Evaluators often find themselves in situations where resources to conduct thorough evaluations are limited. In this paper, we present a familiar instance where there is an overwhelming amount of open text to be analyzed under the constraints of time and personnel. In instances when timely feedback is important, the data are plentiful, and answers to the study questions carry lower consequences, we build a case for using a machine learning, in particular a sentiment analysis. We begin by explaining the rationale for the use of sentiment analysis and provide an introduction to this method. Next, we provide an example of a sentiment analysis leveraging data collected from a program evaluation of an engineering education intervention, specifically to text extracted from student reflections of course activities. Finally, limitations of sentiment analysis and related techniques are discussed as well as areas for future research.
Various coding initiatives and materials exist such as those on Code.org site to promote students’ computational thinking (CT). However, little is known as to (a) whether such materials, in fact, p...