Carnegie Mellon University

Eberly Center

Teaching Excellence & Educational Innovation

Predicting Performance From Features Of An Instructional Design And Their Use

In the design of online instruction, decisions about which resources and activities to
include, how many of each to include, their order, and their cognitive level are made
repeatedly—and often implicitly—in the composition of the whole. Empirical work
provides clearer guidance on how best to design individual learning events than
synthesize an entire course. Educational data mining has tackled this problem by
assessing features of courses more broadly and determining their impact on
performance. The current study extends this work by defining fine-grained features of
the instructional design (ID) and analyzing how these features predict performance alone
and when combined with student-use data. Contributions of this study include defining
ID variables that correlate with learning outcomes and analyzing individual assessment
items in relation to how student use may moderate the predictive power of the ID
features of the materials and activities meant to teach these items.

Delahay, Anita
DC, Psychology
Lovett, Marsha
DC, Psychology/Eberly Center