I read this post earlier today, and was 'inspired' by its eye-brow raising implications:

Election Fraud likely Benefited Clinton Over Sanders - By a Lot
by Steven D — 06/10/2016

Steven D summarizies the study results, with which I agree:

"[the authors] states where there was no paper voting trail (i.e., e-voting machines) and found Clinton did significantly better in those states than Sanders compared to states where a paper trail existed."

That should be a very concerning 'trend', if true.

Steven also concludes with:

"If you're a statistician, please feel free to examine their paper at greater length."

{Thanks SD, I was looking for a "challenge" today, LOL.}

Well, I'm not a "statistician" in real life, BUT I played one at University many years ago.

So here goes a 'brief dive' into into the p-values "significance" of that eye-brow raising study:

"Are we witnessing a dishonest election? A between state comparison based on the used voting procedures of the 2016 Democratic Party Primary for the Presidency of the United States of America," dated June 7, 2016

First a few boring definitions, hopefully made palatable:

The p-value is defined as the probability of obtaining a result equal to or "more extreme" than what was actually observed, when the null hypothesis is true.[4][5]

Null Hypothesis, Huh, what's that? (Think "status quo" vs the nuevo theory that will "upset the applecart" and your halfway there.)

In inferential statistics, the term "null hypothesis" usually refers to a general statement or default position that there is no relationship between two measured phenomena, or no association among groups.[1] Rejecting or disproving the null hypothesis — and thus concluding that there are grounds for believing that there is a relationship between two phenomena (e.g. that a potential treatment has a measurable effect) — is a central task in the modern practice of science, and gives a precise criterion for rejecting a hypothesis.

Notice that last bit: It is one of the jobs of Science, to overturn those "Null Hypothesis" (ie. those "status quo" theories) -- with better, more accurate Hypotheses ... (of how things really work, to the best of our accumulated knowledge).

Onto the first conclusion of that multi-nation University study (minus some the 'eye-glazing' numbers):

​The [data] show a statistically significant difference between the groups [paper-trail states vs non-paper trail states]. States without paper trails yielded higher support for Secretary Clinton, than states with paper trails ... As such, the potential for election fraud in voting procedures is strongly related to enhanced electoral outcomes for Secretary Clinton. P = 0.003

Remember the definition for P-value: "The p-value is defined as the probability of obtaining a result equal to or "more extreme" than what was actually observed, when the null hypothesis is true."

The lower the p-score, the less credible is the "status quo" theory is ... given the data.

IOW there is a 0.3% probability (less than 1%) of seeing this same difference in results or worse (between those those 2 groups of states) -- by chance alone -- assuming the non-paper trail systems were NOT tampered with in favor of one candidate (assuming the Null Hypothesis holds, = No Tampering). Or more simply the conjectured theory of the "rigging" occurring within the group of Electronic system states, in the favor of one candidate over the other, IS statistically significant. (with an "effect size" stat of d = 1.19)

IOW the Apple-turning Theory has statistical merit. (ie. that there WAS Tampering in that e-machine group, = the Null-canceling theory).

But what does this d-stat of 1.19 mean?

Let's let some eggheads from the citadels of Science explain it:

How to Calculate Effect Size -- ncbi.nlm.nih.gov

Cohen's term d is an example of this type of effect size index. Cohen classified effect sizes as small (d = 0.2), medium (d = 0.5), and large (d ≥ 0.8).[5] According to Cohen, “a medium effect of .5 is visible to the naked eye of a careful observer. A small effect of .2 is noticeably smaller than medium but not so small as to be trivial. A large effect of .8 is the same distance above the medium as small is below it.”[6]

So statistically speaking, if a d of 0.8 has a large real world impact, a d = 1.19 much be even more largely impactful as well.

Now to what I find to be the most interesting result of the recent study, concerning the Exit Polls vs the Actual Results (broken out by the Voting-method Groups)

As I recall, there were similar system-wide skewing of results in favor of GW Bush vs Kerry back in 2004, but it was swept under the national rug after a few days of media preponderance. It seemed the national media at the time had a "intellectual crisis" or two, as a long-trusted and statistically accurate tool of theirs (Exit Polling), had to be quietly ignored and tossed into the trash bin of "Expediency" post haste. ... I recall hearing about court testimony concerning these Exit Polls "anomalies", where a renown Statistical Expert explained it to the Judge: that for such Exit Poll anti-skewing to happen like that, in so many places --by pure statistical chance -- with no human interference, would be something like "a Trillion to One odds."

Numbers like that should definitely raise some eyebrows (and ire's) ... and so should those these (again from the opening study):

On the overall, are the exit polls different from the final results? Yes they are. The data show lower support for Secretary Clinton in exit polls than the final results would suggest ... P = 0.002, d = 0.71.

