Teen birth rate, deviations from the trend….

Posted on September 19, 2009 by David Hume

Below in the comments I noted that a “quick & dirty” check of the data yielded an r-squared of 0.14 for the proportion of teen birth rate variance on the state level by the percentage of the state’s population that is black. That is, the black percentage as a variable can explain 14% of the variance of teen birth rate (assuming a linear model). For Non-Hispanic whites the r-squared was 0.18. These are modest values, but I thought readers might be curious as to which states lay above and below the trend line. Below is the scatterplot of teen birth rate vs. % Non-Hispanic white by state.whiteteenbirth

This entry was posted in data, Uncategorized and tagged . Bookmark the permalink.

5 Responses to Teen birth rate, deviations from the trend….

  1. nerdbound says:
    September 19, 2009 at 6:15 pm

    Actually, it almost seems like the best additional variable to add would be ‘latitude.’ The frozen North is having less sex?

  2. kurt9 says:
    September 21, 2009 at 10:25 am

    Do a further analysis by comparing teen birth rates with teen welfare rates. That should tell you the rate of true single motherhood vs the rate of young marriage or relationship.

  3. Chris says:
    September 22, 2009 at 1:27 pm

    @nerdbound: Try “poverty”. (Actually, I would have tried that *before* race, since they’re highly correlated in the US.) Urbanization might work too, although I wonder how much explanatory power it has independent of its correlation to poverty. Most of the high outliers (and most of the high teen birth rate states in general) are highly rural.

    Standard warning about the ecological fallacy applies, of course (in addition to the fact that this is a very weak relationship to begin with – but you can pretty much see that from the scatterplot).

  4. Jon says:
    September 22, 2009 at 8:21 pm

    Mississippi is the most religious state, while Vermont is the least. That pretty much says it all.

  5. Chris says:
    September 25, 2009 at 12:51 pm

    @Jon: Religion would probably show a correlation too, but I think it might be serving as a proxy for poverty. Poor people are more likely to be badly educated and desperate, both traits that increase religiousness. (Religion, in turn, resists social and technological change that could alleviate poverty, completing a vicious cycle.)

Comments are closed.