# Low voter turnout in elections

Before the election mania is over let us do some analysis of the trends in elections so far. The relevant data is presented in the following table

$$\begin{array}{|r|r|r|r|r|} \hline\\ \text{Year} & \text{Voted} & \text{Visited Election} & \text{Visited Site} & \text{Eligible} \\ \hline 2010 & 174 & & & 394\\ 2011 & 225 & 378 & 615 & 1050\\ 2012 & 394 & 677 & 1299 & 2559\\ 2013 & 437 & 1575 & 2137 & 5313\\ 2014 & 1425 & 3731 & 3951 & 11408\\ 2017 & 2161 & 4655 & 9261 & 22919\\ 2018 & 2106 & 4559 & 9069 & 25960\\ 2020 & 2368 & 6152 & 11568 & 46680\\ \hline\end{array}$$

Here are my observations:

• Roughly one third to one half of the people who visited the election page also voted. The year 2013 is rather an exception when this ratio is around $$1/4$$.
• All the numbers show an almost increasing trend which indicates that more and more people are taking part in this process.

However my main concern here is the low voter turnout. Looking at the number of people who visited election page, the ratio of people who actually voted barely reaches 50 percent. This indicates that people are interested to look at the candidates but a significant number does not vote. My guess is that they don't find any of the candidates suitable from their perspective.

What could be other possible reasons? How can we try to improve the situation? We need to hear out those who did not vote (but did visit election page). Perhaps we can start a meta thread during nomination phase which allows people to raise concerns which have not been addressed by existing candidates (this should not however lead to an extra questionnaire thread).

I would like to hear opinions about this. If there are any more disturbing observations from the table given above please discuss them so that we can look for more areas in need of improvement.

• From previous elections: math.meta.stackexchange.com/questions/27219/… – Asaf Karagila Aug 5 '20 at 12:43
• Thanks @AsafKaragila for linking that thread. I was aware of it but I forgot to link it. Given this thread in 2017 the turnout was similar in 2018 but this time it was much lower. Maybe voters knew that chances of a candidate winning this time was low (just 1 mod position) and hence some of them abstained from voting. But we can't be sure. – Paramanand Singh Aug 5 '20 at 13:14
• I'll just throw in my reaction that I was impressed by the number of people who voted, and that the comment by Joonas Ilmavirta in the thread lnked by Asaf represents my views pretty well. As someone who has always loved math, I've gotten very used to the things that I think are important not being important to others. And I think that my interest in the meta portion of this site is similar. Glad to have others who care, open and curious to hear from those who have a good-faith issue with how things work and want to suggest changes, but don't feel a pressing need to ramp up this qeustion. – JonathanZ supports MonicaC Aug 5 '20 at 14:22
• I'm not so sure those numbers are a helpful representation of the process - visiting the election page is just clicking on a link that is prominently shown to you, so is mostly passive. Voting requires you actually do something. I don't have much of a sense of what a reasonable ratio between those numbers would be, but 50% doesn't strike me as outrageous. – Milo Brandt Aug 5 '20 at 14:58
• One may also want to compare the voting data of the questionnaire to the election voting data. – user9464 Aug 5 '20 at 15:03
• Does the number in "Visited Site" include both meta and main? – user9464 Aug 5 '20 at 15:09
• @T.S: there is no such demarcation in elections page (or opavote page). My guess is that this refers to visits on main site so that this gives number of active users on main site. – Paramanand Singh Aug 5 '20 at 15:12
• @ParamanandSingh: thanks for that. I just thought that number of visits to meta may be more relevant; it would be more useful if one can see both numbers (meta and main). – user9464 Aug 5 '20 at 15:16
• @MiloBrandt: given the large jump this year in eligible voters and visitors to election page I find the voter turnout rather low. – Paramanand Singh Aug 5 '20 at 15:22
• @ParamanandSingh I hope you run next time, btw. – rschwieb Aug 5 '20 at 18:49
• I think also, at least in the US, where I see lots of people tune into candidate debates, and follow the news, and debate positions, the majority of those people don't end up voting. Due to apathy? Due to a sense of powerlessness to effect change through a vote? But it happens regularly, that those who vote determine who the subsequent leaders will be. And oftentimes, those who don't vote, complain afterwards, no matter the result. So I don't know, and would not suspect, that there is any inherent problem math.se or SE. Those who vote, and those who don't, equally determine...the future. – amWhy Aug 5 '20 at 19:05
• Just again, as a single data point, I followed the election fairly closely, read some of the answers, but almost missed voting because I had a very busy week leading up to the cut-off. That might argue for elections to stay open longer, but I have no idea how many people are having the same experience. – JonathanZ supports MonicaC Aug 5 '20 at 19:44
• @rschwieb: this is taken from election page. Well, these are written there just as text and I had to put them in table using mathjax array. – Paramanand Singh Aug 7 '20 at 15:40
• If it’s just a matter of 150 rep, it seems entirely plausible to me that a great deal of people were enfranchised when reputation was recalculated to be 10 per question upvote instead of 5 – rschwieb Aug 7 '20 at 23:38
• @rschwieb: I really wonder why I missed this part about recalculation of reputation. Simplicity often eludes us. – Paramanand Singh Aug 8 '20 at 1:32

