Great example of testing/measuring! In the article, there's a simple pie chart leading to a simple question, but without measurement in the first place, the question would never have been asked. ...and after half a dozen replies to the article, it seems like there are already good answers... This is a very compelling demonstration.
Title should be Why people don't install firefox after they've downloaded it.
I was interested in this but 300 people is a small sample in the computer world and I was hoping for a write up on why people are sticking with IE and such after knowing about firefox.
* about 5000 people saw the feedback form (presumably because they canceled the install)
* of that group, about 330 actually submitted something
Strictly as a matter of statistics, 330 is a big sample of the population who canceled. More than sample size, I would be concerned with sample bias; I expect that there is a systematic difference between people who submit the form and people who don't.
I agree with you that it is large for the group but I would be more interested in the people who did cancel as you said. Especially with the example comments like a problem with firefox already running et cetera. I was just was trying to voice my disappointment in the article's given sample.
We would have been delighted with a larger sample size as well (although I think we got some great answers out of the number of responses we did receive).
Of course, the most important factor for our implementing this survey was to not annoy or violate the privacy of people who elected not to participate. So that means not automatically collecting any information.
Note that a "small sample" does not mean the same thing as a "represntative sample". You really care about the latter, and the former will only sometimes skew the latter.
"Sample size is too small, so I don't want to consider X paper" always drives me nuts. A better argument is that since its voluntary, people who are lazy will cancel and never explain and perhaps they are a majority of the user population. Please use that kind of reasoning instead of the knee jerk "300 sounds like a small number".
Neat series. Hopefully they can extend this to be able to get better measurements of people who cancel earlier.
---
Of the group who complained about admin/directory rights, it is my supposition that they are largely at-work users who don't have permission to install Firefox, but don't realize it, or didn't before they started to install it.
