The Challenges of Large Scale Analysis

The data today might not be as comparable to, depending on how much you go back with data that originates from say manual coding. We do the large scale analysis that try to look at gender imbalance and there are a few things we pay attention to. One is that we don't scrape the data because typically the sample that you would scrape has gone through so many layers of filtering. You will not learn a great deal in general about how different genders behave on a certain platform. And then the last thing is around evaluation and that's something we already touched on. When we try to benchmark the results, we try to find correct comparison sets. Is it better or worse than it was last year