I was just looking at W3Schools and Wikipedia for browser share statistics, and they report drastically different numbers for IE, especially. I understand that for all surveys, there will be some discrepancy, but the IE numbers are very different here. For September 2009, W3Schools reports: IE7: 15.3%, IE6: 12.1%, IE8: 12.2%. Wikipedia, on the other hand, reports, for September 2009, that IE has 64.66% market share, with higher values for each. Is that a significant difference that cannot be explained by margin of error issues? What's the issue here?
Tell me more
×
Super User is a question and answer site for
computer enthusiasts and power users. It's 100% free, no registration required.
|
|
||||
|
If you scroll down W3Schools says this is just the data collected from their site.
Wikipedia linked to the counter who does not say how they got their data. But I assume it is from all the sites they monitor traffic stats for. They seem to be a pay service. So it seems to depend on the sites being logged and they time of users they attract. The more tech savvy the lower the IE numbers will be. |
|||||
|
|
Getting "Real Statistics" is hard to achieve since many things come into play when determining accurate data. John Resig has a good article on his blog called " Determining Browser Market Share". It explains this topic better than I could here. |
|||
|
|
|
There's numerous problems with trying to figure this out. Usually, the data is collected from what the browser claims to be, and browsers are not always honest. It isn't as much of a problem anymore, but for a while there were quite a few websites that required certain browsers (mostly Internet Explorer) and enforced it with the user-agent field, so a lot of lesser-used browsers are set up to lie about what they are. For this reason, IE will tend to be overreported. They're also collected from hits, but one hit isn't necessarily one person. Many people do their browsing using caches at some level (perhaps their company, perhaps their ISP), and so one hit on the website with one browser could result in lots of people reading that page on various browsers. The effects of this are probably random, although it may be that certain browsers have a greater tendency to be used behind proxies and caches, and hence are overreported. There's also the question of who's using the website, since different websites have different demographics. I'd expect Linux websites to get fewer Internet Explorer users, while websites devoted to things like web design will get users who think about their web browser, and don't just use what's convenient, so they'll get more odd browsers and less IE. The average computer user will use what's there, which will normally be IE. Bear in mind that most sites interested in tracking browser types will tend to attract the more computer-literate users, and therefore will probably underreport IE usage. |
|||
|
|
|
I recently published some browser statistics for two NON-technical websites that might give a more accurate picture of what the average user is using. Check it out here:
http://www.blog.ofitall.com/development/browser-stats-for-non-technical-websites/ |
|||||
|
[damn-lies]tag. :) – quack quixote Oct 29 '09 at 2:05