General Election 2021 Results Analysis

[Roger Mexico]

10 minutes ago, Andy Onchan said:

But why would there be a last minute rush to register? 

Because people realise they're not registered and want to vote.  They may have moved recently and not got round to it, they may suddenly find the forms in a pile of papers, they might meet someone canvassing who tells them they're not on the register, they may have turned up to vote at the local elections and discovered they didn't have their name down.  There's dozens of reasons including having moved in the last few months  In the case of Douglas East I suspect the high increase there may be due to the MLP's signing people up who weren't registered (they would have done this in the locals as well.

Also it's not all 'last minute' but since 1st April because of the way they only update the register quarterly.

[Barlow]

1 hour ago, Declan said:

 

So if you look at my list from Josem in 38th upwards only Ciapelli in 24th and Wannenburgh in 29th didn't benefit from at least one of these factors. 

Both Damian Ciapelli and John Wannenburgh are well known in legal and business circles and generally.

[Roger Mexico]

While we don't have the official (or even reliable) turnout figures yet, we do have media-reported ones from eight constituencies.  We also have the number of people who voted[1] and I though a side-by-side comparison with 2016 would be interesting:

Most turnouts are down (though I'm a bit dubious about the Douglas East figure as it implies the electorate has grown by 30%) but the fall in Ramsey is quite dramatic.  perhaps it shows the need for a longer campaign than one that starts on nomination day.  The two increases so far are Douglas South[2] and Glenfaba and Peel where 2016 turnout was clearly depressed by the choice of candidates.  In some others a big increase in electorate may be responsible for some of it for reasons discussed before - not just Douglas East but Ramsey.

Despite what will probably be a drop in overall turnout it's  interesting that the number of people who voted actually went up by over a thousand.  This is because of higher electorates, but though all constituencies have more potential voters, not had more voting.  Douglas South and Glenfaba and Peel both have big jumps as you'd expect (though it doesn't result in much increase in turnout because of population growth).  But Douglas North shows a drop, presumably because of an uninviting selection of candidates and so does Onchan where perhaps the current two were seen as certs.  Ramsey's shows the turnout fall wasn't just due to new voters and Douglas Central maybe also suffered from a late campaign.  Rushen's increase is probably down to good campaigning by good candidates.

 

[1]   This excludes spoilt papers which are usually included in the turnout, but as ever the number of these is low (under 100 in total).

[2]  We already saw this in the by-election where the turnout was the same as at the previous general election.  This itself is incredibly rare - (Wiki keeps a list and the last UK one was 1981) and suggests a certain amount of political energising going on - for which I suspect Christian must take some credit.

[Roger Mexico]

Just when you thought you'd escaped my long, boring, number-y posts, I thought I'd give you another one.  This one concerns what you might call the unused votes.  That's the difference between the number of votes that could be cast (2 X the number of people who voted) and those that actually were.  It's the measure of how many people plumped for just one candidate.  If 10% of votes were unused that means 20% of voters plumped.

These are the figures for 2021 compared to 2016:

The most interesting thing about this is how consistent things usually are.  There's a slight drop over the five years, but the overall percentage is about the same.  And this time it's pretty even over the constituencies.  Roughly 20-25% of voters plumped everywhere with two exceptions in the mid-teens.

It was more or less similar in 2016 with two glaring exceptions.  In Glenfaba and Peel over 40% plumped of those who bothered to vote (turnout was down and there were a lot more than normal spoilt votes) presumably the small number and quality of the candidates made choosing even one a struggle.  That's been reversed this time with the second lowest score, possible because many wanted to make sure the current MHKs didn't get in.

The other 2016 exception was Garff where I always reckoned the low score showed that the count was flawed and result was inaccurate even before we started hearing rumours of missing ballot boxes and so on (the same was true of the second lowest Ayre and Michael of course).  Garff is now back to 'normal' which suggests that analysis was correct - not that it would ever been admitted.

The biggest anomaly this time is Ramsey, though it's at the edge of what you would historically expect.  I suspect this may link to @Rhumsaa's comment on the previous page.   He said that Allinson and Hooper  got about 100 plumps each.  There would have been a bit over 400 plumps in total on these figures and (very roughly) those two got a quarter of the votes each while the remaining half was spread over the other six candidates, so that would tie in.   What got them back in was the 700 votes that chose them as a 'ticket'.  Fewer candidates might have put them more at risk and the lowish number of plumps could suggest there were some who wanted them both gone as in G&P.

[SleepyJoe]

It's at the edge of the known world

[finlo]

51 minutes ago, SleepyJoe said:

It's at the edge of the known world

Says a Peel guy!

