Why use benchmarks?
By now you probably noticed that the size and color of icons are pre-adjusted or "benchmarked" against a reference data
point. By default the reference is "all other data we've ever collected" but you can customize that benchmark to be something specific and logical to your question.
Benchmarks are important! This is what raw data from a typical collection would look like without benchmarking:
First, everything is green. People are 10 to 20 times more likely to be positive in stories than negative, depending on how they were prompted.
Second, the results now emphasize the unevenness of the sample across age and gender. Like the positive story bias,
this uneven sample is nearly always present in data. Instead of throwing out all of the data, we display results
weighted against a large reference sample, so that any collection of randomly selected stories will look like this:
Every icon is yellow (neutral) and medium-sized, because they are a representative sample. If you asked 100 people to participate,
this would be the breakdown of who responds, based on asking over 60,000 people so far.
But if you wanted the benchmark to be 100% women, you can select that with "compare" tool
, explained later.
Using quotes to search for phrases
You searched for "peace maker".
Notice that putting quotes around two words narrows the number of stories that appear in your collection.
However, a more powerful and flexible way to search for stories is you put a bunch of words inside parentheses,
like (peace war soldier retribution reconciliation) instead of None.
The search engine will include stories that have any one of these words
inside the parentheses.
Try searching for
(peace war soldier retribution reconciliation)
by typing it into the search box as shown: