[uday]

I recently stumbled across an old ACBL survey from 2005 that claimed that the USA hosts 25 million bridge players.

Yeah, me too. So we decided to run a quick Google Consumer Survey, asking just one question

"Do you know how to play Bridge, the card game?"

This survey seems to claim that 12.4 % of the sampled population ( +/- about 3% ) knows how to play bridge. The population sampled were US adults (supposedly) with Internet access

I don't know enough to know how valid this survey is.

So, opinions on the methodology and results?

References

Results of Google Survey

Original ACBL survey (PDF)

Nate Silver on which polls fared best in the US presidental election

Pew Research Center about google surveys

[hrothgar]

Comment 1: Did the Google Survey report confidence bounds? If not, this would be more than a bit worrisome)

Comment 2: The Pew article seems to provide a good description of the pro's and con's of the approach.

As with any survey, the big question is "How accurate do you need the information to be?".

Last comment (and this is an important one).

Kevin Lane posted some very interesting information in a recent thread on Bridgewinner. (I am attaching that post in its entirety at the tail end of this one)

Simply put: The number of bridge players in the US is going to decline precipitously over the next 10 years.
Any value that you derive via a survey today is going to be inappropriate for any kind of long term planning.


----------------

Hat tip to the ACBL's Melissa Donehue who provided me some anonymized raw data. The following represents an age breakdown of current ACBL members who earned masterpoints in 2013 (about 65% of members). Disclaimer: the data isn't perfect but the imperfections don't seem to skew the results:

10% decile = 60
20% decile = 65
30% decile = 68
40% decile = 70
50% decile = 72
60% decile = 74
70% decile = 77
80% decile = 80
90% decile = 84

Median = 72
Mode = 71

—And for those wanting a breakdown of the youngest 10%:

10% decile = 37
20% decile = 46
30% decile = 50
40% decile = 53
50% decile = 55
60% decile = 57
70% decile = 58
80% decile = 59
90% decile = 60

Again, these are players who EARNED MASTERPOINTS in 2013.

[uday]

The survey says +/- blah, i assumed (no good reason) they were using the standard 95% confidence.

I'm mostly interested in whether there really are a huge mass of non-duplicate players out there, who we can serve with simple games like the couple we hang off our website.

The 10 year problem is real, I suspect. It isn't like i know what to do about it, though. Maybe a good start would be to find a way to get the more casual players to come play simpler web games with less of a fuss.

[Rain]

The google survey claims there is a large group of young people 18-24 who knows how to play bridge.

Do you have any theory why that is so?

[hrothgar]

Can the Google survey folks check for specific cookies?

As I recall, BBO places some cookies on local machines.
The Google survey folks scan for cookies and use this to make demographic assumptions.

I'd expect that the number of self identified bridge players who lack the BBO cookie would be a useful data point...

[helene_t]

It is peculiar that young people are more likely to know how to play bridge than older people are. I wonder if young people are more likely to overstate their proficiency. Some may have said "yes" on the basis that they have a vague idea about the objective and mechanics of the game.

But at least it is good that there is an "uncertain" option. You know the old joke:

Q: Can you play the piano?
A: I don't know, I never tried.

BTW men are more likely to know how to play bridge. For the youngest age group the difference between the genders is huge (28% vs 8.7%).

[WellSpyder]

hrothgar, on 2014-March-06, 09:05, said:

Hat tip to the ACBL's Melissa Donehue who provided me some anonymized raw data. The following represents an age breakdown of current ACBL members who earned masterpoints in 2013 (about 65% of members). Disclaimer: the data isn't perfect but the imperfections don't seem to skew the results:

10% decile = 60
20% decile = 65
30% decile = 68
40% decile = 70
50% decile = 72
60% decile = 74
70% decile = 77
80% decile = 80
90% decile = 84

Median = 72
Mode = 71

—And for those wanting a breakdown of the youngest 10%:

10% decile = 37
20% decile = 46
30% decile = 50
40% decile = 53
50% decile = 55
60% decile = 57
70% decile = 58
80% decile = 59
90% decile = 60

Again, these are players who EARNED MASTERPOINTS in 2013.

Those are pretty dramatic figures, hrothgar. To look at them another way:

- only half of acbl members who earned masterpoints in 2013 were aged 72 or under
- only 1 in 5 was aged 65 or under
- only 1 in 10 was aged 60 or under
- only 1 in 20 was aged 55 or under
- only 1 in 33 was aged 51 or under
- only 1 in 50 was aged 46 or under
- only 1 in 100 was aged 37 or under

I knew there was concern about the age of membership of the acbl, but I didn't know quite how precarious the situation was. I wonder whether there is any way we could get comparable figures for England, where I play? My impression is that it isn't as bad as this, but that may be biased because I mostly see the population of tournament players rather than club players.

