I was thinking about it the other day, and there's been a number of articles in the past on secondary assists and the suggestion that they tell us a bit less about offensive talent than do goals and primary assists. Intuitively, this might make sense, particularly if you look at the entire player population. But what if you split up defensemen and forwards, on the presumption that there is typically a distinct difference between where they play their game in the offensive zone?
Admittedly, part of my inspiration for this post came from a compelling comment by reader "The Men from the Ministry" on Neil Greenberg's great Fan Post from last week. It led me to think about how important your team can be to a defenseman's shot from the point versus a typical shot made by a forward. How frequently does a goal from the point occur because of a screen? If I had to wager a guess, I'd say very. But I think few defensemen are trying to pick corners from there in the first place, settling for just getting it on net and letting their teammates take care of the rest. Which leads to my second thought: could secondary assists mean more for telling us about the offensive talent of defensemen? Obviously, a deflected shot into the net is a primary assist, but when you get into the world of rebounds and tic-tac-toe plays generated from the point, there's a possibility that there's a disparity between forwards and defensemen. Furthermore, good rebound opportunities are not likely to develop if a defensemen can't consistently get the shot on net.
I took the last 3 years' data for 5-on-5 Corsi Rel and ran its correlation to goals/60, primary assists/60, and secondary assists/60 for forwards and defensemen playing at least 20 games. Here's what I got:
Forwards
Goals: 0.3725
Primary assists: 0.3638
Secondary assists: 0.3301
Defensemen
Goals: 0.2034
Primary assists: 0.1982
Secondary assists: 0.2363
So it's possible that, because forwards compile more of the goals and assists, their disparity has masked another interesting one within the player population during 5-on-5 hockey. What if we were to run the same kind of study for 5-on-4 (minimum TOI 1min/60)?
Forwards
Goals: 0.1660
Primary assists: 0.2268
Secondary assists: 0.2305
Defensemen
Goals: 0.2547
Primary assists: 0.2632
Secondary assists: 0.2337
Well, that makes things interesting again. Defensemen rely on the powerplay for their points more so than forwards, but the powerplay accounts for far less time on-ice (which can fart around with our Corsi Rel, though I attempted to reduce that effect); therefore, we are brought essentially back to square one by seeing the weaker correlation for secondary assists. That said, square one doesn't mean secondary assists aren't important, as you can see our correlation in 5-on-4 for defenders is actually higher than any of the forward correlations.
What I take from all of this is that the powerplay, in general, is an important driver of defensemen point totals (for you fantasy hockey players out there, determining whether a defender will get PP time is almost as important as finding out whether a player is the starting goaltender). Further, if we are to assess a defensemen's offensive capabilities using boxcar stats, secondary assists probably shouldn't be excluded. What does this mean for forwards? Secondary assists do mean a little less, as 5-on-5 points count more towards a forward's production. But I think it's important to look for these little complexities in the data before we completely discount a statistic, boxcar or otherwise.
I think J.P. over at Japer's Rink said it best:
There's nothing inherently "better" about a primary assist – they can be every bit as ugly and cheap as any secondary helper.
