Thesis Defense: Iris Howley
THESIS DEFENSE Leveraging Educational Technology to Overcome Social Obstacles to Help Seeking Iris Howley COMMITTEE Carolyn Rosé (chair, HCII) Vincent Aleven (HCII) Bob Kraut (HCII) Marsha Lovett (Psychology) Stuart Karabenick (School of Education, University of Michigan) DOCUMENT http://www.irishowley.com/website/pubs/2015howley_dissertation.pdf ABSTRACT This dissertation provides initial empirical evidence for Expectancy Value Theory for Help Sources and generates design recommendations for online courses based on the newfound understanding between the theory and student behavior. My high-level research goals are pursued in the context of help seeking in the presence of reputation systems in MOOC discussion forums. Educational technology can be intentionally designed and introduced in such a way as to maintain the benefits of existing technology while reducing negative impact on learning-relevant behaviors. I do this through the lens of student expectancy and values for the help source, and costs of pursuing that help. Within this thesis I present three online survey experiments and two field experiments. The first survey supports the existence of beliefs for help sources, although careful design of value manipulations are necessary to isolate value beliefs from expectancy beliefs for the help source. The remaining two survey experiments are designed to further investigate the results of a system for help exchange through the lens of Expectancy Value Theory for Help Sources. The two field experiments experience progressively less controlled variables yet we still see support for our original hypotheses: evaluation anxiety inducing manipulations negatively impact student peer helper selection and other forum behaviors. From these experiments I generate a series of design recommendations for instructors of online courses implementing discussion forums: (1) reputation systems have an overall positive effect on student engagement in discussion forums, but there may be a negative effect on help seeking and other vulnerable learning-relevant behaviors, (2) The negative impact of up- and down-voting can be mitigated through the use of either help giver badges or using only upvoting instead of up- and down-voting, (3) When providing knowledge about a potential helper’s expertise, anything not considered an elevated amount of expertise will emphasize a potential helper’s lack of utility, and (4) Email prompts with dilute implementation have questionable impact on student contributions in discussion forums.