A Characterization Study of Q&a Behavior on an Online Forum for Autism

Friday, May 16, 2014
Meeting Room A601 & A602 (Marriott Marquis Atlanta)
H. Hong1, G. D. Abowd1 and R. Arriaga2, (1)School of Interactive Computing, Georgia Institute of Technology, Atlanta, GA, (2)Georgia Institute of Technology, Atlanta, GA
A Characterization Study of Q&A Behavior on an Online Forum for Autism

Background:  Recent work has demonstrates the potential of using social media to help young adults with autism increase their independence by providing socially-acquired advice about their daily needs. While much is known about the general area of challenges faced by individuals with autism, little is known about what they are really concerned with on a daily basis and how they address such concerns in the form of question-asking or requests for help through social media. 

Objectives:  The goal of this study was to understand how individuals with autism seek advice on everyday life challenges through the use of an online forum, AspiesCentral (AC). We wanted to determine how well this forum served as a social Q&A system for these individuals, and how it might be improved. By considering what questions they ask, and how those questions are asked and answered, we seek design insight to improve the online experience. Specifically, we want to determine whether and how to improve the quality and the timeliness of answers to the kinds of everyday questions that individuals with autism have.

Methods:   we present a study of an online forum, AspiesCentral (AC), for adults that self-identify as being on the autism spectrum. Using a dataset of 28,960 posts organized into 1,945 threads, this paper explores in depth question-asking behaviors of individual users. Drawing on a combination of quantitative analysis and qualitative observations, we examine the posts we collected to better understand the pattern of question-asking behavior. Finally, we explore how the quality of responses to questions posed on the forum (i.e., rate and time) is affected by other properties of the question, such as the target responders and the perceived urgency of the response explicitly and implicitly indicated in a question.

Results:  We found that AC was used to receive advice and answers on a variety of topics. AC users asked question differently compared to what is known in other general population social Q&A systems. First, they tended to reveal more personal details in their questions. Second, they asked questions to simply confirm that others feel as they do or that others empathize with the issues raised. Despite the effectiveness of AC as a social Q&A system, it did demonstrate drawbacks: 1) unless specifically requested, responses for urgent matters were not addressed in a timely fashion; and 2) some topics requiring particular interests or expertise could not be sufficiently covered by existing AC members. 

Conclusions:   These drawbacks represent real design opportunities to either improve online forums like AC or to create other social networking solutions. Scaffolding of questions through a smart template or natural language critiquing system can prompt responders for quicker answers to urgent matters. This scaffolding can also better support questions that make clearer the intended form of response. In addition, mechanisms or incentives need to be in place that will encourage qualified crowdworkers from the general population to provide answers that represent more diverse and needed areas of expertise.