My Master Thesis of the MBA did not have much to do with business. I wanted to create new knowledge about Online Communities, and luckily the Institute for Technology and Innovation Management of the University accepted the subject I proposed: “User-Generated Content in Online Communities: The Effects of Group Size and Topic”. 

Among many other things that I won’t describe in this post, I wanted to find out if people contributed more to an Online Community (OC) if the total numer of participants was higher. By contributing I mean posting content (User-generated content, or UGC) in any way. In this post I’ll try to summarize what I found, avoiding scientific jargon whenever possible.

What Does the Theory Say

There are four theories that describe what happens with the level of individual contribution of the members of an OC depending on its size:

• Critical Mass Theory: It implies that there is a minimal number of users necessary for the sustainability of an OC. People will feel more encouraged to contribute to a community once it has enough users. In other words, this theory predicts a positive correlation between the size of an OC and the level of activity of its members.
• Information Overload Theory: Information overload occurs when the volume of information supply exceeds the capacity of an individual. When an OC reaches a certain number of members, the total amount of contributions may be too large for each member to assimilate, which can generate frustration and diminish the interest of the users, who may even leave the community.
• Social Loafing: It is defined as the tendency to reduce individual effort when working in groups compared to the individual effort expended when working alone. This implies that in larger online communities users will contribute less than what they would in a smaller one because they can benefit from other’s contributions. In other words, social loafing contributes to a negative relationship between group size and individual contribution level.
• Common Ground Theory: Common ground is “a combination of mutual knowledge, mutual beliefs and mutual assumptions”. Establishing a common ground facilitates the communication and thus increases it. Theoretically, it is more difficult to establish common ground in larger groups and therefore there should be a negative relationship between group size and amount of communication.

So, if we take all of them together, one would think that the level of contribution of the users in an OC starts climbing at the beginning due to the effect of the critical mass theory, but when the OC reaches a certain size, people no longer contribute too much because the other three theories kick in. That was the hypothesis of this part of my work.

The Experiment

To test the hypothesis, I manually grabbed data of 1.547 Facebook groups (yes, it was a pain in the ass, but luckily I was able to outsource most of the job to a friend of one of my sisters). For each group I got (among lots of other data relevant for the rest of the thesis) the number of members and the number of wall posts. With that information I could see if the number of wall post per user was related with the total amount of members of the group.

To be able to work with the data I did the following:

• First, I clustered the whole data set in 1.000 clusters, each containing groups with x to x + 1.392 members. For example, the first cluster contained all groups with 0 to 1.392 members, the second cluster contained all groups with 1.393 to 2.784 members and so on. Since the group sizes on my sample varied between 2 and 1.392.000 members, that cluster structure would provide exactly 1.000 clusters, which is why the number of 1.392 was used.
• Then I calculated, for each cluster, the average number of members per group and the average number of wall posts per member. These variables would allow me to see differences in the level of activity in the different clusters. Some clusters did not have any groups; in those cases, a linear interpolation between the previous and the next cluster was used.
• Only 17 groups were contained in the clusters 101 to 1.000 (clusters with more than 140.000 members), so they were not considered in the analysis for they would not have been representative.

The Results

I was able to confirm my hypothesis, as the charts below show:

For groups with up to 70.000 members, the level of individual contribution grows slowly with the total size of the group, but when a group gets too big, the effects of information overload, social loafing and/or common ground begin to appear, inducing a reduction in the level of participation of each user.

I think it’s interesting that the number of wall posts per member achieves a peak at a certain group size and then declines. These results obviously apply only to Facebook Groups (actually the 2009 version of Facebook Groups, which have changed a lot since then), but maybe each OC has its own “critical number”, a certain size that, when surpassed, is followed by the decline of the engagement level of its members.