3262: Sports Commentary

Sports Commentary

|< < Prev #3262 (June 22, 2026)

Title text: The plural of anecdote may not be data, but the singular of data is anecdote.

|< < Prev Random Explanation

Explanation

This is one of 45 incomplete explanations: This page was created at a statistically insignificant time. Don't remove this notice too soon. If you can fix this issue, edit the page!

P-hacking is the academically problematic practice of attempting to come up with a question for which the data offers a significant p-value (probability value). This is in contrast to correct scientific analysis, in which a question is formulated clearly and then answered with data. A common way of doing p-hacking is analyzing subgroups to attempt to find significance when the full dataset does not yield statistically significant results; for instance, restricting the analysis of medical data to male subjects to derive a significant p-value when including female subjects would make the p-value insignificant, when the scientific question is largely gender-independent. Sports commentators do a form of p-hacking in which they cite a fact that's made to sound more significant by restricting the situations it applies to. Randall satirizes this with an example in which the restriction uses very specific criteria largely irrelevant to gameplay patterns in order to narrow down the subgroup sample size to a measly two games. Obviously the 0-2 record reflects random noise much more than any significant insight. As well as being irrelevant to gameplay, their p-hacking also makes the game sound like jargon, which can be confusing and difficult to understand. This is ironic given a sports commentator's job is to explain matches and suchlike, rather than making them harder and more vague. P-hacking was also the subject of 882: Significant.

This comic was published 11 days into the 2026 FIFA World Cup. The World Cup was also the subject of 3260: Messi, published the previous Wednesday. Sports commentary was also the subject of 904: Sports.

Transcript

This is one of 28 incomplete transcripts: Don't remove this notice too soon. If you can fix this issue, edit the page!

[Cueball and Ponytail are sitting at a table, looking at the wall behind them. On the wall is a screen showing a soccer field with some unreadable score information above it.(The only readable information is that the score is 2-1.)]

Cueball: They could be in trouble. Over the last 36 years, they've gone 0 for 2 when they've scored in the 37th minute to lead 2-1 against a team whose country comes before theirs alphabetically.

[Caption below comic:]

I wish sports commentators hadn't discovered p-hacking.

  Add comment        Create topic (use sparingly)       Refresh 

Discussion

F1rst p0st! I'll do this explanation. 185.36.194.22 04:32, 23 June 2026 (UTC)

Did this example actually happen? 47.151.65.120 04:33, 23 June 2026 (UTC)

This comic reminds me of 1122: Electoral Precedent and 2383: Electoral Precedent 2020. Generalizing coincidences.

I am not a native English speaker. What does " they've gone 0 for 2" mean? Obviously it cannot be the score, since they are already leading 2-1? Or does this refer to a previous match? And on a more general note, I am really surprised to discover the second football themed comic strip in a few days. OK it's the World Cup, but I always thought that Randall doesn't really care about sports? --92.209.171.90 08:37, 23 June 2026 (UTC)

I am a native English speaker, but it was also a bit impenetrable to me. In part, perhaps, because it was intended to sound impenetrable (as part of the joke). But, even if not, it may be because it's using Americanized sports-talk phrasing that just isn't (yet!) used so much in my more native Anglicised commentaries that I'm used to.

However, I think they're saying that "in the two specific occasions in which all those other conditions occur, they won in neither of them".

A simpler version being perhaps to state that a given team/player has gone nought-for-two in previous matches with their current opponent(s). The results of those contests might have been anything (the winner having gone to 3-2 after penalties, 6-love/6-love/6-love, a par-4 advantage or getting them all out for 178 — depending upon the sport), it's just the win/lose (or win/not-win) count thats "0 for 2".

But this is a case of Overly Narrow Superlative (overlapping with P-Hacking), making it a dubious analysis. Starting with ignoring all the games there are in which a given svoreline was not achieved in a particular minute of play. I think part of this set-up is the difference between Gridiron 'football'/"hand-egg" having tons of points scored, whereas this football (Soccer) often turns on comparatively low scores that (one-nil can be a worthy and entertaining win/loss, and even a no-score-draw might have been fun to watch if your side isn't in desperate need for a win). These commentators, or at least the US audience they're commentating to, are used to spieling things about "the last time they were down on the forty-yard line in the fifth quarter, with two home runs and a shot from the free-throw line in hand..." (look, I know I don't know what they'd really say, to any accuracy, there was no point even trying!), at least to fill in the copious down-time/time-out pauses. (Which isn't actually as easy with low-scoring but more ever-moving 'soccer', where there's often much to be said about current player and ball movements almost all the time; although a five-day international cricket test match(!) commentary on the radio does rather famously lapse into 'filler' like discussing the nice cake that was sent to them by a listener, in the gaps between balls being bowled...)

