Analyst Gene Munster surveyed 45 Apple developers at last week’s WWDC to find out how many developed on other platforms. While the number of developers he spoke with is quite small (considering there were 5,000 registered for the event), there are some interesting results.
Only 7% are also developing apps for the Mac, down from 50% in 2008.
Nearly half (47%) write apps for Android, 36% for Research in Motion’s (RIMM) BlackBerry, 13% for Microsoft (MSFT) Windows Phone 7 and 7% for Hewlett-Packard’s (HPQ) WebOS.
iOS developer survey: 47% are on Android too, only 7% on Mac [Fortune]
This is interesting. But the fact that the study was done on 45 people discredits it completely.
Also, percentages do not make any sense giving that the sample was inferior to 100.
These numbers might be right (though highly imprecise), but they could also be totally wrong.
What is .15 of a developer?
As in 45*0.07= 3.15 developers…
It’s a small sampling, but taken within a specific group. Hoping Munster can aggregate this with more data over time.
I would like to see 1,000 or more.
This also assumes that the attendees of WWDC are an accurate cross-section of developers in general. Is it fair to make that assumption?
There’s no way to confirm that the attendees this year constitute a statistically unbiased selection of developers. A tiny sample from a biased group is unfortunately meaningless.
Not only was the same small, but when was it taken? Did they ask people leaving Moscone right after an iOS session ended? Would the results be different had a Mac-based session ended?
It is completely useless. A sample of a random population of 400 is a minimum requirement for a statistically significant population. This population is nether random nor large enough to draw any valid conclusions.
The sample size is small, but it’s probably enough to see the trend.
I have been a Mac developer in the past, but at the moment I’m exclusively developing for iOS. It has nothing to do with a dislike of the Mac, but where the products I’m working on need to be used.
Asking about history would have been interesting, though. I wonder how far the 7% would have increased?
I disagree with your assertion that the group “WWDC attendees” is meaningless. It shouldn’t be assigned too much meaning, of course, but it’s certainly an interesting collection of iOS/Mac developers.
The point isn’t what % of the public is an app developer. The point is a breakdown of developers themselves.
agreed. it is poor reporting to cite these “percentages” as anything other than inaccurate conjecture.
not peer reviewed.
you missed his point — to be valid, they would have needed to take 400 *random* samples from the developer population at WWDC. nobody is talking about the non-WWDC public.
nope, 45 is not enough to see any trends. its simply too small to publish as anything other than a guess. not peer reviewed statistics.
WWDC has had a high churn rate — in the past few years over half are brand new attendees. Furthermore, the fact that it sold out in a day kinda takes the sampling of the developer community interested in attending WWDC and futher skews it toward those who were quickest on the BUY button (a lot of people were left out).
Furthermore, the app store business model is just now coming to Mac OS, so give it time. It’s relatively easy to create an app for iOS, MacOS not so much, given everything involved in actually making money. But that’s changing. (yay!)
I keep track of development resources, and there are something like 4-6 times more development resources available to programmers for iOS than for Mac OS. It makes sense that more people are following the path of least resistance, the easier business model notwithstanding.
My point wasn’t the percentage of the public. Do you understand the statistical meaning of the word population? In this case, the population are all developers. It’s just way too small a sample to be meaningful. Just like flipping a coin. You might get 10 heads in a row in a population of 100 flips, but that doesn’t mean in a population of 1000 flips you’ll get 100 in a row. In fact, with more flips the chances of another 10 flips in a row becomes less likely.
I wonder if Apple did this kind of survey. (Good luck figuring out if they did.)
Point taken. But were do you pull out the number 400? There is no hard and fast law of statistics that demands that number form the sample size. The question is whether you can have confidence in that number from the sample size. Given that sample size, 7% is little more than noise. But you can be guaranteed that at least *1* person answered that question in the negative. No one is going to be writing books or publishing papers on these numbers. But they are the basis for a more detailed analysis to see if indeed the crossover between iOS and Mac OS is minuscule.
Point taken. But were do you pull out the number 400? There is no hard and fast law of statistics that demands that number form the sample size. The question is whether you can have confidence in that number from the sample size. Given that sample size, 7% is little more than noise. But you can be guaranteed that at least *1* person answered that question in the negative. No one is going to be writing books or publishing papers on these numbers. But they are the basis for a more detailed analysis to see if indeed the crossover between iOS and Mac OS is minuscule.
Peer reviewed statistics? This isn’t “The Lancet”. it’s a informal poll. He could have Jim’s requested 1000 people in the poll and it would still never get a “peer review”.
Note: strangers haggling over semantics don’t amount to “peer review”.
Peer reviewed statistics? This isn’t “The Lancet”. it’s a informal poll. He could have Jim’s requested 1000 people in the poll and it would still never get a “peer review”.
Note: strangers haggling over semantics don’t amount to “peer review”.
This is an informal poll. 7% of 45 is 3 people.
Think about that: 42 people didn’t do Mac development.
Damn right that’s significant outside of a statistics class. Making major decisions based on? Probably foolish. Remarking it’s a trend? Completely reasonable.
And again, the sample was too small to really tell anything. It was just under 1%.
AND yet I put the real fault here on Jim because his article actually endorses this ‘data’ as if it had any real validity. One would expect him to be the first to point out how flawed using such small sample is with more than a passing mention of the full headcount.
No one will be writing books but given that this was a stock ‘expert’ folks might be selling or buying based on this info.