I think the main reason for the perhaps low-seeming voter count is that most users are not very interested in meta matters like elections. I thought so back in 2017 and I still agree with what I wrote then.

If you want more useful statistics, I don't think the number of eligible users is a very descriptive number. As the site grows, especially newer users will see it less as a community they have a role in and perhaps more as a playground for learning and teaching math (and getting some internet points). What I would like to see is data for the number of "engaged and eligible users", which could include additional qualifications like "has voted to close or reopen ten times" or "has accessed three review queues" or "has given over three votes on meta" (maybe within the last 12 months). I suspect that the number of these users has not grown as fast as the number of eligible users. More nuanced data is needed to estimate the size of the "effective electorate", the set of users who can vote and care about it.

I cannot remember the details, but I assume the election is run and advertised a little differently. As the execution evolves, so does voter interest. I don't know how big an effect this has, but I'll share a personal view: The election questionnaire is too long, and I think a minority of those who voted cared to take a look at it. I am interested in the health and future of this site, but I could not bring myself to read the whole thing; I picked one or two key points and quickly extracted the relevant bit of the candidates' views on those and entirely skipped the rest. Perhaps people feel that a voting decision is behind too many walls of text?

Some might also be upset about last year's storm regarding moderators on the network — and this storm hasn't quite settled fully yet — and might refrain from participation in choosing moderators over lost trust. Most users will probably not know or care about the events, but I expect that people that care most about that are strongly correlated with the people who would typically vote. I don't think this is a major factor here, but it's hard to be sure.

Not all users will care about how the site is run and I wouldn't strive for it. I think it's perfectly fine to have an active core group, and I see nothing to worry about when that group can be measured in thousands and is on a slowly increasing trend. I think the meta participation on this site is sufficient for drawing meaningful conclusions about the opinion of the community.

Also, a lot of users are inactive. If someone became eligible in 2016 but stopped using the site in 2017, they should be excluded from your data. With the current data I see no reason for worry, but I would like to see what the trends are with more refined numbers. If, say, 40 thousand out of those 47 thousand users are here (if they are here at all) for math only, so be it. They might be disinterested in making the rules, but that is no issue as long as they are interested in following them — making the rules is a meta matter, following them is a main matter.