Sorry, that was a long and convoluted paragraph. (But then, so was the Explanation, before I decided to say this down here. I hope it's been tweaked since then. I'm only really guessing about the Leftpondian commentator-speak being parodied here, and ball-sports aren't really my main interest in the sprorting sphere itself. (But, regarding balls that aren't themselves spheres, I'd happily discuss Rugby League or Rugby Union, and why they're 'better'... though I would totally acknowledge Aussie Rules as a class of its own as far as such contact-sports go.)

HTH, HAND. 82.132.236.84 10:08, 23 June 2026 (UTC)

I'm also English, and it's totally alien to me too. GSLikesCats307 (talk) 11:53, 23 June 2026 (UTC)

I don't know what prompted the rant above, but if you don't care to read it, "going 0 for 2" means having 0 successes out of 2 chances. In the context of this commentary, it's referring to winning 0 games out of the 2 games that meet the criteria. It's not intended to sound impenetrable; it's a common phrasing.163.116.145.33 13:37, 23 June 2026 (UTC)

Worth noting I think that it's a common AMERICAN phrasing. It's hardly ever used in England.

I don't know what prompted you to think it was a rant. It's certainly quite lengthy (in the context of discussion comments here - not in the grand scheme of things), but that's not really the definition of a rant. 82.13.184.33 14:57, 23 June 2026 (UTC)

Because it was lengthy, but it almost entirely ignored the question, focusing instead on your opinion on sports and their commentary styles. Going off on one vs. ranting...it's a blurry distinction. You absolutely did the former; possibly the latter. 0 for 2 (pronounced as a letter "O") simply means zero victories for two games played. It's not obscure terminology. Yorkshire Pudding (talk) 15:50, 23 June 2026 (UTC)

Yes, it is obscure. The only context in which “0 for 2” makes sense to me is in cricket: 0 runs scored, 2 batsmen out. The 0 of / out of / from 2 described above is just not something which would occur to me, and it's largely because of that ‘for’. If it's common, why have I never heard it on (for example) Match of the Day? Randomnonsense (talk) 16:44, 23 June 2026 (UTC)

It's hard to argue that a usage common to the US - the largest English-speaking country in the world - Canada (https://www.google.com/search?q=%22went+0+for%22+site%253Athestar.com), and Australia (https://www.google.com/search?q=%22went+0+for%22+site%3Aabc.net.au, including their coverage of American sports leagues, local sports leagues, international tennis, and international cricket) is obscure. Your personal knowledge is not a good metric for obscurity. 163.116.145.44 18:50, 23 June 2026 (UTC)

Precisely what Yorkshire Pudding said; you could have stopped after your third paragraph, or really even after your second, and you wouldn't have lost any relevant information. I'll add that the rantiness is enhanced by saying things like Gridiron 'football'/"hand-egg", and the very "sportsball"-coded "the last time they were down on the forty-yard line in the fifth quarter, with two home runs and a shot from the free-throw line in hand...," which, apart from their irrelevance to the topic, have a very superior air. 163.116.145.44 16:51, 23 June 2026 (UTC)

Closest match I can find is Germany - Curacao but there Germany took the lead in the 38th minute (not the 37th). I leave the deep dive on Germany's record against teams alphabetically before them when they have taken the lead 2-1 in the 37th/38th minute to someone else...

And, of course, Germany destroyed Curaçao 7-1, just like they did to Brazil (which is also alphabetically before Germany!) 12 years prior Wilh3lm (talk) 12:38, 23 June 2026 (UTC)

As I recall, the Germans had already scored 7 before the Brazilians scored, and quite a few people independently came up with Ger-many Braz-nil… Randomnonsense (talk) 16:48, 23 June 2026 (UTC)

I did a limited look into possible edge-cases, but didn't get anywhere so far as to confirm all minutes-of-goals for those. Concentrated on (fairly recent) games with more than a 2-2 final score, as the requirements are that they must have at least been 2-1 and then at least got levelled up by the lagging side, before the possible tie-breaking penality shootouts.

