Posts belonging to Category stat attack

Fun fact I have just discovered

I was just writing the comment for Interceptor in the Gameshow General Election (yes it is in the Top 30) and discovered the following startling fact:

Sean O’Kane is only 45.

Come on ITV, I think you understand what I’m saying here…

Stat Attack on The Krypton Factor Part Deux

Just before I start, I’m disappointed that last night’s Minute to Win It is currently nowhere to be “found”. Sort it out, America!

Also I can’t read the lyrics to our Eurovision entry (“You bring the sunshine, I’ll bring the good times!”) without adding “…and I’ll get to Scotland befoooorrre you!” in my mind. I largely suspect this is just me now but you’ll all be doing it by May.

Also also, it’s the last night of Man in Box this evening, the thing that literally nobody’s talking about. I am interested in what happens tomorrow with the reveal, though.

Anyway, you might remember me doing a post before this year’s Krypton Factor attempting to lay to rest the idea that the general knowledge round is rather unfairly weighted and came to the conclusion that although it might look like that sometimes, generally speaking the numbers suggested otherwise – it gives the impression that anyone can win (and that is, in fact, the point), but the person leading going into it still usually wins.

That was based on KF 2009 numbers, where the show was five rounds and the GK round was 70 seconds. This year there were only four rounds and the length of the GK round had increased to 90 seconds. Now is it properly game changing?

Here are the numbers, numbers in brackets are the scores at the end of the round. The first heat ended in a tie, I’m not including the result of that for statistical purposes.

Heat 1: 26 (30), 16 (16), 20 (30), 10 (22)
Heat 2: 22 (23), 16 (16), 14 (19), 16 (25)
Heat 3: 10 (16), 14 (23), 14 (14), 30 (36)
Heat 4: 18 (27), 18 (23), 14 (14), 16 (15)
Heat 5: 16 (26), 18 (23), 14 (12), 22 (22)
Heat 6: 10 (18), 22 (20), 20 (29), 18 (22)
Heat 7: 16 (16), 18 (20), 22 (26), 12 (11)
Semi 1: 18 (20), 10 (13), 20 (21), 18 (25)
Semi 2: 16 (20), 14 (21), 14 (14), 26 (38)
Final: 20 (19), 18 (27), 14 (26), 14 (22)

What does this show?

  • Averaging (mean) scores on a like-for-like basis (all the best scores, all the second best scores and so on) reveal the averages for the round to be 9.3, 6, 2.9 and (hilariously) -0.6. Rather closer to the traditional round scores than last time. Total of averages is 17.6, so still below the amount given in other rounds. On three occasions a player scores more than the 10 points on offer in other rounds.
  • More interestingly, the leader going into the round only wins five out of the ten shows. The average lead going into the round is 5.2. The lead of the five players who go into the round as the leader averages at 7.2 (this includes two rather large outliers of 10 and 16), but leads of people who don’t go on to win is only 3.2 which seems an entirely reasonable turnover.
  • The most interesting result is the one for the final in which loads of points are scored, but if you were to convert the performances into 10, 6, 4 and 2 would still give broadly the same result (in fact Pete and James would tie on 24 points each).

What does this all prove? Beats me, only that the current system used is more ‘alright’ than ‘not alright,’ and this year’s series had me feeling rather bored which definitely wasn’t the case last year. So let’s instead look at the New Zealand 1990 KF final and possibly the dullest Mental Agility round I can recall seeing. Your ten seconds of concentration starts… now!

Stat Attack

That’s right! Not only do I like gossip and pointless conjecture at Bother’s Bar, on occasion I actually do some hard research as well.

Final rounds of gameshows are difficult to perfect – you want to be fair, but at the same time you don’t really want to disenfranchise the people lagging behind – miracle comebacks are good telly after all.

I was never a fan of “let’s double the points!!!!” because it doesn’t really solve a problem – sure, you make it easier to catch up. On the other hand, you make it just as easy to pull away, and chances are if they were doing better than you before they’re going to be doing twice as good as you when it really matters.

The Krypton Factor  was always slightly different in that it tests ability in a number of different disciplines. Most of these events are scored the same way (usually 10, 6, 4, 2). But the final round, General Knowledge plays slightly differently: a number of questions on the buzzer against a time limit with points on for answering correctly and points off for answering incorrectly. Incorrectly answered questions are not thrown over. The points and time limit have varied wildly over the years, but with the potential for one person to blast everybody out, is there too much weighting towards it?

So I thought we’d take a look at some stats from the 2009 series (we will do the same at the end of 2010 and compare). In this year, 2 points are awarded for a correct answer, -1 point for a wrong answer. 70 seconds is the time limit, which in practice meant between 9 and 14 questions.

The scores given are before general knowledge and in the brackets the final result:

Ep 1: 26 (31), 20 (21), 18 (17), 28 (30)
Ep 2: 22 (22), 26 (32), 20 (21), 24 (26)
Ep 3: 30 (32), 28 (30), 22 (29), 16 (20)
Ep 4: 26 (30), 12 (14), 18 (18), 34 (37)
Ep 5: 26 (26), 22 (26), 28 (30), 14 (23)
Ep 6: 30 (30), 20 (20), 20 (28), 22 (24)
Ep 7: 26 (27), 20 (23), 16 (18), 30 (36)
Ep 8: 24 (30), 10 (12), 22 (21), 36 (36)
Ep 9: 22 (31), 30 (32), 18 (19), 24 (23)
Ep 10: 14 (15), 20 (20), 22 (22), 34 (47)

Doing a little bit of number crunching shows up some quite interesting things:

  • Given that other rounds are typically scored 10, 6, 4 and 2, if you were to average the best, second best and so on scores for GK in 2009 it would be worth 7.3, 2.5, 1.1 and 0.
  • That equals 10.9, a far cry from the 22 points on offer for all the other rounds.
  • The person leading going into the round ended up winning the contest nine out of ten times, but on only four occasions did they actually score the highest on general knowledge.

To me, this leads to a surprising conclusion that far from being weighted too heavily, general knowledge wasn’t weighted enough compared to other rounds of the contest. Whilst on only one occasion did a contestant score more than the 10 points on offer from other rounds, those further down lose out more.

I will return to this at the end of the 2010 – similar scoring but a longer 90 second time limit and seemingly faster questions. It’s also the fourth round of four rather than of five or six as it used to be.

If anyone wants to compile statistics from older series as shown on Challenge, go on right ahead.