Featured Photos


Baseball Hall of Fame - 8/23/11

Featured Video


Avery's QuEST Project - It's Healthy!

House Construction


The Completed Home Renovation


Home Renovation - Complete!


Our House Construction Photoblog

RSS Feed


« | Main | »

Playing with Numbers, Part 1

By Brian | September 28, 2006 | Share on Facebook

I’ve been playing around with trend data lately, and have discovered some interesting relationships that may provide a little context to current events. For instance, compare the historical price of oil to President Bush’s approval ratings throughout his presidency:


(chart links to a spreadsheet with the underlying data)

The correlation between the two datasets is -81.14%. In other words, over the last five and a half years, every time the price of oil has dropped $1, the President’s approval rating has increased by 0.8% (and vice versa). For the real statistics geeks, the R-Square value is 65.84%, suggesting that 65% of the movement in Bush’s approval ratings are explainable by the price of oil.

Now, I’m no statistician (two intro courses in statistics at The Wharton School were enough to sway me away forever), but I know enough to avoid confusing correlation with causality. This is not proof that the price of oil directly affects Bush’s approval ratings. It merely suggests that the historical data suggests a relationship. There is always the chance that this relationship is purely coincidental.

That said, (and based solely on my own opinion now, not the statistics), I think there is a causal relationship here. When the price of oil drops, the stock market tends to rise and the price of a gallon of gasoline tends to fall. These items have real world impact on most Americans and when they’re all moving in the right direction, I can easily imagine it affecting their attitude towards their president and their responses to polling questions.

What’s most surprising to me here is the extent of the correlation. If this data does reflect causality, then it means that each new opinion poll on the President’s job performance is not the stark referrendum on the Iraqi war, the erosion or our civil liberties, or our treatment of foreign detainees that one side or the other (depending on what the data says) claims it to be. Instead, around two thirds of any movement in the polls is tied to the price of a publicly traded commodity which is, at best, only indirectly controlled by the President himself.

Topics: Money Talk, Political Rantings | 4 Comments »

4 Responses to “Playing with Numbers, Part 1”

  1. Jeff Porten says at October 3rd, 2006 at 5:48 pm :
    [choking sound of Starbucks in spit-take]

    Dude, this is obvious. Oil prices are directly correlated to how badly things are going in the Middle East. Bush’s approval rating is inversely correlated to the same thing. It is fairly simple to demonstrate that these two correlations, in fact, are causation.

    Therefore, Bush’s approval rating *is* correlated to oil prices, that correlation is *not* causation, and his being viewed as one of the worst and inept adventurists in American history is his own damn fault.

  2. Brian says at October 4th, 2006 at 1:12 am :
    Oil prices are directly correlated to how badly things are going in the Middle East.

    Any source on this? Because I can think of a good dozen other things that oil prices depend on: growing demand in India/China, hurricane damage in the gulf coast, OPEC’s supply decisions, the Federal Reserve’s interest rate policy, Hugo Chavez’s mood, hot summers and/or cold winters, etc.

    Even before the war, Iraq produced less than 10% of the world’s oil. This is not to say the instability caused by the war hasn’t affected oil prices, but the data I’m looking at covers six years, and lots of other factors are involved.

  3. Jeff Porten says at October 4th, 2006 at 7:45 am :
    Sorry, the best source I can quote is “common knowledge”, which I realize is unsatisfying and frequently inaccurate. But note that it’s not just Iraq’s supply we’re talking about here — Iraq destablized the entire region.

    As for the age of your data — it wasn’t exactly a secret that the neocons were looking for reasons to invade Iraq back in January 2001. Certainly your graph doesn’t show much correlation until early 2002, when it became clear that Bush was going to get a blank check to invade downtown Montreal if he wanted to.

    I think my explanation wins the Occam’s Razor kewpie doll. If you want to say that the perception of Bush incompetence is based on bad luck, well, I think the ball’s in your court to make a stronger explanation. Mine is pretty well self-contained.

  4. Brian says at October 6th, 2006 at 1:45 pm :
    Well, no – I didn’t say it was based on bad luck. I said it was based (primarily) on the price of oil.

    Let me test your assumption a different way: last week, the price of oil dropped below $60/barrel for the first time in over a year AND Bush’s approval rating broke 40% for the first time in just about a year.

    Your explanation would suggest that things haven’t been this good in the Middle East in about a year. Agree or disagree?

Comments

Comments will be sent to the moderation queue.