One of the most interesting statistical tests is Correlations. It compares every possible pre-review category (such as Area, Tag, or Idea Group) with the overall score from H2H. 

In the example below, all possible category-value subsets were tested against the full set of ideas and three of the 40 were significantly different from the whole. Ideas with “Idea Group = Selling Assets – Capital” were better than overall, while ideas with “Area = Cost Avoidance or Sharing” were definitely worse :

The test used is the Kolmogorov-Smirnov test and the words “probably” and “certainly” correspond to proper statistical thresholds like p < 0.05 or p < 0.001.  
Released in October 2015, we anticipate this analysis will support an approach to refining challenges over time with data that supports focus within specific areas.