While an effect size of 0.71 is quite substantial, and suggests a considerable difference between exit polls and outcomes, we expected that this difference would be even more exaggerated in states without paper voting trails. Indeed, the effect size in states without paper voting trails is considerably larger: (d=) 1.50, and yields more exaggerated support for the Secretary in the hours following the exit polls ... P < 0.001.

In contrast, the effect size is much smaller in states with paper trails, ... P = 0.04, d = 0.58.

IOW there is a 0.1% probability (less than one-tenth of 1%) of seeing this same result or worse (of those contra-predictive Exit Polls, illustrated by the pie charts) -- by chance alone -- assuming the Null theory that the Exit Polls should, by construction, always favor the winning candidate.

IE. the conjectured "rigging" of the Electronic systems in favor of one candidate over the other one, on the basis the Exit Poll baseline, IS VERY statistically significant here. (with an "effect size" stat of d = 1.50, which in the language of Science is HUUUUUGE!!!)

The "control" version of this theory on the other hand (ie. the results in just the paper-trail states) have a 4% probability of seeing Hillary beating the exit poll sampling, by chance alone. This effect d = 0.58, shows a "medium effect" in the deviation from normal.

IOW the results of the contra-predictive BIAS displayed, in the states where Voting Machines have NO paper trails, are WAY more "extreme" {at over-turning the Null Theory}, than what occurred in the more 'run of the mill' control group of states that do (have a paper audit trail).

Notes: Here are some of my 'objective' concerns with this study, "scientifically" speaking:

1) Was there any reason for voters lie during the Exit Polls? (en-mass, in so many places? ie. were there scads of Hill-leaners, but afraid to admit it, etc.)

2) Were the Exit Polls truly randomly sampled? (ie. did they ask every tenth Voter leaving; in all precincts, etc.)

3) Is the study assumption that Caucus states can be a "good proxy" for "paper-trail" voting states, valid? (It would have been better if they, Caucus states, had their own "study group" IMO, as those states often had their own "irregularity" issues.)

4) Is there some "other correlation" going on, between the e-machine states and local "voter preferences" that might be effecting [confounding] the study's results? (eg. Do the "closed primary" states tend to also be the e-machine states; what about blue collar vs urban predominance, vs e-machines, etc.)

The study results ARE significant, BUT other factors may also be at play here. So caution, and further study would be well advised.

BUT bottom-line: E-voting machines need a Paper Trails. If we get a receipt at the ATM -- then why No Receipt with our Votes???

Which is more valuable, in its context? Which is more in need of some serious auditing?

(When was the last time anyone got a "surprise inspection" of their e-machine code? And why in the world are these inferno 'ghost-like' machines be 'propriety'? (Could there be an easier way "to buy" an Election ...)

As I observed at the time (cc 2004) one of the e-machine programmers had some interesting statements about how insecure (ie. hackable) those 'inferno' machines really were ... are? -- (although the supporting link no longer exists [www equalccw.com/dieboldtestnotes.html] that once documented his statements):

Diebold Programmer Ken Clark,
explains in an email to a supervisor,
the numerous security problems that exist
in the Diebold Vote Counting Program GEMS

Here's summary of those known security problems:

[See: Appendix A: Excerpts from Diebold internal memos/documents]

• Alteration of GEMS data files with an unapproved product (MS-Access) is common both among Diebold staff and county elections personnel.

• Ken Clark knew about that, and deliberately avoided tightening down security to eliminate the practice because it's "handy".

• Ken admits having the ability to hack elections(!). That means somebody else in Diebold can.

• Ken knew that relying on "operating system security" was inadequate, yet suggested telling the Federal test lab that it was. [was adequate]

This all adds up to intent to defraud.

Or put more simply in terms of the Null 'Status Quo' Hypothesis of the day:

link

Rep Peter King said that on 2004 Election day. As the Exit Polls were coming in, and seriously starting to worry some Bush-backers, and even some Media Pundits. Funny what a little alcohol will do, to help assuage those once-scientifically 'objective' fears ...

And as far as I know, the clowns in congress have done very little in the meantime

to "over-turn" King's theory. That's one apple-cart, the Villagers en-mass fear to touch.

Must.Not.Peer into "the Black-box" ... or you'll go civicly-blind if you do.

Democracy would not survive that searing light of day ...

Nor those brave {political} scientists who dare to ask:

"Where do E-votes go when they Die?" (... or more to the point "when they are flipped?")

.
.
.

And the short Dilbert Answer is:   into the 'winners' column silly, post haste.

And none's the wiser. ... It's not for nothing, it's called a "Black-box" afterall ...