Using GA4 measurements to characterize the use of biostatistics online support documentation

Abstract

An Online Support Documentation has been built, since 2019, for a course of Biostatistics at University of Aveiro. It has been done to better support student study by complementing traditional slide presentations and a book of exercises. This course targets more than 180 students from the first year of the degrees of Biology, Geology and Biology, and Marine Sciences. Last year, 28 students, out of 180, replied to an anonymous inquiry and a complementary interview was conducted with 5 chosen students. We could conclude that the development of the online support documentation should continue but improvements are needed (for instance, increasing the level of details in online texts). To get a complementary view on how the Biostatistics online support documentation is used by students, a new approach is carried out. In 2020/2021 academic year we have automatically collected data usage using "Google Analytics 4" (GA4). The website, holding the support documentation, is a tree-like structure with chapters and sections. Each section is a webpage that can be of two types: a "concept" section or an individual "exercise" section. GA4 collects for each webpage: (1) the webpage title; (2) the number of interactions; (3) the mean time spent in all interactions with a webpage. An interaction (also called "User engagement") is issued by GA4 when a user has the webpage in focus for at least one second. Manually, two more columns were added to data: (4) the type of the webpage: concept or exercise; (5) the chapter number (1 to 6). We present statistical insights about how the online support documentation was used in 2022/June when students had two formal tests about Chapters 4 and then Chapters 5 and 6. Briefly, our findings is summarized as follow: (1) as expected by the common sense, usage of webpages increased before assessments; (2) in Chapters 4 and 5, the "number of interactions", obtained for a webpage, and the "mean time spent", obtained for the same webpage, are positively correlated suggesting there are some pages more important in the learning process; (3) Concerning the mean time spent, its distribution is more skewed to the right for exercise webpages than for conceptual webpages. It could mean that the mean time spent on exercise webpages shows high variability because students look more for suggestions on the "how to do it" or to “confirm their solution” to a certain exercise.