I thought the lab on linear regression was interesting. It was a lot of work doing the algorithms "by hand" so it was a relief to find out there are simple Excel commands to do the job, as well as an easy way of creating a whole page of regression analysis statistics. I thought it would be fun to search the web and find statistics to create my own regression charts with facts slightly more interesting than shoe sizes, so here are a few.
First, I looked to see if how much a person weighs can possibly effect how they vote. I did so on a state-by-state basis comparing % of people considered obese with % of people who voted for Bush in 2004. Here is the chart I came up with..

This chart seems to show a correlation with living in a state with a high obesity rate and living in a state that supports Bush. This brought to mind the stereo-typical overweight businessman in a suit, possibly smoking a cigar, getting into a big shiny car with a "Support Bush" bumper sticker. So I decided to see if the correlation could be due to these highly successful educated folk who can afford a big steak now and again. Maybe these people drive up both the education levels and obesity levels of their states. So I next charted obesity levels with the percentage of people over age 25 with at least a bachelor's degree, per state.

As you can see the above chart disproves the educated large businessman theory as obesity also seems to be correlated with having a lower level of education. Perhaps uneducated people are only able to afford fast food, or are not aware of its effects. I can only speculate.. Maybe it would be interesting to conduct an exit poll of people leaving McDonalds to see who they voted for in the last election..

Another interesting thing I came across when making these charts was the effects of an anomaly. I originally in error left in Washington DC as a state in the 2nd chart. The level of education there is way higher than the highest state. This fact in combination with an average level of obesity flattened the trend line drastically for the entire chart, there almost appeared to be no correlation. This would have been a good example of how people in the media and government can play around with charts and stats to distort perceptions to their advantage. But unfortunately I deleted the chart and now have to get back to work on my Individual Installment also due today...

A final note, if anyone actually reads this post, and does not like the facts presented on the charts, remember you also have the tools learned in lab to create your own charts to discredit my charts. :))