Here's a few question about the format I'd be interested if anyone has any comments about:
Is anyone put off/confused by the amount of numbers and different statistics in the table?
Anything else you would like added? I'm all ears but I want to keep this pretty simple with a low workload so I am unlikely to implement anything much more complex personally.
Does anyone have thoughts on changing the format slightly to the following, which I think is more statistically robust in theory: CTA Race of the Year:
Score: Median (instead of mean)
Tiebreak 1: Mean (instead of median)
Tiebrek 2: Mode
CTA Most Disputed Race
Score: Average Deviation on Median
Tiebreak 1: Average Deviation on Mean (instead of Standard Deviation on mean)
Tiebreak 2: Average Deviation on Mode (instead of no 2nd tiebreak)
Then if races are still tied after all that I would start to look at the 2nd decimal place. I haven't gone through everything but I don't think it would have made much difference. Silverstone would move up to 4th from 6th and the 2nd decimal place would put Japan ahead of Brazil but that's it in terms of Race of the Year.
Yes I am. Personally all I am looking for is the total vote score divided by the number of voters. i.e. the 'voters average' vote.
Maybe the most disputed race etc. could be reserved for a special summer break and end of season special post. Keeping a simple 'voters average' only for the rest of the result posts.
I also find phrases like 'deviation on mode' perplexing and a turn off. I say keep it to a simple 'voters average' I suspect that if you do the voter numbers will increase as it will be more accessible to the average readers that don't happen to be statistic analysts.
Greenlantern101 I suspected this might be the case but let me explain why I have done certain things and what I have or could change. I also know that there are people on CTA who like seeing lots of number crunching and statistics, so there is a compromise to be made.
1) I tried to make the table more understandable by adding the colouring and also the terms score/tiebreak on top of the statistical terms. I could maybe separate the table in to multiple tables for race of the year and most disputed race and not bother with including the stats terms in the tables.
2) The race of the year score, for this season at least, is the simple voters average. However, with the relatively low number of votes we have and the wide range of views, just using this by itself is not really the best way to determine which race people preferred. This is why I've only used the average down to one decimal place and then used the median/mode as tiebreaks. The median (middle vote) itself is less sensitive to being influenced by a few votes with scores a long way from the general consensus... which is also why I would like to use this as the primary score for race of the year.
I'm pretty reluctant to remove the other scores to be honest, but if a lot more people express the same view I can change it. Alternatively I could come up with some way of combining all three in to a single score, but I'm not sure that would really help.
3) I think most disputed race is an interesting concept and one that can give quite a lot of additional insight in to what people were thinking about the race. In fact, it was not my idea initially - it was tooncheese - which again shows the balancing act needed to make this appealing to everyone. As I said, I can separate it from the main table and hide the stats terms. How much people agree or disagree about a result is what we spend most the time talking about though and that is what most disputed race shows, so I'm not really a fan of only updating it a couple of times in the season. I'll try to do a better job of explaining it as well.
Any more thoughts? Basically, if there is a majority that just want one number and nothing else I can do that but otherwise I will do what I can to make everything as transparent and easy to understand as possible.
I like having all the stats avaliable. However may I suggest you update the table after race to show them in order of highest scoring rather than in order of GP. That way it will be easier for the none statos to see whats going on at a glance.
Other things like most disputed can be worked in at leisue but I think a 'leader board' table is essential for the casual viewer to catch the drift of things.
Yes I can change that, my original thinking for not doing that was to keep it for a "big reveal" at the end of the season but I will make separate race of the year/disputed race leader boards as well as the full table throughout the season next year.