Statistics Hell
I have entered statistics hell for the past 3 weeks. I’m working on a massive research project at work, and my brain is nearing critical mass. Soon, my head will explode and my brains will have to be scraped off the walls rather than being donated to science.
I work for a consulting company, and naturally we do lots of research to tie our materials in with client organization performance. If we can’t show that our consulting helps the bottom line, it’s pretty tough to sell. I’m in charge of the department that does the research, so that means when a really icky project comes through, I get to get my hands dirty on it.
Well, this one qualifies. A 250 point assessment with a ton of responses, needing to be internally validated as well as tied to other “tried and true” assessments and statistically correlated to our mounds of other data. Of course, none of the data is in compatible systems, so there’s a lot of manipulation to get it all to work.
The reason it’s not all in the same system, is that previously the research was done by what I like to refer to as the “seat of the pants” method, in which research was typically done for an individual client and was crunched by the consultants actually assigned to the project. That was destined to insure that every assessment was done completely differently.
Basically that means I’m tossing all of the previous summaries and recrunching everything from the raw data export, then combining things one at a time into SPSS.
Then, after running everything through SPSS, I get to export section by section into Excel (my Excel is souped up with after-market addins) where I do cascading summaries until finally it results in a 40ish page report.
Well, I’m about 10,000 statistical tests into the SPSS portion of the work. My eyes are bleeding from looking at SPSS output.
I know you don’t care about any of this… I’m cleansing my brain by writing so I can go back to it.
Pete
Monday, June 18th, 2007