I suppose I could have first narrowed it down to every game with a reported 37th minute goal (given the rarity of that exact event, by apparently common agreement) then tallying up the precise game-state at that time, plus the final result. But I was looking into the other fine details first. 81.179.200.152 23:14, 23 June 2026 (UTC)

Don't assume "bad" p-hacking without looking closer at the data: Sometimes what looks like P-hacking is really finding previously-unseen patterns. If you have a drug trial on a drug that you have no reason to think will show gender differences and you are asking "is this drug better than existing drugs" and the results are inconclusive, then you do "p-hack" subgroups and find that in males between the ages of 18 and 50 it demonstrates superior results, you MAY be cherry-picking results or you MAY have found a hidden pattern. Assuming your sub-group size isn't ridiculously small, you can legitimately claim that you need more funding for a follow-up study or at least a follow-up analysis of this subgroup in previous studies of the same drug. 150.221.155.241 13:35, 23 June 2026 (UTC)

Yes, more data is usually the solution. The comic deliberately uses an extremely small dataset. You can make up almost any hypothesis and find 2-3 datapoints that fit it. Barmar (talk) 14:24, 23 June 2026 (UTC)

Or, perhaps more commonly, if you have a sufficiently large dataset you can mine through it and come up with two or three interesting-looking 'hypotheses' that it'll appear to support, even if you didn't have any to start with. 82.13.184.33 15:02, 23 June 2026 (UTC)

The difference is that taking a lot of data and looking at many possible patterns to that data is likely to reveal artefacts of mere chance.

Considering a single possible pattern and looking to see if it is justified is (even if, as an individual pattern, exactly as much a coincidental artefact or not) useful. So one might legitimately be able to suggest that the male subjects may show a legitimate long-term improvement to a treatment (possibly either because it works better with male hormones, or because there aren't the same underlying hormonal variations across each month, or just because it works better for different body-mass/fat distributions that are more typically male than female), or perhaps vice-versa (that it actually works more usefully for females, again for such reasons).

But filtering inconclusive results through many possible sorting algorithms and extracting 'nuggets' of apparent significance devalues those nuggets. Especially if it gets more complexly combinatorial. Imagine the number of criteria you might have considered, together and individually, to establish anything like "the treatment was twice as effective as the control treatment in males less than 30 with an older brother and females over 38 whose job is in education". How you'd even rationalise/explain such a complicated cause/effect relationship is one thing. That you've probably discarded so many other datums you had (out of hundred subjects, you're decided to ignore maybe half the people because of being under-/over-age, in this gendered thing, cutting back far more than that for the family/employment requirements and then of the remainder (probably somewhat <10, I'd guess, depending on where you actually got the initial subject-pool) and then, for such a comparison to stand up, the non-discards must be further split between on-treatment and whatever off-treatment/old-treatment you're comparing to, is an issue of practicality which is the other issue. (At least until you come up with a decent reason to recruit these subsets explicitly for a follow-up study, assuming the relationship doesn't just vanish/revert-to-the-mean when you do as it was pure chance after all.)

Realistically, you should weight the apparent significance of 'results' according to how many results you actively looked for. (e.g. had fifty possible things, divide the significance of any positive 'match' by 50. Although I can see reasons why division by 50**2 or ln(50) would be better, depending upon the scenario relevant to this kind of siftingt.) But better just to avoid that.

Which might be a big problem in AI-derived research results. You're asking an algorithm to make an uncounted large number of separate interpretations (including of exactly what question it is that you're asking) and then it's returning the 'nicest' results that seemingly fulfil its mysterious 'training' insofar as they aim to be the single/several results from the largely randomised treatments that are judged to be the least incomprehensible. 81.179.200.152 20:00, 23 June 2026 (UTC)

The saddest part of all this is that Randall, in the title text of this comic, might be one of the last persons in the English-speaking world to recognize that "data" is a plural noun, the singular of which (icymi) is "datum". 147.81.27.244 15:38, 23 June 2026 (UTC)

I can assure you that Randall knows full well how "data" operates and that's part of the joke. If all you have is a single data point, you can extrapolate whatever you want from it; which makes it no better than noise or anecdote. 74.202.210.170 19:05, 23 June 2026 (UTC)
        Add comment