Good call Brian. I ran the regression again, this time omitting the first two (high) data points. The result ... http://www.heimhenge.com/downloads/regression-2.jpg is indeed a horizontal line for best fit. Curiously, the correlation goes to zero now (at least to the two significant figures computed). That tells us that what we have is essentially a random data set. Not surprising for weather trends. Still, it does seem like I cancel observing more often these days. It would be interesting to run this analysis using data for the Maricopa County area, where all of my cancellations happened. Dan Heim DFAC President http://www.dfacaz.org ----- Original Message ----- From: "Brian Skiff" <bas@xxxxxxxxxx> To: <az-observing@xxxxxxxxxxxxx> Sent: Wednesday, June 13, 2007 2:31 PM Subject: [AZ-Observing] Re: Flagstaff nighttime cloudiness/sky brightness > >> Granted, the correlation is a shaky 0.09.... > > ...and if you fit a horizontal line (i.e. no change in cloudiness > since 1980 on average), I'll bet the correlation coefficient is > also something like 0.09, so the declining slope and zero slope > have the same (non)significance. Basically, take Dan's plot > (http://www.heimhenge.com/downloads/regression.jpg) and hold > your finger over just the first two (high) points, suddenly it > looks like the best-fit line is a horizontal one. > About my only misgiving about this stuff is that I didn't start > noting it the day arrived here in 1973. > > \Brian > -- > See message header for info on list archives or unsubscribing, and please > send personal replies to the author, not the list. > > -- See message header for info on list archives or unsubscribing, and please send personal replies to the author, not the list.