Re:Why is the American dream dead in the south?
by magicalist on Sunday January 26, 2014 @05:58PM
(#7127455)
That's a really really bad map. Bucketing to "lower than", "same as", and "higher than US rate"?
Even accepting their statistical significance qualification, you could have a .1% difference in the percentage of single-parent families between two states, and as long as you have enough samples, your result could be statistically significant and so bucketed with different colors, implying a much greater disparity since you would also bucket two states with a 20% difference in the same way.
And you can see similar issues with their bucketing (why cut off at those particular numbers?), but far more importantly they provide the actual percentage per state. And here we can see (to take the submission's comparison) that what makes California and North Carolina different when it comes to single-parent households is only 4% (33% vs 37%).
Now there may be more to this when we look at the numbers for smaller regions, but at least at the state level that your map showed, there are significant confounding factors that single-parent status would not explain.
Even accepting their statistical significance qualification, you could have a .1% difference in the percentage of single-parent families between two states, and as long as you have enough samples, your result could be statistically significant and so bucketed with different colors, implying a much greater disparity since you would also bucket two states with a 20% difference in the same way.
Here's a different map:
http://www.thenationalcampaign.org/state-data/state-comparis...
And you can see similar issues with their bucketing (why cut off at those particular numbers?), but far more importantly they provide the actual percentage per state. And here we can see (to take the submission's comparison) that what makes California and North Carolina different when it comes to single-parent households is only 4% (33% vs 37%).
Now there may be more to this when we look at the numbers for smaller regions, but at least at the state level that your map showed, there are significant confounding factors that single-parent status would not explain.