“E-Associations? Using Internet to Connect Citizens: the Case of Meetup.com” by Thomas Sander
Paper Prepared for American Political Science Association Conference, Sept. 2005
- 18-
• Demographic factors (respondent’s age, education level, marital status, ethnicity,
number of years lived in community).
• Meetup characteristics (size of Meetup observed, size of prior month’s Meetup,
type of Meetup, whether there was a Meetup leader, % of Meetup that was ages
18-30, % of Meetup that was over age 50, number of times the Meetup group had
previously met, type of Meetup)
• Respondent’s Meetup behavioral characteristics (how many personal friends
respondent had at first Meetup, how many new personal friends R met through the
Meetup, number of Meetups and other groups previously attended, whether R
recommended to others that they attend, how many public meetings on
town/school issues R attended in last year, whether R had gotten together with
new Meetup friends or acquaintances outside of Meetup)
[Insert Tables 1-1c here]
These results of an OLS multivariate analysis, controlling for all the items listed in the
above bullets are shown in Table 1. Table 1a shows the results of a standard OLS model
with robust standard errors on the variables significant from Table1, Table1b shows the
results of a 3-level random intercepts model that adjusts for the clustering of data at the
city level and Meetup level, and Table 1c shows the results of a 3-level random intercepts
model with ordered logit.
As can be seen in Tables 1b and 1c, the only variables that were statistically significant at
predicting "intent to return" (at p<=.05) were:
• Doing something with new friends/acquaintances outside of Meetup;
• Whether the Meetup group is a Hobby group; and
• Attending a greater number of previous Meetups on this topic and on other topics
(in the ordered logit, clustered model);
To some extent this tells a fairly pure social capital story, in which unless you develop
meaningful new social capital with group members (i.e. new friends or acquaintances that
you do something with outside of Meetup, rather than merely pre-existing friends),
you’re not as likely to return. The role of having been to more prior Meetups (on this
topic and other topics) could be something of a tautology. If you’ve been to more such
Meetups, you’ve presumably found value in the Meetup proposition. So it is little
surprise that the cumulative impact of all these individual decisions to return to the
Meetup group is driving the decision this time (that is after all what one would expect
unless the most recent Meetup was so bad as to swamp the long-term average of whether
the participant had felt the urge the return).
51
51
It is unclear why hobby groups were outliers. In the hobby groups, 8 out of the 37 groups we examined,
all had high averaged expressed likelihood to return and 3 of the 8 had 100% of respondents who said they
were “very likely” to return. This was much higher than the average expressed intention to return where
74% indicated they were very likely to return. (The hobby groups were 2 anime groups, 2 knitting groups, 1
boggle group, 1 chess group, 1 pinochle group, and 1 livejournal group.)
Convention