• I upvoted your post in '17 for it probably paints an accurate picture of the situation. Yet, that state of affairs is not very satisfactory. In terms of your soccer club analogy, in addition to working on their skills in controlling the ball and their bodies (=the math), they also need to have some agreements about a few basic rules like off-sides and where the sidelines are etc. Such "house rules" are also discussed in meta. Some club members not being aware of those rules reduces the enjoyment of the game for everybody. – Jyrki Lahtonen Aug 6 '20 at 5:06
• @JyrkiLahtonen Good point! I added one sentence at the end to make an important point I had left out: I only meant disinterest towards making and discussing the rules, not following them. If every new soccer player can say "it seems I can only kick the ball here and I need to pay my share of the rent — I'll do that", their lack of involvement in administration is no issue. If they are unsatisfied with the rules (whether they're playing math or soccer), they should join the discussion. Obedience and awareness of the rules doesn't require any further interest in them. – Joonas Ilmavirta Aug 6 '20 at 6:12
• Looking at the comments and answers in the entire thread here I can gather than there is no significant reason to get alarmed over the numbers presented in my question (cheers for everyone). Accepting this to remove it from the unanswered queue. – Paramanand Singh Aug 7 '20 at 5:02

As regards the turn-out to me the situation looks relatively stable. As detailed in other contributions already I think we can basically forget about the "eligible" number. Manifestly many accounts turn inactive and get abandoned entirely.

Speaking from my experience on some other SE sites, I am also not that interested in elections there, say, on Academia or Chess. In the end usually I tend to vote, but it really can go either way. I don't think that's a problem. Some sites one is actually involved others one uses more casually.

You say:

My guess is that they don't find any of the candidates suitable from their perspective.

I really don't think that this is the main reason. Especially in this election there was a quite diverse field. What may be an issue is name recognition; several of the candidates while highly qualified as candidates arguably were not among the most visible users.

That said, it does come up at times that there should be an option to actively not vote, and cast something like a blank ballot, which would allow to quantify that phenomenon a bit more.

• Regarding the last paragraph, this is a relatively new feature in Indian elections called NOTA (none of the above). However this is not used by voters too much. – Paramanand Singh Aug 7 '20 at 0:51
• Indeed, there are some highly reputable users which answer multiple questions per day. I think such a candidate would attract a lot more voters. – QuantumSpace Aug 8 '20 at 15:54
• @ε-δ honestly I don't think it would be "a lot", likely a bit but no all that much, and there is historical precedent to back that up, e.g., hardly anyone was more active than Jack D'Aurizio when he ran. – quid Aug 8 '20 at 16:11

Speaking as someone who occasionally visits stackoverflow sites, and has looked at the election pages, I decide I have no idea what makes a good person to vote for. I don't know anything about these people beyond their own statements, and my experience of other elections is personal statements are not that useful.

Therefore I choose not to vote, rather than introduce random noise.

• Not voting also introduces random noise. – Gerry Myerson Aug 16 '20 at 21:58
• Well here is my view. It is quite possible that a user visits the election page by clicking the prominent link of upcoming events, browses it casually for a minute and leaves the page. However if someone spends time reading the nominations then it is most likely that the user will vote. I also think that people can't be casually interested in elections, they may be disinterested instead. – Paramanand Singh Aug 17 '20 at 9:25
• I'm going to be honest, I find this reply a bit annoying. You asked for opinions, but your reply to me reads (to me) like saying "Most people aren't like you, if they read the nominations they will most likely vote". I would recommend, when asking for opinions, to not reply with your own opinion, as that might shut down discussion of other viewpoints. – Chris Jefferson Aug 17 '20 at 10:25
• I apologize if you feel offended and if you wish I will delete my comment. I wasn't really aware that when one is trying to find opinion of others one isn't allowed to express his own. Sorry again, offending anyone is not at all my agenda here. – Paramanand Singh Aug 17 '20 at 12:10
• Don't worry, I was probably too annoyed, and wanted to write out what I was thinking (I did go back and make my message less mean from what I originally thought). – Chris Jefferson Aug 17 '20 at 13:32

I have this in a comment, but I think it partially explains what happened and deserves the visibility of a solution.

On November 13, 2019 there was a change which converted upvote bonuses for a user's question from +5 to +10. I have been told that the only thing required for election voting eligibility is 150 reputation.

Considering how low the 150 threshold is, it seems likely that a good number of users were drawn into the pool of eligible voters by this change, inflating the number. (I'm not good enough with the SEDE queries to adapt this query to count exactly how many, but that is a good starting point. Any takers?)

Given this, the drop in the proportion of participants would be not as dramatic.