Caution: statistical misinformation

    Have you ever wondered how big the so-called statistical error can be? Especially - on a global scale. Especially - with seemingly reputable and repeatedly cited agencies or online services. I will try to show you how sad everything is. Naturally - on the example of browser usage statistics (who knows me will understand why). I hope that after reading this mini-study you will be more critical of the next high-profile statements of various publications about the market share of a particular program.

    So, let's begin. For example - from a fairly well-known and respected resource W3schools.com . According to this source, over the past year, the share of Opera browser in the world grew by one percent and amounted to 2.4%.

    image

    At first glance - excellent performance, a noticeable increase in the user base. However, I was never convinced by W3schools.com data, and here's why.

    In fact, the statistics on W3schools.com are also strongly, to put it mildly, inaccurate, as well as on all other similar "import" resources that are threatening world statistics. The fact is that as a source data, the log of visits to the W3schools.com site itself is used. At the moment, according to Alexa.com , the audience of visitors to this resource by country looks like this:

    image

    Neither Russia, nor any of the countries of the former USSR is even included in the TOP-25 list of visitors to W3schools.com. Meanwhile, it is we who make up the lion's share of users of Opera browsers. Surprise, right?

    However, the statistics of W3schools.com, one can say, are still “favorably” different from, for example, the well-known service HitsLink , which stubbornly does not want to give Opera more than 0.71 percent.

    image

    Everything here is also very simple: for the most part, HitsLink only tracks American resources (that is, it actually equates America’s indicators with global ones), and at least W3schools.com has Germany, India - countries in which Opera has a certain weight.

    Another proof of the fallacy of HitsLink data can be obtained from the browser developers themselves. For example, Google recently announced that the number of Chrome browser users has exceeded 10 million . At the same time, Opera Software in the third quarter of last yearvoiced the figure of 30 million users of the desktop version, not counting 21 million users of Opera Mini (which, incidentally, often falls into the statistics on desktop browsers - the kernel running on a remote server is the same as in the usual "desktop" version of the browser ) But what do the HitsLink numbers say? As you can see, Chrome - 1.04%, Opera - 0.71%. These are the incidents.

    Even more interesting is the data from the GetClicky service , which offers such a “global” balance of power, in which the Opera browser is inferior even to some mobile browsers (being at the same time a recognized leader in mobile web surfing):

    image

    Of course, one could simply not notice such statistical “oversights” and quietly and peacefully use one’s favorite browser, if not for one “but”: the media often use these statistical calculations in their news or analytical articles, while operating with loud the terms "global market share", etc. etc. As a result, the average person gets the impression that Opera is an eternal "loser", not worthy of special attention. What is the result? Fewer people will decide to try this really good browser, the user base is growing much slower, and web developers continue to ignore Opera when creating their "masterpieces", thereby creating problems for me and tens of millions of other Opera users.

    Actually, that’s why you are now reading this article;) If, however, someone wants to at least roughly determine the real balance of power in the browser market, I can advise Wikipedia - at least they tried to approach the issue from different directions, so do it yourself calculate the percentage is not difficult. But, I repeat, this is not the ultimate truth, but only a very approximate picture of the correlation of forces.

    Also popular now: