4 ms·
If there are enough data points (which it seems like there is), and each UI is chosen a random 1/4 of the time, then it will be very easy to isolate and determi
by rubergly 16y ago
If there are enough data points (which it seems like there is), and each UI is chosen a random 1/4 of the time, then it will be very easy to isolate and determine the effect of the slider UI. Problems would arise if the slider was chosen 60% of the time in Firefox, but only 30% of the time in Opera, and if Opera had a 20% higher population of hackers, who had a distribution centered around 6 as opposed to the aggregate population centered around 5; as is, it sounds like there's plenty of random sampling solving these complications.