Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:84860
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 74.125.82.53 as permitted sender)
Message-ID: <5505C875.1030202@gmail.com>
Date: Sun, 15 Mar 2015 17:59:17 +0000
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Thunderbird/31.5.0
MIME-Version: 1.0
To: internals@lists.php.net
References: <CAAyV7nEikYZNmaAnRy9hkRBnDDP=Ey43-nZbJGMHmQ_dAC_vgw@mail.gmail.com>
In-Reply-To: <CAAyV7nEikYZNmaAnRy9hkRBnDDP=Ey43-nZbJGMHmQ_dAC_vgw@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] Voting irregularities
From: rowan.collins@gmail.com (Rowan Collins)

On 15/03/2015 14:19, Anthony Ferrara wrote:
> All,
>
> I ran some numbers on the current votes of the dual-mode vote right
> now. There were a number of voters that I didn't recognize. So I
> decided to pull some stats.
>
> The following voters never voted before the dual-mode RFC went up:
>
> dom - no
> eliw - no
> kguest - yes
> kk - no
> nohn - no
> oliver - yes
> richsage - yes
> sammywg - no
> spriebsch - no
> srain - no
> theseer - no
> zimt - no
>
> Some of these names I recognize from list (sammywg and eliw), but many I do not.
>
> The interesting thing happens when you look at the voting direction.
>
> Currently, the RFC is slightly losing 70:37 (65.4%).
>
> If we look at percentages, 4.2% of yes voters have never voted in a
> prior RFC. But a whopping 24.3% of no voters have never voted before.

I think calling this an "irregularity" is going a bit far. It's an 
interesting observation, but since this is such a contentious issue, the 
question I would be asking is what these people have in common that 
makes them likely to vote no - are they from a particular part of the 
community whose voice is less often heard, for instance?

As I've just said on Twitter before seeing this thread, these are really 
small sample sizes, and the way you've framed the statistics there makes 
it sound more significant than it is. Wolfram Alpha tells me that if 12 
people chose their vote by tossing a coin, there's a probability of 
0.073 that 9 of them would vote the same way, which is higher than the 
threshold of 0.05 traditionally set for significance. I don't know if 
that's a valid statistic, but it's at least as scientific as your 
"whopping 24.3%".

If you look at those users as a proportion of the complete "turnout", 
you get 11.11% (1 in 9) votes coming from first-time voters. The net 
impact is 6 votes out of 108, which is about 5.5%; that happens to be 
enough to stop this vote crossing the line right now, but only because 
the vote is so close anyway.

> If we adjust the votes to remove these "never voted" accounts, it
> stands at 67:28. Which is 70.5%. Which is basically where the vote was
> prior to the competing RFC opening.

If you exclude an arbitrary subset of votes in a close ballot like this, 
it's easy to edge it past the finishing post, but that's really an abuse 
of statistics. For instance, you could say that if the vote was closed 
on date X, the result would have been Y, but you can't know that there 
weren't people who'd already decided which way they were voting, but 
hadn't got round to logging in, because they knew the vote wasn't due to 
close yet.

With more research, you could come up with other interesting subsets, 
like people who've voted less than X times, or not voted in the last X 
months. But if you're going to play with statistics, you should be 
rigourous in defining your hypothesis, and how you'll measure the 
significance of your result. Alternatively, leave the statistics out of 
it, and say that you're interested to know why these first-time voters 
voted how they did.

Regards,

-- 
Rowan Collins
[IMSoP]