mininglabs

Data mining, analysis and visualization

Digging through Rapleaf’s study on gender and age in social networks

The social network users study by Rapleaf

Online reputation company Rapleaf recently published a study on gender and age in social network users.

The study was made using the public available data gathered from the social web on hundreds of millions of people. Although no real statistical estimation of the validity of the study can be computed, the number of scanned profiles are quite huge and the trend we see across different social networks seems to be consistent with one another. Analysing “people who are on at least one social network and in which there exists age information on these individuals”, they took into account 5.9 million users on facebook (to compare to the 90 million active users total on facebook) and 3.18 million users on myspace (supposed to have 110 million active users).

Scanning almost 50 million profiles, the study (featured on RWW) reveals a few interesting highlights along with some nice graphs.

First and foremost, the numbers confirms something we already know: in terms of demographics, the main difference between the various social networks is it’s users’s age. Another interesting point highlighted by the study is that women outnumber men on most social networks (with the notable exception of LinkedIn and Flickr).

Figure #1: Rapleaf’s figure: percentage of social network users across all ages

Rapleaf also published the raw data so we dug through the numbers to try to give a more graphical insight on the data.

Sex ratio

From the previous figure, it is clear that social networks can be divided into 3 categories:

• the vast majority of social networks exhibit the same pattern than myspace or facebook
• bebo or myyearbook shows a sex ratio bias whatever the age of the users. Those social networks seem to be mainly populated by women (myyearbook could even be seen as a female only network)
• linkedin and perfspot can be viewed as two “outliers” in this study, showing a sex ratio bias toward male whatever the age - except for teenagers on linkedin. At least linkedin is more business focused.

Looking now at the average profile of the sex ratio (as depicted in the above figure and computed by weighting each social network profile by the corresponding number of users) we can see also that there is a small sex ratio bias toward men for 30-40 year old demographic but this tendency is reversed just after 40 years old. It is interesting and also funny to note that this pattern is generally the same in dating networks, where there are more men looking for women for 30-40 year old demographic and vice versa after 40 years. Finally, note also the significant enrichment of men for 60-80 year old demographic.

Clustering of networks by demographic occupancy

Another interesting pattern to look at is the age frequency profiles. For a given social network, its age frequency profile is simply the distribution of the network users as a function of their age. We then used a simple k-means algorithm to explore how the networks clusterize together and how many groups of networks can be found.

As depicted in the previous figure, there are three main groups of social networks - based on their demographic occupancy. As expected the number of users is decreasing exponentially with age whatever the network. It is interesting to note that facebook user peak is shifted toward older users comparing to the myspace user peak. Reporting this result, Rapleaf was assuming that:

Our assumption is that once Myspace users graduate from high school, they switch over to using Facebook as their primary social network

Conclusion

As social networks grow up they collect an unprecedented among of information about it’s users. Using their user profile database, Rapleaf managed in this study to highlight a few interesting facts that we tried to illustrate by various visualizations.

This kind of in-depth analysis of social network users is not only interesting on his own but could also think that this might be used to evaluate the market value of social networks.

If you enjoyed this post, please consider to leave a comment or subscribe to the feed and get future articles delivered to your feed reader.

This work is licensed under a Creative Commons License permitting non-commercial sharing with attribution.