saying that the optical voting machines in Florida did not give irregular results:
In the aftermath of the 2004 Presidential election, a significant number of activists and
researchers have argued that the optical voting machines that are used in a majority of
Florida counties caused John Kerry to receive fewer votes than “Direct Recording Electronic”
(DRE) voting machines. Implicitly, these researchers wish to estimate the causal
effect of using optical versus DRE voting machines. There are two difficulties which must
be confronted when one attempts to estimate this causal effect. First, the type of voting
machine that a county uses was not randomly assigned. Second, the 67 counties in Florida
are extremely heterogeneous. For example, the portion of the population which is white
(including Hispanic whites) ranges from 0.41 to 0.96,1 and the proportion of the population
which is registered Democrat ranges from 0.24 to 0.89.2
Both of these issues are common when one tries to make causal inferences from observational
data. The issues give rise to the problem of confounding—i.e., the distributions
of pre-treatment variables, such as party registration, past votes and demographics, differ
greatly between those counties that use optical voting machines and the counties that use
DREs. These baseline differences must be accounted for before any valid causal inference
can be made. A common way to account for these variables is by comparing matched observations.
Matching estimators do not make functional form assumptions and are hence
appealing.3 For the optical voting machine question, a matching estimator greatly reduces
the imbalances in baseline variables and produces the most reliable statistical results we
currently have. The estimated causal effect of optical voting machines on the Kerry vote
is indistinguishable from zero. These results give no support to the conjecture that optical
voting machines resulted in fewer Kerry votes than the DREs would have.
http://jsekhon.fas.harvard.edu/papers/SekhonOpticalMatch.pdfI don't know about all of the fine points of their arguments, matching, and so on, but I do know that using a group of academics who have worked together on a subject in the past is no way to get an unbiased evaluation of a problem. If nothing else, they have an investment in the point of view of their research.
http://elections.fas.harvard.edu/index.htmlIf fairness and an impartial analysis of results was truly the intention of this committee, why didn't they invite researchers with differing views?