Since there were no seismographs (and, even better, no seismologists!) around much more than a century ago, our evidence about how often Cascadia subduction zone earthquakes occur comes from what a lawyer might call "circumstantial evidence". Mostly, from tsunami deposits--characteristic sediments left behind when high tsunamis ran inland and left a layer of sand behind. Also from coastal tree stands in which the all the trees died at once, due to being submerged in saltwater as a result of deformation associated with a big event (and which we can date precisely using tree-rings).

Chris Goldfinger, at Oregon State has shown that Cascadia earthquakes also cause submarine "turbidity currents" that are essentially mudflows that cascade down offshore canyons, and that can be sampled using sea floor coring devices, and the events dated. The size, or magnitude, of these ancient events may be related to how widespread the shaking was (and thus how widespread their resulting turbidite deposits are found). He's established a chronology of Cascadia subduction events that seems to indicate that the subduction zone is "segmented". He interprets the data as evidence that the southern portion tends to go off in relatively smaller earthquakes (M8-M8.5?) relatively frequently. But about half as often the entire margin breaks (like it did 310 years ago), from Cape Mendocino to Vancouver Island, in a walloping M9 earthquake. From a seismologist's perspective, I think these findings are all logical and make a sensible story. But since I'm not a geologist I retain a skeptical attitude about just how well one can identify the turbidites arising from individual earthquakes in different submarine channels.

Ok, now for the probabilities. Yes the accepted 50-year probability of occurrence for a Cascadia Megaquake (sounds like something Hollywood would invent, no?) is generally regarded as 10%-15%. You could do worse than saying that if in any 500 year period I'm likely to have one M9 earthquake, then in 10% of that time interval (i.e. 50 years) I've got 10% chance of an M9. I know I know you think that a 50 year period starting at 450 years after the last earthquake will have a higher chance than one starting 100 years afterward. We call that "time dependent" probability, and it sounds grand. The problem is that you need to know a LOT about the variability of your "500 year events", or time-dependent probabilites don't do any better at forecasting likelihoods of earthquakes then the simple calculation we just did.

Goldfinger's results indeed suggest that the likelihood of a strong southern Cascadia Subduction Zone earthquake is something like twice as high as for the northern part of the zone, which only breaks in great M9 beasts.

Now for my answer to the triggering question: Not that I know of. While it is clear that stress changes and seismic waves from one earthquake can indeed cause other earthquakes to happen, simply not enough is currently known about how this works to enable us to guess how likely it is to happen with any particular earthquake. But my guess is that it is probably fairly rare since I'm not aware of any historical examples of subduction earthquakes being followed by a large crustal earthquake in the overriding plate (and I was looking for this after the 2004 Sumatra earthquake and last month's Chile earthquake).