest of this entry »

Tags: , ,

This past Summer, I taught two sections of the same French course during the same six-week session at a large, public R1 university. The main difference between these two sections was the mode of distribution: one was delivered asynchronously, while the other section met synchronously over Zoom for 80 minutes twice per week. I wanted to record some of my observations, in an attempt to explore some of the different student outcomes. Note that this is an informal comparison, and no personally-identifiable data are being shared.

I’ve broken down this post into several sections:

  • Course description and distribution model
  • Modifications for online distribution
  • Student population
  • Unexpected roadblock
  • Interaction and engagement
  • Grades
  • Discussion
  • Conclusions
Read the rest of this entry »

Tags: , ,

https://www.xkcd.com/327/

Turns out that Learning Management Systems aren’t as secure as they want you to believe–SQL injection vulnerability is a pretty grievous, novice error to make for company like Blackboard.

The US Department of Education has several online resources regarding the collection of students’ personally identifiable information, and even discusses how online education platforms might collect usage metadata from all of our students–but this doesn’t violate FERPA as long as any shared metadata is not directly linked to identifiable information. So there shouldn’t be any issue with that, unless tech companies start using these data and metadata to create a profile of you, even if you don’t have an account with them.

It’s not a stretch to say that a company like Google could take all of these data and metadata from schools’ Google Apps for Education accounts and match them to personal accounts of people whose personal data match. As long as Google keeps the data for themselves (because only disclosing it would violate FERPA), there are no legal protections for any of us, especially our children, from a company that decides to use these metadata to create psychographic profiles for targeted advertising.

“Oh, so Google can show my A-student ads for colleges, and the C-student can get ads for tutoring? What’s the big deal?” The big deal is that an unscrupulous advertising firm might target students based on psychological traits rather than grades alone (which would be Huxleyan enough). Students whose schoolwork shows a lack of critical thinking, or a reactionary mindset–students who are quick to jump to conclusions just by seeing a headline without reading the whole article (or who don’t bother to read the instructions)–they might be susceptible to the kinds of propagandistic voting campaigns that a company like Cambridge Analytica boasted about.

This doesn’t even touch special education plans, disciplinary records, medical records, or even family information (some of my students’ files have had notes about their parents’ divorce arrangements). We need to start taking students’ data more seriously, otherwise they’ll have to worry about their “permanent records” for the rest of their lives.

Tags: , , , , , , , , ,