Is Europeia Actually Less Political Than Before, Or Is It All Nonsense?

HEM

former
Jorts Connoisseur
Honoured Citizen
Citizen
Pronouns
he / him / his
"Is Europeia Actually Less Political Than Before, Or Is It All Nonsense?"
Written by HEM

One of the most frequently debated questions in Europeia over the last few years has been whether the region is more "boring" politically than it used to be. Are citizens less engaged in politics, and more devoted to the social scene of the region? Is our region slowly transitioning from the world's premier political region, to simply a popular hangout spot?

To answer some of these questions, I turned to Senate election statistics. Here's how I did it.


METHODOLOGY

I went through the entire Voting Booth archives and collected three pieces of information from every Senate election.

(1) The number of candidates running
(2) The total number of votes cast
(3) The number of Senate seats up for election.

Using this information, I defined the following variables.

Activity = The total number of votes divided by the number of Senate seats. That's to say, if a lot of people are voting in our region, the region is more active. More votes = more members = more activity.

This measure of regional activity, at least upon first glance, appears to be superior to the ancient "post count" measure of activity. This measure shows how many members are engaged enough to be on the forum during a 24 hour period, and find a poll to vote in it. This seems to be a better measure than "post count" which can be unduly influenced by spam games and special events.

Political Activity = The number of Senate candidates divided by activity. That's to say, the more people we have running for Senate in proportion to people casting votes, the more political activity we have.

To illustrate my methodology further, here is a basic example:

Let's assume Europeia has 10 citizens, and that all 10 citizens are running for Senate, and all 10 citizens cast 7 votes for a 7 person Senate:

Example!
Senate candidates:10
Senate votes:70
Number of seats:7
Activity: (Votes per seat):10
Political activity: (Number of Candidates / Activity):1


In this situation we have "perfect political activity" because every single citizen is participating in Senate elections by both voting and running for a Senate seat. This scenario is (of course) totally infeasible. But it helps to explain how the data will be displayed. The closer the political activity coefficient is to "1" the more political activity Europeia has. The closer the political activity coefficient is to "0" the less political activity Europeia has.


POLITICAL ACTIVITY OVER TIME

Here is the resulting graph of "political activity" over time:



There has been, on average, a slow decline in political activity over the course of Europeia's existence. This general trend was briefly rebuked throughout 2014, but since, has seen a dramatic drop-off in political activity and participation.

As political activity reaches its lowest points in history, ironically, regional activity has reached some of it's highest — by far:



More people have been voting in Senate elections — relative to the number of seats — than any point in Europeian history (by far). Our citizenry is very attuned to elections, and very keen to vote, even in Senate elections that have typically attracted far less fanfare than Presidential elections.

Why is Europeia experiencing the highest activity ever at the same time we are experiencing the lowest political activity of all time?


IS PERMANENT GROUPCHAT TO BLAME?

Over the last few months and days, many people have saddled Europeia's "groupchat" culture with the blame. While Europeians have often used off-site messaging services (prominently MSN Messenger), only since January 2014 have we seen permanent-standing group chats such as EuroChat on Skype, or our Discord servers. While previously off-site services were used for individual chats, or temporary group chats, the bulk of social activity took place on forum. Now, permanent group chat rooms can allow for an increasing level of socialization (and perhaps politiking) to take place away from Europeia. Some people have claimed that this has watered down Europeia as a region, and resulted in a decreased interest in politics.

I split up some of the data and compared the median political activity from July 2007 — December 2013 and February 2014 — October 2016.

The differences are telling:

Median Political Activity
Pre-EuroChat0.46
Post-EuroChat0.32

Remember 1.0 is perfect political activity, 0.0 is no political activity

When comparing the two eras, there is almost a 15% drop in political activity after the introduction of permanent group-chat services.


ARE THERE OTHER FACTORS?

Maybe. Here I took the political activity chart and added some lines to represent important events in Europeian history that might be impacting these numbers.



Key
ColorEvent
PinkIntroduction of Civil Service
GreenCreation of Electoral Panel
OrangeShift of Senate elections from concurrent with Presidentials, to Midterm style
Yellow Popularization of "EuroChat" on Skype
Red Popularization of Discord

In order they appear on chart from left to right​

Just looking at the chart qualitatively, it's difficult to see any clear causation. Even if the introduction of EuroChat on Skype is followed by the political activity "resurgence" through 2014, which seems to suggest a positive impact, not negative (though again, the numbers do suggest that median political activity is down compared to before its introduction).

There are many who believe the shift from having Senate elections concurrently to midterm elections was a boon for the Senate. You could argue that the negative impacts of EuroChat on Skype was partially offset by the shift by the long-term effects of the new election system.

