Have the FJs been getting easier? An analysis

[Jasoni22]

I went through all the FJ polls for Season 29 and compiled them into a spreadsheet. To make sure we're comparing apples to apples, I removed all tournament games. (Yes, I used Excel. Yes, I know it's inferior to Minitab and SPSS. Deal with it.)

Graph 1

This first graph shows the % correct on the JBoard poll for each FJ. As you can see, it's incredibly noisy. Peaks and valleys seem to occur regularly, and no high or low plateaus are apparent. How about fitting some trendlines? A linear regression yields a line that increases from around 55% to 65%, but its R^2 value is 0.0254, so it's pretty much useless. We can try to smooth the data out by applying some SMAs (simple moving averages), but a 5-day SMA is no help, and even a 20-day SMA isn't much better, although the latter does show an increase from the 50-60% range in 2012 to something solidly in the mid-60s in 2013.

What about if we look at them on a weekly basis? (To make it easier to ignore tournament data, let's assume the week starts on a Wednesday.)

Graph 2

Again, we see a graph with no real discernible pattern, but again, a 4-period SMA (4 weeks, or 20 days, like in the previous graph) shows a small but definite increase these past couple of months.

Conclusion? It is likely that the FJs have been easier recently (p < seriously, you're expecting me to calculate p-values at 3 AM?), although not to the degree that it had seemed to be to some here (including me).

[immaf]

Perhaps boardies have gotten better? Or perhaps those who don't always do so well have stopped reporting?

I'm not saying you're wrong, but there are other factors in play.

[Linear Gnome]

One thing I noticed on j-archive is that the contestants have had a lower-than-usual correctness percentage on FJ so far this season, and I've been wondering if that has prompted a change in the aimed-for difficulty for the rest of the season.

[NYCScribbler]

I think they're also going for more potential TOMs instead of flat-out knowledge, and since boardies have been trained to drool when a TOM is given... or something like that... we're getting them at higher rates.

[the_phil]

I think it'd be interesting to compare this year's in-season trend to any previous ones. I only say this because I've noticed that across my "practice" games there's a general trend within any given season. For instance, October/November are almost always my worst months. My best showing came in season 25, when they were my 6th and 9th best months of the season, respectively. January, however, is always a strong month for me - it's been no worse than 3rd thus far.

[StevenH]

I agree that the FJs have been easier in the last month or so. I have found that Jeopardy! boards are like the moon: they go in phases, between both difficulty and the quality of the clues. This has been especially true in the last 10 years or so; before that the boards were more consistently on the easy and well-written side. The difficulty of the FJ clues should soon return to normal.

[Austin Powers]

I agree that the FJs have been easier in the last month or so. I have found that Jeopardy! boards are like the moon: they go in phases, between both difficulty and the quality of the clues. This has been especially true in the last 10 years or so; before that the boards were more consistently on the easy and well-written side. The difficulty of the FJ clues should soon return to normal.

Yes, it's very dialectical, and clearly tied to how the show is trending. The frustrating thing is that the show can trend easier in the middle of a TOC.

[OrangeSAM]

...seriously, you're expecting me to calculate p-values at 3 AM?...

That's some real dedication, man. Or some serious insomnia.