Personality, Party, or Era?

The world is full of interesting small data sets these days. UC Santa Barbara has historic data about the length of State of the Union addresses since 1966. This article investigates a small regression built from that and other sources.

The UC Santa Barbara data set was augmented with party data.

With DataSplash we can easily upload the newly made CSV to quantify and visualize any relationships it might contain.

It turns out that speeches have been growing in length over time and it appears that GOP party members are decidedly less chatty. On visual inspection, it looks like President Clinton could be a big outlier. If we remove his data does the relationship reverse or go away?

If we left-click and drag over the data between the years from 1994–2000, DataSplash presents us with the option to delete President Clinton data.

The strength of the relationship seems to change a bit, but it’s still there. The Regression Results tab makes these findings clearer because 0 is outside the illustrated confidence interval for both variables.

DataSplash is great for quick analysis and also for recreational data exploration. So what happens when we add cats?

As expected, the cat effect dominates all other effects and entirely reverses the previous trend. In the new model, the cat ownership effect is the sole significant variable.

Pet data courtesy of the Pet Food Manufacturers’ Association. Download this article’s finalized data set here.

In 1980, as Carter was leaving office, America had a healthy ratio of dog to cat ownership, but this data makes a little-known trend clear.

The fact is, in 2001 there were as many cat-owning households as dog-owning households. It is not at all clear what this rise of cats will entail for the future of our country.