[EL] Limits of scientific method for policy decisions (was: Incidents of fraud ID ...)

Wed Aug 6 18:42:06 PDT 2014

Jon:

You respond as if both sides in the debate are equal in their approach to
evidence.  This is not the case.  Repeatedly those who believe that wide
spread fraud exist have have had their evidence debunked over time.  The
line of debate, at least the way I approach it, is between those who assert
that fraud exists despite what the evidence says versus those (when I argue
it at least) who make it an empirical claim to test.  Yes there are limits
to scientific knowledge and you wax well about it in your e-mail but all
that you argue is mostly irrelevant to what is supposed to be an empirical
debate.  My point is that arguing with those who believe in voter fraud is
like arguing with those who believe a second shooter existed in Dallas in
1963.  No matter what you argue you cannot win.  And that is the point--in
the end claims about voter fraud are not empirical, they are either
political or simply based on belief.  They are powerful tools  to move
people.  Finally, I just disagree with your assertions about the use of
evidence in the making of public policy.  I have written an entire book on
this topic that you should read after you read my book on election law.
Perhaps if we had more evidence-based policy the government would do fewer
dumber things like enact voter id laws, subsidize professional sports
teams, or enact laws that restrict immigration or public assistance on the
belief that immigrants and welfare recipients are a financial drain on the
economy.  Too much of our policy making is based on fear, ignorance, and
prejudice.

On Wed, Aug 6, 2014 at 6:34 PM, Jon Roland <jon.roland at constitution.org>
wrote:

>  As powerful as scientific method <http://pynthan.org/> is, for
> situations that lend themselves to it, it is important to understand its
> limitations, particularly for making policy decisions that do not always
> allow for the level of certainty that scientific method can provide, but
> have to be made anyway, on perhaps flimsy evidence, if the evidence on all
> sides of an issue is flimsy.
>
> The old saw is, "What is the rate of occurrence of undetectable crimes?"
>
> Game/decision theory can provide some insight. A confidence function
> applied to a proposition assigns a value on the unit interval, [0,1], where
> 0 is falsity and 1 is certainty. Those extreme values are only for
> propositions in mathematics or logic. Empirical model propositions get
> levels of confidence in between.
>
> A policy choice is usually cast in terms of which of several alternatives
> have the highest level of confidence, but it is possible for every
> alternative to have a confidence level of less than .5. The alternatives
> might even all have levels close to zero.
>
> The obvious question then is, if the confidence levels of every
> alternative are close to zero, why choose any of them? The answer is that
> there are other confidence measures that come into play.
>
> One of these is confidence in the estimates of the costs and benefits of
> deferring a decision, and of getting the choice wrong. If the estimate of
> the chance of an adverse outcome is low, but the cost of choosing the
> alternative is low, and the cost of such outcome is high, one may
> rationally make a prudent choice on weak evidence of the chances of the
> alternative happening. (Precautionary principle
> <http://en.wikipedia.org/wiki/Precautionary_principle>.)
>
> There is a separate assessment of when a decision is needed. It is often
> possible to have a high level of confidence that a decision is urgent, even
> if one cannot have a high level of confidence in any of the decision
> alternatives. This leads to a second dimension of analysis, in which
> confidence levels in the costs and benefits of alternative decisions,
> including not choosing any of the positive actions, are plotted against
> time, which may result in an assessment of the optimum decision time, which
> could be immediately, or it could be never.
>
> Systems analysis, combined with a knowledge of human nature, provides
> another approach. One such approach is opportunity analysis: If a system
> affords an opportunity to commit an offense ("defect"), there are players
> with motive, method, and opportunity to do so, at little risk and a high
> payoff, then one can reasonably presume that if it can happen, someone is
> doing it, even it the defections are undetectable. It may be worth the cost
> to close off some such opportunities ("temptations") even if no violations
> are detected.
>
> In the field of elections, we need to analyze the kinds of vote fraud that
> can be done (such as ballot access, registration, impersonation, box
> stuffing, counting, and reporting), and the costs/risks/benefits of each
> method that might be used. Clearly, the payoffs are very high, so one can
> expect the temptation to cheat will be irresistible to some. That leads to
> comparing the relative costs/risks/benefits of each opportunity and
> exploitation method. There is no strong incentive for fraudsters to risk
> impersonation fraud if they have rigged the counting.
>
> My own admittedly informal analysis is that the cost and risk of
> impersonation fraud is too high for the payoff of getting a few more votes.
> That leaves registration and counting. Clearly, voter ID cards will not
> prevent registration fraud, unless people try to vote multiple times at the
> same polling places with different ID cards and some poll worker recognizes
> them. That means such impersonators would have to scatter their illegal
> votes at multiple polling places and that adds to the cost for the small
> payoff.
>
> That leaves box stuffing and  counting. If measures are taken to prevent
> the first, then it is counting where the system vulnerability lies, and
> where efforts for reform need to focus.
>
> On another note, it is unconstitutional to require government-issued ID as
> proof of eligibility, because that that adds a qualification not in any
> constitution. It is reasonable to require proof of citizenship and
> residence, but government-issued ID gives anonymous bureaucrats the power
> to decide who gets to vote, and that amounts to an opportunity for
> registration fraud.
>
> -- Jon
>
> ----------------------------------------------------------
> Constitution Society               http://constitution.org
> 13359 N Hwy 183 #406-144               twitter.com/lex_rex
> Austin, TX 78750 512/299-5001  jon.roland at constitution.org
> ----------------------------------------------------------
>
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election
>

-- 
David Schultz, Professor
Editor, Journal of Public Affairs Education (JPAE)
Hamline University
Department of Political Science
1536 Hewitt Ave
MS B 1805
St. Paul, Minnesota 55104
651.523.2858 (voice)
651.523.3170 (fax)
http://davidschultz.efoliomn.com/
http://works.bepress.com/david_schultz/
http://schultzstake.blogspot.com/
Twitter:  @ProfDSchultz
My latest book:  Election Law and Democratic Theory, Ashgate Publishing
http://www.ashgate.com/isbn/9780754675433
FacultyRow SuperProfessor, 2012, 2013
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20140806/ad23f3b1/attachment.html>