Data analytics are everywhere, informing business and education practices, helping target marketing campaigns, and even estimating what type of social media content we are likely to engage with. They are so ubiquitous that we don;t question their validity or correctness; they are the hard quantitative answers to seemingly qualitative problems. Data analytics has this reputation for good reason, since it is most often a helpful and fairly accurate way to inform decisions.
But is it always so?
The article by Watson, et al. (2016) entitled "Small data, online learning and assessment practices in higher education: a case study of failure?" begins to address this question. In this case study of a single student who struggled in an online master's-level class, they argue that sometimes looking at single cases ("small data") can teach us something that big data cannot (p. 1032). The subject of this case study is Jay, who failed the course despite that fact that according to the usual data analytics used to identify at-risk students, he engaged with the material in the same way as a successful students. For this reason, they argue, she would not have been identified as an at-risk student using the usual data processes.
This gets me thinking about my own institution, and how we go about trying to identify at-risk students. We are a community college, and we struggle with retention, especially between the first and second semester We have a large at-risk population, with many students who are first generation, single parents, working full time, homeless, etc. We have data analytics that we use to help identify those students and connect them with services that can benefit them. That is a good thing, and it does help.
However, how many Jays are we missing? And how do we begin to set up some way of identifying them before it's too late?
References:
Watson, C., Wilson, A., Drew, V., & Thompson, T. L. (2016). Small data, online learning and assessment practices in higher education: a case study of failure? Assessment & Evaluation in Higher Education, 42(7), 1030-1045. doi:10.1080/02602938.2016.1223834
That is a really interesting question! In large classes, it would be really hard to notice some of the signs.
ReplyDeleteReading this post kind of made me think about qualitative vs. quantitative studies. In qual, we usually use fewer participants(smaller data) to get a deeper view (big data maybe?) but in quant, we use more participants (big data) to get a more general high level representation (smaller data maybe?).
ReplyDelete