Early statistics caveats

[RyanChristiansen]

The following paragraphs appear in the Research Notes for a book that I'm wrapping up. Does anyone else have any thoughts about caveats that should be mentioned about statistics from old newspaper articles?

It’s worth noting that sports-writing during and before the early 1930s provided different, less exact information than it does today, especially in terms of statistics. Sportswriters and team statisticians had no film to watch later to double-check observed yardage gained, and while one source might give a runner two yards on a play, another source might give three yards. Observers typically rounded up for their own players and rounded down for opposing players. Depending on weather and field conditions, the disparity between observed yardages might be wider. Whenever possible, I deferred to using information from sources local to the home team.

Also, because the forward pass was relatively new in importance, some observers awarded the entire distance of a pass from where the passer stood behind the line of scrimmage to where the receiver caught the ball, while others awarded the distance of the pass from the line of scrimmage to the receiver. Some tacked on the receiver’s yards gained after the catch, but not all observers awarded that yardage. Some broke down a pass into pass yardage and run yardage. Modern practice awards the distance of the pass from the line of scrimmage and includes the yards gained after the catch. Whenever possible, I share passing yardage gained using the modern practice.

Sometimes, accounts of the same game in different newspapers awarded touchdowns and other scores to different players. Accuracy depended on knowing who was whom and being able to read numbers on jerseys, if they had them at all and if they were clean, and if the players wore their correct numbers. In these situations, I deferred to using information from sources local to the home team.

In rare instances, newspapers shared the events of a game that were recorded second-hand from firsthand observers who relied on memory, which is very fragile.

A key observation I can share about newspapers and statistics is that early on, when football was primarily a running game, most articles only mentioned the final score, starting lineups with last names, and how many first downs each team gained. As football progressed to include more of the passing game, fans grew more interested in the game, and newspaper articles began to share more statistics.

[rhickok1109]

I've already mentioned this in another thread, but it might be worth repeating...

Newspaper accounts tend to mentioned extremes. For example, an account might tell you that a runner "galloped" for 18 yards or lost 3 yards on a crucial 3rd-and-1 but all the 3- and 4- and 5-yard gains are ignored. Or they'll mention a long completion or an interception without mentioning shorter passes and incompletions.

[TanksAndSpartans]

This is just meant to help, not criticize, but I would cut this paragraph:

A key observation I can share about newspapers and statistics is that early on, when football was primarily a running game, most articles only mentioned the final score, starting lineups with last names, and how many first downs each team gained. As football progressed to include more of the passing game, fans grew more interested in the game, and newspaper articles began to share more statistics.

I'm working on an 1890s project and I can say some Pittsburgh and Westmoreland county newspapers did play-by-plays every week. And it was only a running game at that time. Also pre-NFL, Ohio League, at least one of the Canton papers did play-by-plays for big games. I don't think it's all about the passing game. Some cities were super interested in their teams even before passing was allowed. Not only would the games be covered in detail, but there was "gossip" during the week. Just my opinion - not saying the idea that the passing game increased popularity is incorrect, I just personally wouldn't tie it to everything. Just my 2 cents.

I like the rest of it. Good points. One thing I'm struggling with on stats, off the top of my head, is them not differentiating between punt return yardage and rushing yardage, stuff like player x had runs of 40, 52, and 60 yards, then you read the play-by-play and learn two were on punt returns . Another thing I wondered about is it really wasn't the papers job to be stat keepers - I think I read somewhere the pre-1932 NFL stats were just "lost" - I don't think that's true, but I think I heard it someplace.

[RichardBak]

I agree with Tanks re cutting that one particular graf. I've seen some game accounts from 1890s-early 1900s giving an almost blow-by-blow account, even when the action consisted almost entirely of runs and punts.

[JeffreyMiller]

I'm going to be the negative nellie here ... I think trying to reconstruct reliable stats from the league's first decade (or before) is an exercise in futility on many levels. Firstly, trying to use partial stats to paint some sort of accurate assessment or quantification of a player's worth is both unfair and misleading. What if the three or four games you are missing from a player's particular season are his worst? Secondly, You have to come to some sort of hierarchy when it comes to determining which paper is the one from which you might wish to cherry pick your stats. Will it be the one that says Thorpe's run was 55 yards or 50 yards? Or will it be the one that calls his gain a run or a forward or a return? Third, there were some teams that had very little coverage as opposed to others (such as Canton, Providence or New York) which had in-depth accounts, meaning some players from some cities will have more comprehensive stats than others (which is why I don't like the All-Pro teams of the early years, since some of the writers didn't even see every team in the league that year!).

[TanksAndSpartans]

I think there's value in research that attempts to construct early statistics, but it's like any other research - you don't want to over reach with your conclusions, you want to compare apples to apples, and you want to document all of the limitations you mention which are all valid and nicely explained. This was an article I found particularly good that pulls it all off:

http://www.profootballresearchers.org/a ... 02-396.pdf