[waivek]

Take any App on the store. Count the number of reviews. Then count the total number of installs.

The total reviews will be less than ten percent of the installs. The negative reviews are a fraction of that percentage.

As such you cannot observe via reviews what the majority of the users think of the app.

Then, by the definition you have just given, app reviews aren't empirical.

[hungerstrike]

Incorrect again. The fact that you can go onto any apps review history and see evidence that people are unhappy with updates that break their stuff is exactly the definition of empirical evidence.

Sorry, but none of your badly formed, hand-wavy reasoning has proven that wrong. Also, nobody is arguing that "the majority" think something - I'm arguing against your completely anecdotal and un-evidenced claim that it "rarely" happens.

What evidence do you have that it rarely happens? None that I can see so far...

[waivek]

How can app reviews be used to draw conclusions of how users feel about the app if less than ten percent of the users write app reviews?

It's like surveying a small percentage of an electorate by asking them who they'll vote for for and deciding the result based on the survey.

[hungerstrike]

Actually, 10% of a population is a damn good sample size for just about any scientific or medical study. It’s quite large to be honest.

Anyway, once again... where is your evidence that this rarely happens?

[waivek]

It's a biased sample because people who are fine with their app don't write reviews. The sample is polluted.

My evidence can drawn from the fact that ten percent of people write reviews. A fraction of that are negative. If an occurrence happens say five out of a hundred times,that's a rare occurrence.