New article - Characterizing Whole Class Discussions About Data and Statistics - in JMB

2022/08/02

In May 2018, Seth Jones, a faculty member at Middle Tennessee State University, connected on Twitter over the topic of students in math and science classes working with data. I mentioned that we were moving to Tennessee, and we then connected over email. I had to search an old email account to find our correspondence:

I’d love to have the chance to meet up (virtually) - and hopefully F2F at state events. My research is focused on science education but I’m really interested in, well, exactly the kind of work you do with respect to “work with data.” I’m sure you have a ton going on and a ton of collaborators, but if we could find a time to talk for ~30 minutes, it would be great just to learn about what you’re up to. Maybe we could Zoom some time convenient for you in June or July? My schedule is pretty flexible so if any time works for you, just let me know and we can get it on the books.

Seth wrote back affirmatiely. He also wrote:

Iā€™m also just realizing that you are now my academic nephew since my advisor is your grandadvisor šŸ˜‚

Seth’s advisor, Richard Lehrer, was my advisor’s (Matthew Koehler) advisor.

We connected over shared interests and kicked off a project that combined data Seth collected on whole-class discussions in mathematics classes (about data and statistics) and the methodological approach and the package in R I developed to carry out LPA using R.

We met, worked a bit, met, worked, had life come up, met, worked, presented at a conference, and then eventually submitted a paper last fall. After a few cycles of revisions (major and minor), the paper was at last accepted - more than four years after the project started. The result was a unique paper that I am keen to build with research (including research with Seth!) around data science education.

Here’s the abstract:

Whole class discussions (WCDs) are an important pedagogical tool for mathematics classes but are challenging to characterize across large numbers of observations because of their dynamic and complex nature. In this paper, we report on an exploratory method to characterize WCDs in mathematics classes across large numbers of observations that we refer to as Conversation Profile Analysis (CPA). CPA uses Latent Class Modeling (LCM) with live observation data to generate profiles of WCDs in middle-grade mathematics classes. We report on our exploratory use of CPA to analyze observation data from 259 WCDs about data and statistics in middle school classes making use of an innovative approach to instruction called Data Modeling. We identified 4 profiles of WCDs and found that these profiles varied in likelihood across time and were associated with different ways students talked about key mathematical ideas. We also discuss broader implications of the CPA approach to studying WCDs in math classes.

The paper can be accessed for approximately the next 45 days via this link. We’ll try to add a post-print so it can be open;y accessed later. Thanks to Seth for this collaboration that has led to a friend - I think my first in Tennessee! The reference for the paper is here:

Jones, R. S., & Rosenberg, J. M. (2022). Characterizing whole class discussions about data and statistics with conversation profile analysis. The Journal of Mathematical Behavior, 67. https://www.sciencedirect.com/science/article/pii/S0732312322000645