More snow fell this morning, which makes me want to share the pain. I'm a giver like that. However, I'll keep this really short just in case your head explodes from memories of high school (or college) statistics class.
Contingency tables are the fancy name for what many corporate folks put in their business presentations when trying to prove correlations and relationships through data. Most of the folks have no idea the why or wherefore of the input data; that's the domain of the research monkey in section 124G of Cubicalville. Typical practice is to look at the table and say, "Ooo, maffs! Numbers. Let's play!" Office twonks cut and paste the Contingency Table, add a few summary bullet points to support their case -- they may or may not be proven by the table -- then feed the report up the food chain. Each layer of management shaves off a chunk of information and repurposes the bullet points until all that is left is the pretty picture. When it reaches the executive offices, that chart becomes and addendum to a bullet point that has nothing to do with the initial data.
(Fortunately, most executives never ask for the Chi Square. Now, politicians, lobbyists, and academic might might ask for the independence test.)
Head 'splode yet?
What does this have to do with writing or publishing? If you've been following the latest salvos in the Self-Pub versus Trad Pub War, then the answer is everything. A lot of raw data was recently released. A lot of analysis was done by both sides. Both sides reenacted a typical corporate reporting process and their loudest voices claimed victories. How is that possible when there is data? Statistics don't lie. Any researcher, analyst, or business wonk will tell you:
It's easy to manipulate the data. It's easier to manipulate the audience.
Contingency tables show correlations and relationships. Before you accept data as a weapon, think about those relationships.