5 things you should ask about any graph or figure

1) Do the data come from questions asked of people, a records database, or observation

Note: some "findings" reported in the media are judgements of committees

2) What is the actual statistic(s) reported?  And how are they displayed (e.g., over time, grouped). In other words, what are the x and why axis?

3) Was there an experimental manipulation? Or is this just a presentation of "observational" or "correlational" data? If the latter what other things could correlate with the trend or difference that the author is trying to assert?

4) Does the finding match with other findings you know. If it doesn't, that doesn't mean it's wrong.  How solid and thorough was your previous knowledge anyway?

5) Is the change or difference "statistically significant" or outside of what would be expected by chance (through some accepted metric of variability quantification)? Most popular media reports will not address this. Every scientific report should.

A new blog for a new purpose

I take a lot of notes when I read statistical work. They usually just stay in my own notebooks or I share them with the people with whom I'm working. At the same time, I know I've benefitted from blog posts, discussion board posts, and stats how-to websites. So when I spend a lot of time researching a topic I feel some urge to share what I've learned or found. Plus, I want to keep my work organized and in a place that's easy to share with others.

Setting up a "website" (whatever that means anymore) is one option, but I've found the blog format easier to maintain and update.

The posts here probably won't often be catchy "5 things you should...", but sometimes they will (that's often how I learn and synthesize statistical material). But I'm keeping this blog more as notes for myself and colleagues than statements about the field (although I'm sure some of those will slip in).

Please comment and share as you see fit.