Just Add Social Noor on social media

This semester, I decided to take a statistics class at the UC Berkeley extension in Redwood City.  Stats has never been one of my strengths so I decided to take the class to bolster my skills and refresh my memory.  For my class project, I decided to conduct a survey of engagement with social media sites, focusing on Facebook, flickr, and del.icio.us.  My main question was whether there is a relationship between a user’s age and their engagement with a social network site.  To assess engagement, I looked at two metrics - size of social network and size of digital artifacts collection. I defined digital artifacts for each site as Facebook wall posts, flickr photos, and del.icio.us bookmarks.  In my sample, I found a weak negative association between age and level of engagement, meaning that younger users tended to be more engaged.  Nonetheless, this relationship was fairly weak.  These results seem counterintuitive to me - I hypothesized a strong negative association.  I believe that these results are due to my sampling procedures, which were not random at all and most participants tended to be around my own age.

Out of curiosity (and because the Excel CORREL function is super easy to use), I started correlating all sorts of various metrics, hoping I can find a strong association somewhere in my data!  I found a strong positive association with the size of a del.ico.us user’s network and the size of her bookmarks collection.  This means that as a user’s social network increased in size, so did their bookmarks collection.  These results are interesting but not too surprising given the nature of del.icio.us.  del.icio.us users with larger social networks are exposed to more bookmarks, merely because they know more users who are posting more bookmarks and those bookmarks are winding up in their Network page.  And once those bookmarks end up on their Network page, they can easily save those bookmarks themselves.

For those who may be interested, my project slides are now available online.

For whatever reason, my fervor in blogging SXSW did not carry over to ICWSM, even though I enjoyed ICWSM far more than SXSW.  I’ve considered going back through my notes from the conference and posting my impressions of various talks but that seems a bit futile now that it has been almost a month since the conclusion of the conference.

At any rate, I did want to post some of my impressions regarding the conference, especially compared to last year’s.  As much as I enjoyed last year’s conference, this year’s conference was considerably better!  Compared to last year’s selection, I noticed a significant improvement in the quality of papers presented at this year’s conference.  This noticeable improvement may have been due to this year’s 26% acceptance rate (I’m not sure about last year’s acceptance rate).

Of all the sessions I attended, the set of papers in the psychology session were my favorite.  Two papers come to mind - What Elements of an Online Social Networking Profile Predict Target-Rater Agreement in Personality? and Thin Slices of Online Profile Attributes.  In the first paper,  David Evans, Sam Gosling, and Anthony Carroll used a Facebook application to determine if users’ impressions of their own personality matched how their friends perceive them.  They then built a social network site and conducted the same test but with random users, where users made assessments about others’ personalities via their profile information.  In both instances, they found that users tended to understand one another (or put more appropriately - the way that users perceive themselves tended to coincide with how others perceived them). In the second paper, Kristin Stecher and Scott Counts looked at various aspects of online profiles and found that users can generally learn about other users with a minimal amount of profile information (”think slices”).  They also considered the utility of various online profile attributes.  Not surprisingly, profile photos bubbled up to the top.

I’m looking forward to reading the four papers from this session.  Overall, ICWSM 2008 was a great conference and I’m glad I had an opportunity to attend and present.

This morning, I’ll be presenting my social TV paper at ICWSM.  For those who may be interested, I’ve posted the slides to Slideshare.  I really enjoyed working on this study and I’m looking forward to the talk.  Being that this is the last day of the conference and so early in the morning, I fear nobody will be there!  Hopefully, I’m proven wrong.