NUMB3RS meets the rhetoric

A statistical analysis by Stanford and George Mason professors was recently published by the Society for Risk Analysis. The subject was illegal aliens and enforcement efforts.

The formulae look something like this sample:Their findings aren’t very promising.
1) There aren’t nearly enough detention beds.
2) If you don’t detain them, the likelihood they will actually leave the country is only 13%. (Now you know why Obama’s aunt and nearly 600,000 other fugitive absconders are running around the country.)
3) Expedited removal procedures have shortened the detention time to an average of 19 days before deportation, which greatly increases the probability of deportation.
4) Increasing the number of Border Patrol agents alone will not be effective unless more detention beds are also provided.
5) Not detaining non-violent illegal aliens “until their fourth apprehension is nearly equivalent to not detaining them at all.”
6) Given that Mexican workers can earn 450% more here in the United States, worksite raids and employer fines will have to be increased significantly to offset the value of potential earnings.
7) The probability that a non-Mexican terrorist can successfully enter the United States through our Southern border is 97.3%

The bottom line is that more border agents, more miles of fence, more virtual fence (assuming the manpower is provided for apprehension), and more worksite enforcement really do work. But we are so far behind the curve in these areas that it will take time and money to reverse the trend.

The entire study is, of course, a chalkboard exercise. The researchers themselves tell us, Immigration is a difficult issue to mathematically model: even restricting our attention to the small part of immigration that affects homeland security leads to an unwieldy system of equations and a formidable parameter estimation task. The level of detail in our submodels is dictated by the level of detail in the available data.…and …
Consequently, the model’s numerical output is not intended—and indeed is unable—to capture the quantitative impact of various decisions with any degree of accuracy, and so the model is incapable of directly guiding policy, except in a very crude manner. Rather, this study—by framing the immigration/ homeland security problem in a way that captures most of its salient features—is meant primarily as a vehicle for rational dialog about a complex problem that often elicits strong emotional responses.

Thank you once again, Washington.

The study can be found here:
http://www.stanford.edu/~amotskin/index_files/USMBorder.pdf

Update 2/3/2013.  The link above is broken, but the study can still be viewed here:
http://graphics.stanford.edu/projects/lgl/papers/wlm-ahsumb-09/wlm-ahsumb-09.pdf