Indeed, so much change was happening in Europeian culture from 2013 - 2015 there are many ways to interpret the data. What was helping? What was hurting? What wasn't having an impact at all? It's hard to say.


CONCLUSION

Based on the data, it seems undeniable that political activity in Europeia is down from any point in history. However, it also appears that "total" activity is up, perhaps at unprecedented levels.

The data also seems to suggest that the introduction of permanent group chat services like EuroChat on Skype, and Discord, have had a negative impact on political activity (a 15% drop). However, there are a number of other factors that could have resulted in the decline, and it is thoroughly difficult to attribute historical fluctuations to any single event.


CAVEATS & COUNTER ARGUMENTS

There are some other important issues that were not discussed, however.

(1) Bad Data

Unfortunately, a number of polls have been tainted by (evil) people voting after the respective elections. Indeed, there are a few polls from 2007 that have a political activity coefficient over "1" which is literally impossible. While I imagine that political activity in 2007 was very high, we don't know exactly how high, or how many bogus votes that were cast.

(2) My metrics

My argument hinges on the notion that the number of people running for Senate can be effective stand-in for total political activity in the region. The biggest rebuttal to this may be that this is not an effective metric. Some may say that more people may simply be interested in executive involvement, and may be seeking more assistant ministry jobs than running for Senate.

This is an important argument, however I am skeptical. Political activity (as defined by the number of people running for Senate divided by total activity) after the introduction of the original Civil Service (the biggest expansion of the executive branch in our history, was fairly steady through the next two years with only "normal" fluctuations. If executive involvement were solely responsible for people not running for Senate, I would expect a huge drop-off in the months after the Civil Service's creation.

(3) Specialization Rebuttal

This is similar to (2). Basically, as group sizes get larger, people tend to specialize. That is to say, as Europeia grew in numbers and total activity, people gravitate toward "focus areas". While in 2007 it might have been necessary for someone to hold jobs in two different branches — so maybe almost everyone runs for Senate — in 2016 it is possible to just focus on cultural pursuits, or Cabinet jobs.

That is to say, people may argue that Senate candidacy may be a bad metric for "political activity" because less people will tend to be interested in law-writing as the region grows larger.

This argument is probably the one I was most cognizant of when using Senate figures to extrapolate conclusions about the region. However, I hold the belief that specialization cannot explain the drop off in Senate candidates better than decreased political activity and interest does. Specialization assumes that people are going elsewhere to be politically involved while the political activity theory assumes these people are simply not interested or engaged in Europeian politics at all. For the specialization theory to disprove the conclusions I've made here, it has to shown me where people have gone. However, the Cabinet tends to have a (roughly) fixed number of positions. Sub-Cabinet positions do exist, however there is no evidence that growth of the executive department with the Civil Service's creation in 2009 drastically impacted Senate elections. This is qualitative, but there certainly is no evidence that the press is more active (on median) than it was previously.

Because of those factors, I do not believe people are "specializing" they are just not politically involved, and hence the use of Senate candidacy figures to stand-in for the measure of political activity is appropriate.

(4) Vote Casting Rebuttal

The use of number of votes per seat to represent total activity hinges on the assumption that all citizens cast the total number of votes available to them through history. Or, at least, the variance among the number of votes cast by each citizen through history is consistent. If, for some reason, voters were more picky with their Senate votes in early Europeian history, and suddenly they became less picky over time, it could make it look like less Senate candidates are running compared to total votes per candidate when actually just more votes are being cast.

I can't think of any reason this would be the case, but it is a valid counter-argument.
 
This is a very timely (and doubtless it was a time consuming) piece. I think I'll examine this closely later (and it deserves close examination) but to be charitable for the time being: it should be relatively easy to adjust the metrics to total citizen count.
 
I have concerns about how the metric of 'political activity' is generated. For one behavior could have changed over time (for example voters only voting for party endorsed candidates, or only for a few candidates in order to strategically help those candidates) that decreases your score but does not indicate lower political activity.

Also we could see more activity over time by your measure (votes per senate seat) because the electoral panel sets seats and may not be increasing the senate seats at the speed that 'activity' at least would suggest is appropriate.

I will think more about this and comment again when I have put more energy into it. Thank you, in the meantime.
 
PhDre said:
I have concerns about how the metric of 'political activity' is generated. For one behavior could have changed over time (for example voters only voting for party endorsed candidates, or only for a few candidates in order to strategically help those candidates) that decreases your score but does not indicate lower political activity.

Also we could see more activity over time by your measure (votes per senate seat) because the electoral panel sets seats and may not be increasing the senate seats at the speed that 'activity' at least would suggest is appropriate.

