CSO article / AV Test

Let me know when you've got it ready and I'll start a thread on it.

 

Its not about 100%, its about consistency, thats all that matters.

 

You have pretty much summed it up.

 

To the contrary, this only thing this test proves is the proponent's ineptitude.

 

I find it disappointing that the individuals and institutions performing anti-virus comparisons appear to be so blissfully unaware of the basics of the science of testing – especially the concepts of reliability and of measurement error.

Even ignoring any methodological flaws that may be present in this test, the results are questionable in the absence of knowing the degree to which they are repeatable and consistent. For example, internal consistency reliability is an approach that may be applied. In this case, a random sample from the population of malware under consideration (N=36,438 ) is used for the test of anti-virus products, and that sample-test sequence is repeated over several iterations. If the results from each iteration are highly similar (dissimilar), it lends credence (doubt) to the test itself.

Additionally, such techniques allow a researcher to report a “standard error” or an estimate of the precision of the performance of each anti-virus product. Products whose estimates reside within one or two standard errors of each other exhibit “equivalent” performance and are not statistically distinct with a high degree of certainty, for example.

Someday, maybe someone who performs these anti-virus tests will sprinkle a bit a science into the mix?

 

I normally don't bother reading test results,I just like reading the reactions of various users to the results depending on how well/badly their chosen product has done,that can be amusing!
This time however I took a look at this test and what I find amusing this time is the test itself the way its conducted is a joke!
What is "a semi-scientific comparison of antivirus software"?how can something be "semi-scientific"?:-I suppose the author is covering himself from all angles!,does that make the results "semi-empirical"?(semi-accurate!)