Data Analyst
Jacob Fricke
I help organizations run better experiments and make clearer decisions. Currently building analytics infrastructure at Campact and studying for an MS in Analytics at Georgia Tech.
Recent Writing
-
26 APR
Why You Can't Trust 'We Tried It'
Comparing a new approach to the old version doesn't tell you what you think it does. Randomized controlled trials are the only way to be sure the change is what caused the difference.
-
24 MAR
Is the Pattern Real?
Your cross-tabulation shows different results across groups. The chi-squared test tells you whether those differences reflect a real pattern or just random noise in who ended up where.
-
23 MAR
The More You Look, the More You'll Find
Run enough tests on the same data and something will look significant by pure chance. Multiple comparison corrections keep you from fooling yourself when you test many things at once.