Is Real Data Useful? (in pre-graduate education)

Kim Kastens, Senior Research Scientist, Lamont-Doherty Earth Observatory

(Please note that the graphs below present a hypothesis not observation. They are presented with the intent of stimulating thought and debate. Educational researchers are encouraged to compare this hypothesis to real educational data.)

Traditional earth science laboratories often include examination and manipulation of block diagrams, artists' conceptions of geological features, or data sets that have been simplified to clarify a specific concept or relationship. My intuition and experience suggest that, with such teaching aids, students' learning curves rise steeply at first, and then taper off through time as they extract all of the available insight from the simplified visual presentation (i.e. the blue curve below).

One of the commonly-stated claims to fame for computer networks in an educational context is that they will permit the delivery of genuine and complex data sets to large numbers of students. Examples are often taken from earth and planetary science. The unstated premise is that bringing data to the masses is good for education. My experience in teaching labs with real data is that, at first students flounder trying to understand the conventions of the data display, and trying to sort out the confounding complexities from the profound underlying patterns (red curve below, left end).

If they persist long enough, however, they do begin to learn and eventually their level of understanding surpasses that which could be achieved from the simplified display (red curve above, right end). In other words, the substitution of real data sets for the traditional simplified schematic representations could result in either a net gain or a net loss in educational effectiveness, depending on the time invested on the concept under study. It also seems likely to me that introducing real data would affect different students' learning unevenly. Some students, those with steeply-rising learning curves, will certainly learn far more using real data than they ever would have learned using an artificial representation designed especially for pedagogical purposes (right side of graph). But it seems plausible to me that other students would actually learn less, lost in the myriad of detail and complexity of the real data set (left side of graph).

If we believe these pictures, either based on our own classroom experience or through confirmation by educational researchers, a few strategies present themselves. One is to pick your battles carefully, in other words introduce real data sets only in cases where the data is so fundamental and the potential insights so profound that sufficient time can be allowed for the investigation to be pursued past the cross-over point in most students' individual learning curves. Another strategy is find ways to shorten the flat section of the real data learning curve and thus reach the rapidly-rising part of the learning curve sooner (i.e. move from the red curve to the green curve below). Tools that make the data easy to manipulate and visualize should have thiseffect by reducing the time required to learn to understand the data conventions. But I think the initial segment of the real-data learning curve will still be relatively flat because of the inherent complexity ofthe earth.

ILT Conference Series 6 | Kim Kastens | LDEO | graphs

Is Real Data Useful? (in pre-graduate education)

Presented at ILT CONFERENCE SERIES 24 February 1995

Presented at
ILT CONFERENCE SERIES
24 February 1995