I have been looking at a long-running behavior score experiment and the communication score part got me confused. Some people say it only moves by 100 per summary, but the numbers in the experiment seem to point higher than that.
What are people seeing in practice, and how should those score changes be interpreted without overreading the graphs?
StunTicker:
Yeah, the pixel counting part makes it feel more like a rough estimate than a perfect measurement. The trend matters more than the exact number.