Ok, besides using the same typing test each time, what other things should be observed?
Sometimes when I'm taking a test I make a lot of errors, stop and try again a few minutes or hours later. Somehow I feel this is "cheating" and skewing the results. But on the other hand, I could also take a test while being very sleepy or drunk and get a low score and it would not bring anything interesting other than to measure my performance in that particular state. So I guess it's important to know the purpose of the measurement. I came to think about this while testing for my Experiences thread, so that is the purpose.
But should that be about best speed or just "typical speed"? The max speed is maybe not that interesting since it could be a lucky text and also just a speed you can't typically reach. For example, my typeracer max is 92, but I still rarely go over 85. Typical speed on the other hand is not easy to measure, at least for me because I start getting competitive. Then I can get the feeling that "I won't complete this text because the excessive punctuation slowed me down" or "I don't feel like including this result in the mean because it is low". If I get a good score while measuring for my progress graph, I might feel like not doing more measurements that day because it would lower the mean for that day. While the latter is not "cheating" compared to excluding scores from the mean, it doesn't feel scientific at all.
What is a good way to go about this?