[jogs]

The median age of those who earned a masterpoint last year was 72.

What was the median age of members of the ACBL?

[hrothgar]

jogs, on 2014-March-06, 10:39, said:

The median age of those who earned a masterpoint last year was 72.

What was the median age of members of the ACBL?

The information that I provided is the only such demographic information that I have available.

[barmar]

I think "earned masterpoints" is intended as a proxy for "actively plays bridge". Clubs only report masterpoint winners, not everyone who played in the games, so it's the most complete information they have. Since most club games and tournaments are stratified, even poor players get points once in a while, so masterpoint winners is not a bad approximation of active players.

Statistics about people who are members but never actually play don't seem very important. They're probably also a tiny fraction of the membership -- how many people would keep paying ACBL dues if they don't care to play? So I doubt the difference biases the statistics significantly.

[dustinst22]

The major problem with a survey like this is that I think you will get a higher response rate amongst those who actually play.

I imagine when you ask a random survey question to a random Internet user who isn't forced to answer, those who know the subject will be more willing to answer the poll. Especially if it's a niche subject (in this case Bridge) it will attract more attention to those who know the subject. I think the only way to get an accurate representation for something like this is to somehow "force" every person asked to answer.

On the other hand a poll asking an opinion on who will win a presidential election is likely going to get a fair representation, thus making the Google consumer survey very accurate for those types of polls.

[Vampyr]

helene_t, on 2014-March-06, 10:00, said:

BTW men are more likely to know how to play bridge.

Hugely. Looking around the room at a club or tournament confirms this.

[Bbradley62]

Rain, on 2014-March-06, 09:48, said:

The google survey claims there is a large group of young people 18-24 who knows how to play bridge.

Do you have any theory why that is so?

I think this is because a fair number play some when they are in college, then never play again. At age 20, they'd say "yes, they know how" but at age 30 they'd say "no".

[Bbradley62]

I'd really like to see the 2013 ACBL data next to the 1983 (or so) ACBL data. I think predictions of impending doom are not valid because ACBL has a long history of being able to recruit recent retirees as new members; with the Baby Boomers now reaching that age, I'd actually expect to see an increase 10 years from now over the current membership totals.

[barmar]

dustinst22, on 2014-March-06, 17:44, said:

The major problem with a survey like this is that I think you will get a higher response rate amongst those who actually play.

I assume Google factors this into their confidence level.

Quote

I imagine when you ask a random survey question to a random Internet user who isn't forced to answer, those who know the subject will be more willing to answer the poll. Especially if it's a niche subject (in this case Bridge) it will attract more attention to those who know the subject. I think the only way to get an accurate representation for something like this is to somehow "force" every person asked to answer.

I don't think I've participated in a Google Poll, but I'm guessing you don't get to see the question before agreeing to participate.

[Zelandakh]

As with any survey the first question should be how it was done. For example, if you did a random survey on BBF with this question you might discover that more than 12.4% of respondents know how to play bridge. Indeed, as someone who currently works for a market research company I would say that the chances of this survey matching industry standards are pretty low. And that is in a field where some surveys already have pretty low quality (see the WC election thread for some examples of this). However, I could be wrong - it is possible to produce meaningful data via the internet if you are careful about it.

[helene_t]

My guess would be that they have thought quite carefully about how to avoid biases. They don't poll people entirely randomly but stratify on the basis of websites visited, cookies and IP addresses. And as Barmar said, don't disclose the question or topic before people have agreed to participate.

That said, the concordance between inferred sex/age and reported sex/age is not that great so one could question the quality of the stratification.

[gordontd]

helene_t, on 2014-March-06, 10:00, said:

BTW men are more likely to know how to play bridge.

Vampyr, on 2014-March-06, 19:14, said:

Hugely. Looking around the room at a club or tournament confirms this.

Looking around the room at tournaments, or at the clubs where you play, may support this idea but if you looked at the memberships of all bridge clubs in England you would find more women than men playing.

[helene_t]

Vampyr, on 2014-March-06, 19:14, said:

Hugely. Looking around the room at a club or tournament confirms this.

Well I actually don't think it is true. It may be true in some countries in Eastern Europe where bridge is more a male thing. As I said the difference between the genders, according to the poll, are most pronounced among young people. This may be true. I just wonder if it could be a poll artefact.

[Trinidad]

I printed the data on ACBL masterpoint winners. It is a great way to postpone my midlife crisis.

Rik