[SleepyJoe]

I'm from south of the bridge

[Rhumsaa]

2 hours ago, SleepyJoe said:

I'm from south of the bridge

Ballaugh Bridge?

[SleepyJoe]

South of the south bridge

[Roger Mexico]

Just when you thought it was safe to put away the calculators, more number crunching from me I'm afraid.  I thought I'd look at what you might call the Front Seat Premium - in other words how much better did the MHK who came first do than the one who came second.  As previously I've used the percentage of those voting who actually voted for the candidate as a marker, as that evens out the differences in turnout and electorate[1] and too some extent number of candidates.  Here's the 2021 figures:

This is a fairly extraordinary distribution of numbers.  Normally you would expect a fairly even spread but instead there's two big clumps - five constituencies where the leading candidate was miles ahead and the other seven where it was less that 20% (indeed six less than 10%).  There's nothing in the middle.

It's not what we saw in 2016 when the spread was more even as you would expect:

(Though of course the A&M and Garff figures may be dodgy).

Here only two MHKs had really big leads (though Watterson's consistency is to be noted) and only two under 10%.

I'm not really sure what is happening here.  Clearly in some places it was generally felt among those voted that one person should get in, but they were less concerned with the other.  But all those closer vote need explaining as well.

 

[1]  We still don't have final electorates actually used but for example Glenfaba and Peel had 36% more voters than adjoining Middle in April.

[Zarley]

23 minutes ago, Roger Mexico said:

Just when you thought it was safe to put away the calculators, more number crunching from me I'm afraid. 

Stop apologising for adding to this thread. I for one appreciate your number crunching as it means I don't have to, but I am interested in the results. 

 

So, thanks, and keep crunching!

[Declan]

Is that 106% swing from Allinson to Hooper?

 

 

[Phillip Dearden]

3 hours ago, Roger Mexico said:

Just when you thought it was safe to put away the calculators, more number crunching from me I'm afraid.

I looked at a similar-ish issue. How do the constituencies compare in their ability to distinguish between candidates. For this I looked at the Standard Deviation of votes in each constituency. The results were as follows;

A&M and Glenfaba and Peel come top of the Standard Deviation tables but their results are different. A&M has two clear winners and 5 candidates who got significantly less votes. G&P has one clear winner, one who just beat two fellow contenders and three who were some way off.

The lower end of the SD table, ie the constituencies where votes were not  widely dispersed includes all four Douglas constituencies and Garf. Looking at the votes here, it does seem that the difference between winning and losing is not so pronounced. 3 of these also had the lowest voter turnout (as % of eligible) – does this suggest less voter engagement or more closely matched candidates?

A&M provided more candidates than anyone, a higher proportion of their eligible voters actually voted and this was a repeat of 2016 (ie the highest voter turnout) and the Standard Deviation calculation suggests they are the best at distinguishing between candidates they like and don't like. They seem to be good at this democracy stuff...but not perfect, theirs is the only return of votes not yet available on the Government web-site.

Whether the stronger distinction between preferred and less-preferred candidates is as a result of better voter ability to identify the qualities they want or whether it actually reflects a difference in quality of candidates is a good question, to which I don't know the answer.

 

 

Edited October 5, 2021 by Phillip Dearden

[Roger Mexico]

On 10/5/2021 at 2:56 PM, Declan said:

Is that 106% swing from Allinson to Hooper?

Political scientists have numerous ways of working out swing, depending on number of candidates, number of seats, electoral system, if there are political parties, the country involved and for all I know the direction of the wind.  The only thing for certain is that none of them ever agree.

All I can say is that Allinson went from getting the support of 84% of Ramsey electors in 2016 down to 46% this time.  Hooper did gain some, he went from 42% to 49% (though the overall number of people voting was down a bit) but most of Allinson's lost vote went to the other candidates.  There were six of these rather than three before, but it still suggests that either the Marina caused a lot of the dissatisfaction or there were other reasons.

[finlo]

9 minutes ago, Roger Mexico said:

Political scientists have numerous ways of working out swing, depending on number of candidates, number of seats, electoral system, if there are political parties, the country involved and for all I know the direction of the wind.  The only thing for certain is that none of them ever agree.

All I can say is that Allinson went from getting the support of 84% of Ramsey electors in 2016 down to 46% this time.  Hooper did gain some, he went from 42% to 49% (though the overall number of people voting was down a bit) but most of Allinson's lost vote went to the other candidates.  There were six of these rather than three before, but it still suggests that either the Marina caused a lot of the dissatisfaction or there were other reasons.

I think a lot of people have worked out that he just wants to be King of the hill.