I will think more about this and comment again when I have put more energy into it. Thank you, in the meantime.
Your first concern is theoretically possible, your second is due to a misunderstanding I think.

Total activity is found by dividing the total number of votes by number of Senate seats. The reason the division is even there, is to standardize the number of votes. So instead of looking at total number of votes, we are looking at the number of votes per seat.

Otherwise, the vote metric would be less useful because the number of votes cast each election varies with the number of seats being elected.
 
This is a really cool article and I appreciate all the time it took to put this data together but I think there was a counter argument that you may have left out. The argument however goes of a few assumptions that I don't necessarily have data for.

I don't have the citizen numbers but when looking at the graph it would appear that as the number of citizens increased the number of senate candidates stayed relatively flat (trend line). I would guess that since the number of seats has a relatively small range the number of senate candidates peaked respective to the number of seats avalible, even with the electoral panel. It rarely changes more than a few seats so some people may decide to stay away from pilling on a max of like 15 candidate election because their chances of getting a seat are lower the more candidates pile on. However the number of citizens is much higher so there are more votes per candidate. Thus a lower number in your activity metric.
 
Just a preliminary thought. In your example you did a little trick that Grav and me are trying to tease out. You assumed that the total number of citizens equalled your number of candidates. You then reached the number "1" and gave it meaning. But that meaning literally collapses when I change one thing in your assumption: the number of citizens.

In the example, I can make the number of citizens 300, 50000, or 7 billion and I would still get "1". We would argue that you can't ascribe perfect political activity to that number in the 300, 50000 or indeed anything above 10. The meaning you ascribe to the statistic is sensitive to the total citizen count. Yet, the statistic itself is -not- sensitive at all to the total citizen count.

The number "1" -does- mean -something- but not what you are saying it means.

In general, a competitive field has a lower spread of votes per candidate. A competitive field also has the total number of votes closer to an integer multiple of the number of seats. These two demonstrate two things: votes are not getting clumped because some candidates are clearly unsuitable and that people are using all their votes.

Let's look at the second one in a bit more detail because it is worth exploring. If I have 7 seats, and I have 70 votes then I can safely say that everyone voting has used all their votes. This isn't a proxy for activity, rather, it is an indication of how strong the field is. In your formulation, (2) ÷ (3) gives us activity. I don't think it does though. I think it gives an indication (and I'll explain why it really is only a weak indication) of field strength.

(2)÷(3) will give us numbers like 44.54, 10.3, etc. These numbers are only useful however if we can compare them to the actual number of voters. In our previous example the number 10, because it is an integer has an inescapable meaning. There were 10 voters and they all used 7 votes. However, what would 6.3 mean? Very quickly we see that this metric is only useful if two things are known:
1) Citizen count
2) Number of citizens who actually voted

The second piece is impossible to know.

But this -is- leading us somewhere. If we know the citizen count and the number of Senate seats then we do know the maximum possible number of votes. In HEM example, we know the citizen count is 10. We know the number of seats is 7. So the maximum theoretical is 70. That is reached. So how close the actual is to the maximum theoretical tells us two things:
1) How invested the entire population is;
2) How strong the field is.

I can make that difference smaller by getting the same number of.people to use more of their options or get more people to vote period (or both).
 
Just a preliminary thought. In your example you did a little trick that Grav and me are trying to tease out. You assumed that the total number of citizens equalled your number of candidates. You then reached the number "1" and gave it meaning. But that meaning literally collapses when I change one thing in your assumption: the number of citizens.

In the example, I can make the number of citizens 300, 50000, or 7 billion and I would still get "1". We would argue that you can't ascribe perfect political activity to that number in the 300, 50000 or indeed anything above 10. The meaning you ascribe to the statistic is sensitive to the total citizen count. Yet, the statistic itself is -not- sensitive at all to the total citizen count.

The number "1" -does- mean -something- but not what you are saying it means.

I think you are right here.

In general, a competitive field has a lower spread of votes per candidate. A competitive field also has the total number of votes closer to an integer multiple of the number of seats. These two demonstrate two things: votes are not getting clumped because some candidates are clearly unsuitable and that people are using all their votes.

This part I did address though, at least indirectly. Your assumption here is that most citizens, most of the time, don't use all their votes. Or at least there is high variability in such thing.

If that is correct, I think you are right. My tentative conclusion is that I thought voting patterns in terms of number of votes used were pretty constant. Your counter-argument about weak vs. strong fields is compelling, however.

EDIT: The feedback so far is very rich. Very willing to take advice and re-approach this topic.
 
Also you state in your caveat section that it is impossible to go beyond "1" and some bad people have been voting is historic polls. This is not true.

