> What I'm doing right now is actually averaging the betas from the lags I'm most
> interested in from the tent model, and thus getting a number that indicates the
> average level of activity over those lags. Is this approach susceptible to the
> problem of negative betas, or were you talking about something else entirely?
Yes, the approach is susceptible to the situation with both positive and negative betas, and also it won't be able to differentiate two curves with the same average but different shapes.
Gang