Access/reading statistics for Election Interviews?

Hi, would it be possible to get statistics about the amount of impressions/readers for the election blog posts? I have a suspicion that they are rarely read and it would be interesting to check if they are read more than other posts (to see if the election makes a difference) and if their hits or difference in page hits match the number of voters (to check which percentage of voters are accessing them).

1 Like

I personally read every candidate interview before voting, but I’m curious as well

2 Likes

So do I.

1 Like
Election Candidate Comm Blog Views Discussion Views Votes/Max
Council Till Maas (till) 107 27 49%
Council Tom Callaway (spot) 105 25 75%
Council David Duncan (davdunc) 95 34 46%
FESCo Ahmed Almeleh (ahmedalmeleh) 78 X 16%
FESCo Fabio Valentini (decathorpe) 75 28 54%
FESCo Tom Stellard (tstellar) 73 X 41%
FESCo Miro Hrončok (churchyard) 65 25 73%
FESCo Zbigniew Jędrzejewski-Szmek (zbyszek) 65 26 58%
FESCo David Cantrell (dcantrell) 65 24 47%
FESCo Kevin Fenzi (kevin) 56 30 66%
Mindshare Fernando Fernández Mancera (ffmancera) 75 X 43%
Mindshare Stephen Snow (jakfrost) 62 26 48%
Mindshare Till Maas (till) 59 29 65%

For some reason, the posts I’ve marked with x did not get published here, which I am looking into now. Given the low number of views here for those posts overall (and that the elections app linked directly to the commblog), I don’t think it was election-shifting, but… it’s certainly not good. I’ll figure out what happened and fix it for the future.

The planned site re-org should make these a lot more prominent here, but I think both blog and forum posts for these are kind of hard to deal with, since it’s a lot all at once. I wonder if we could present them side-by-side somehow.

2 Likes

I don’t suppose we have metrics on time spent? It’d be useful to tell how many of those are “actually read it” vs “just opened the page”.

I don’t think we have that for WordPress at all. Discourse tracks time spent but I don’t think it exposes it. I could probably dig into it, but I’m not sure how we’d act on the information.

For most candidates (maybe all, I don’t recall off the top of my head), the number of voters for a candidate is higher than the number of interview views.

The table @mattdm shared doesn’t necessarily tell the story best, as he used the score, not the number of voters. So someone may have received a lower score because people did read their interview and it was terrible. :slight_smile: It’d be better to look at the views:voters ratio. I’ll pull those numbers together as time allows.

3 Likes

Thank you for the details.

These are the number of voters according to results pages at Fedora elections

Council: 252 voters
FESCo: 260 voters
Mindshare: 219 voters

It seems to me that the majority of voters does not care about the interviews. Only for the Council elections a little bit more than 50% might have read the interviews if no voter accessed both the Blog post and the discussion post.

I went back through the last four election cycles. There were a total of 52 candidates on all ballots. Here’s a quick summary of what I found.

There’s no meaningful relationship between the interview views and the number of voters. R² for a linear regression is 0.084. A logarithmic regression brings R² all the way up to 0.105. Even going with a 10th degree polynomial, R² was only 0.395. :slight_smile:

image

But what about the results? Let’s look at the percentage of possible votes received as a function of interview views. R² for a linear regression is 0.115. For logarithmic: 0.113. Again, there’s just no meaningful relationship between the interviews and the votes.

image

Finally, let’s look at the ratio of voters:views. Here are some quick summary statistics ( < 1 means more viewers than voters, > 1 means more voters than viewers):

Statistic Value
Minimum 0.830
1st Quartile 1.534
Median 1.840
3rd Quartile 2.073
Maximum 2.8
Mean 1.812
Standard deviation 0.407

I have long suspected that name recognition is the most important factor in election outcomes. If people recognize your name and they like you (or are at least indifferent from you), you get a lot of votes. If people recognize your name and they don’t like you (or if they don’t recognize your name), you get few votes. I bet if we compared to something like mailing list activity, we’d see a much stronger relationship.

So does this mean that the interviews are useless? I don’t know if I’d say that. I certainly read them, although I tend to base my votes more on my interactions with the candidates than the contents of their interviews. And there’s definitely something to be said for having that transparency, even if many people don’t take advantage of it. Perhaps people are more likely to read the interviews when they’re not familiar with the candidate?

I guess the takeaway for candidates is this: you don’t have to worry too much about what you put in the interview responses.

1 Like

Well, that’s one takeaway, but it’s kind of depressing. How can we get more people to read these? I think they’re important!

How do you make a horse (that you’ve previously led to water) drink? Apart from requiring voters to click the interview links before casting a ballot, I’m not sure there’s much more we can do. We can change the schedule to publish interviews a day or two in advance, but I doubt that would do much to increase readership. I suspect we probably do better in the “read the candidate positions before voting” category than most governmental elections anyway.

I agree that the interviews can be useful in some circumstances. I’m not sure that they definitively are. To some degree, the fact that people are voting based on reputation and contribution history is appropriate for a “do-ocracy”. Actions speak louder than words, etc. The downside is that it disadvantages people who contribute in less visible (but not less important!) ways.

By and large, I think the elections are working pretty well. Or at least, I’ve been personally happy with the results. My preferred candidates don’t always win, but the winners have been solid.

That said, we can certainly make changes. Instead of static interviews, maybe we could make them more participatory. Remember how we used to do IRC town halls? Let’s not bring those back. But maybe we could have each candidate post a discussion thread with a few pre-selected questions and then have it be an asynchronous interview with the community. In this case, we’d start the threads shortly after the release and then lock them (or not) when the voting starts.

1 Like

Like, a Candidate AMA?

Yes! With a few starter questions that we ask of everyone.

Give people a pop quiz about the candidate before they can vote for them :stuck_out_tongue:

(a bit tongue in cheek, I know)