Your "political activity" metric is Number of candidates/activity. But "activity" is total votes/number of seats.

Doing the math this means in a case where "pol' act'" is >1, that:

(Number of candidates)(Number of seats) > (Number of total votes)

This is trivally easy to achieve legitmately. In HEM's example, just make each voter only vote for one person. The "political activity" "coeffiecient" can be greater than one.

 
hyanygo said:
Also you state in your caveat section that it is impossible to go beyond "1" and some bad people have been voting is historic polls. This is not true.

Your "political activity" metric is Number of candidates/activity. But "activity" is total votes/number of seats.

Doing the math this means in a case where "pol' act'" is >1, that:

(Number of candidates)(Number of seats) > (Number of total votes)

This is trivally easy to achieve legitmately. In HEM's example, just make each voter only vote for one person. The "political activity" "coeffiecient" can be greater than one.
I think we missed each other, but I do agree that my interpretation of the meaning of the "coefficient" number was very incorrect. However, I think the number's change over time still has relevance.
 
I know voters don't use all their votes because I have never done so for the last 5 years. Also, it's really easy to check that you don't get whole numbers from (2)÷(3).

I think that the next step is getting citizen number data. I do think the data you currently have tells us -something- but not what you want to tell. I think getting citizen number data will help us tell the story you want to tell.
 
HEM said:
hyanygo said:
Also you state in your caveat section that it is impossible to go beyond "1" and some bad people have been voting is historic polls. This is not true.

Your "political activity" metric is Number of candidates/activity. But "activity" is total votes/number of seats.

Doing the math this means in a case where "pol' act'" is >1, that:

(Number of candidates)(Number of seats) > (Number of total votes)

This is trivally easy to achieve legitmately. In HEM's example, just make each voter only vote for one person. The "political activity" "coeffiecient" can be greater than one.
I think we missed each other, but I do agree that my interpretation of the meaning of the "coefficient" number was very incorrect. However, I think the number's change over time still has relevance.
The number's change is relevant only as far as you can support it. I don't think that (number of candidates)(number of seats) ÷ (total votes) has any meaning whatsoever. Infact, look at the denominator. Total votes are a function of population. So the ratio will, virtually, always go down.

You are saying that this number tells a story. I agree but I don't think it is the story that you think it is or even if it does have meaning, that the meaning is important.
 
Very thought-provoking. I, for one, rarely use all my votes these days. It was more common for me to do so in the past, however. I would be interested to know if newer members are more likely to use all of their votes than older members. In this past election, I used 5 of my 7 votes, but as recently as a few months ago I only used 3. It may not change much how we think about this data, but now I'm curious about it anyway.
 
Also, your activity ratio is really "the number of citizens who voted assuming all the voters used all their choices". So it can be thought of as a weak "number of voters" and this scales with "citizen count".

I think then the political activity ratio makes a little more sense if we think of it as "senate candidate number/citizen count". Then you'll think that this gives you how engaged the total population is but you run into the ceiling effect Grav describes where the candidate numbers are effected by number of available of seats and the nukber of candidates already present. This last point is one that hasn't been made before but is important.

So immeidately what you think of as a useful statistic becomes a number that is pretty much always going down.

The shapes of your graphs are consistent with "citizen count" for "activity" and "candidates per citizen" for "political activity".

Putting it this way, the results are unsuprising and don't tell us much more than our population is increasing.
 
Sopo said:
Very thought-provoking. I, for one, rarely use all my votes these days. It was more common for me to do so in the past, however. I would be interested to know if newer members are more likely to use all of their votes than older members. In this past election, I used 5 of my 7 votes, but as recently as a few months ago I only used 3. It may not change much how we think about this data, but now I'm curious about it anyway.
In all the elections I have participated in I've only not used all my votes once if at all.
 
GraVandius said:
Sopo said:
Very thought-provoking. I, for one, rarely use all my votes these days. It was more common for me to do so in the past, however. I would be interested to know if newer members are more likely to use all of their votes than older members. In this past election, I used 5 of my 7 votes, but as recently as a few months ago I only used 3. It may not change much how we think about this data, but now I'm curious about it anyway.
In all the elections I have participated in I've only not used all my votes once if at all.
Voting for everyone who you think is qualified is not strategic. For example, people who felt super strongly about Skizzy being elected should have only voted for him. I know if I had voted 'strategically' I would have a) waited until the last minute to vote in the general Senate election b) only voted for Skizzy in this example.

Instead, I voted for all three of Calvin, Skizzy and Possibly This earlier in the election. But anyway, being strategic means not using all available votes.




A radical and bad idea I had was for us to vote for anonymous platforms instead of candidates. :sleep:
 
Back
Top