@causeisaidso
Forum Replies Created
-
I kind of have to stifle a laugh at the hypothesizing going on about voting correlation re: casual participants. To me it seemed perfectly obvious that themes with higher numbers of casual participants will garner higher numbers of total votes because if the theme generates more casual interest participants it will generate more casual interest viewers as well.
Hey, your husband started it! 🙂
But I screwed up in my previous post so it turns out you’re mostly right (as indeed should’ve been obvious). The figures I used as casual entrants are actually the regular entrant counts – those that entered 8 or more of the past 15 contests. So, the correlation between regular entrants and total votes is 0.60. The correlation between total casual entrants and total votes is 0.82, a much stronger relationship.
My suspicion is that the threads with a higher ratio of casuals to regulars than the average thread are the ones that coincide with higher vote totals.
I neglected this part of your hypothesis so I ran those numbers (and this time I think I’ve got it correct). It turns out that although the total count of casual entrants affects the total vote count somewhat strongly, the actual percentage of casual entrants does so less strongly. The correlation of percentage of casual entrants to total votes is 0.54.
I think the only thing we can safely deduce from any of this is that more entrants equals more entries which equals more votes. Which is kind of a “no duh!” statement.
But orionid‘s other idea about a “wall of win” got me thinking that there are two ways to measure the “success” of contest – total votes, sure, but also votes/entry. I think votes/entry may better indicate viewer interest as opposed to participation rate.
So I ran another set of correlation calculations for votes/entry and basically there’s nothing really to predict that (aside from the theme itself). The number of entries has absolutely no correlation to votes/entry (0.01). The strongest correlation is for percentage of casual entrants, but it’s a weak negative relationship at -0.27! So, a higher percentage of casual entrants in general somewhat equals a lower vote/entry.
Although December is always a huge fall-off month for participation and voting so no need to panic over this week’s total votes imho.
You’re right there (again). December not only produces a lower total vote count:
but it also produces a much lower average vote/entry:
And finally, for those of you still reading this, as kestrana suggested, here are the top ten themes that garnered the most votes over the past 3 years:
Date Theme Votes
2008-07-30 I Tawt I Taw A Puddytat 5546
2008-06-04 Trees 4092
2010-01-20 Sunrise. Sunset. 3691
2008-03-19 Nature Photography 3559
2008-07-23 Farktography Classic: Blue II 3282
2010-09-22 Farktography Classic: Green 2 3067
2010-10-20 Travel Photography 3047
2009-10-07 Farktography Classic: Black and White 2 3031
2010-01-06 Animals 3024
2009-05-27 Birds 2955and here are the top ten in terms of viewer interest as indicated by votes/entry:
Date Theme Votes/Entry
2009-06-10 Tilt-Shift Faked Miniature Scenes 17.38
2009-03-18 Tableau 15.44
2009-08-26 Training Documents 15.41
2009-12-02 Win, Fail, or "I Like Where This Thread is Going" 15.10
2010-12-01 The Song Remains The Same 14.80
2010-01-20 Sunrise. Sunset. 14.20
2010-09-22 Farktography Classic: Green 2 14.20
2008-07-23 Farktography Classic: Blue II 14.15
2009-01-21 Take Three 14.14
2009-04-29 Farkstronomy 13.80Basically, I figured the overall vote drive is driven by the number of casuals, and the way to find any truth in that would be by creating an algorithm as such…
Edited to add: I screwed up the calculation of casual entrants in this post, so please disregard all statements about casual entrant counts and relationships to votes. All figures quoted as casual entrant counts are actually regular entrant counts.
Dangit, orionid, you got my interest up so I had to go run some numbers. Turns out the data doesn’t support this theory, at least not strongly.
Those with mathphobia may want to skip to the next post…
I only used data from the last 3 years (1/2/2008 – 12/8/2010) to exclude some of the wilder swings seen during the frontier times. I used your criterion for “casual entrant” – someone who’s entering this contest but has entered less than 8 of the previous 15.
I ran a correlation calculation on the various variables against the vote counts. (For those not familiar with correlation, it’s a measure of how closely one dataset is tied to another. A correlation of 1 means the datasets are tightly bound together – a 10% rise in one variable should yield a 10% rise in the other. A correlation of -1 is a perfect inverse relationship where the second variable falls when the first rises.)
The strongest correlations are those that are intuitively obvious. The correlation of total number of entrants to total contest votes is 0.87, total entries to total votes is 0.91 and total entrants to total entries is 0.96. You can see that relationship in this graph:
But the correlation of casual entrants to votes is only 0.60. You can see in this graph that the vote count swings wildly even with the same casual entrant count:
Surprisingly, the correlation of casual entrants to number of entries is even weaker at 0.57. It seems to suggest that the theme itself has more to do with it as it encourages more entries from even the regulars and more entries equals more votes.
