Confirmation Bias, Lineup Utilization, and the Ottawa Senators

After this week's events, fans of the Ottawa Senators are furious - and rightly so. Though the Sens were coming out of the Olympic break with a 7-2 loss to the Bruins, the team had went 9 - 3 - 4 since the turn of the new year, and fans were high on the optimism brought by the Olympics (unless you were American or Russian - my condolences). 4 games later, optimism is at a season-low, with a spanking by Alfie's Red Wings and convincing losses to Edmonton and Calgary (not exactly good teams) highlighting some questionable lineup decisions by Paul MacLean.

Just like everyone else, I was furious. I went to bed upset, raged on twitter, questioned everything, ranted about Melnyk, and only stopped to reflect yesterday. After writing down some of my observations and researching more into them, I realized that my initial train of thought may have been warped by two common psychological phenomena: confirmation bias, and groupthink.

Confirmation bias is a phenomenon whereby people look for evidence to confirm their hypothesis, rather than look for contradictory evidence. In sports, this is extremely evident. We, as fans, don't have access to the locker room, coaching staff, and management to vent our frustration to, and usually, just have an endless amount of questions but no way to get our own answers. We rely on what we see, and make judgments upon what's in front of us (line combinations, player interviews, game footage, etc). Through this, it's a lot easier to pick out events that support our conclusions, and rely on fellow fans who share our opinion to confirm what we already know. This is the foundation for groupthink, another psychological phenomena, where instead of critically evaluating information, group members form opinions that match the group consensus.

We can find many cases for both of these phenomena (Jared Cowen, Eugene Melnyk) but the example I'm going to use is Paul MacLean. Coming off the high of the Senators first playoff series win in what seems like forever, we were ready to enshrine MacLean [hell, I even bought a Paulrus t-shirt]. As pointed out by the great @steffeg on twitter, amazing goaltending and low expectations may have shrouded some of MacLean's questionable lineup decisions - ones we are now being exposed to now. Everything is a lot easier to accept when you're winning (Hi Toronto and Montreal!), and the glutton of injuries led to there not being many NHL bodies left to fill spots, and also an outlet to blame any bad performances on. This year, expectations were high - the Bobby Ryan trade, Melnyk + Murray both proclaiming that the Sens were ready to contend, healthy Spezza + Karlsson + Michalek + Cowen, but the team has failed to deliver. Thus, the blame has been put on many individuals, with MacLean and his lineup decisions getting their fair share right now. For a quick glance, search ‘Paul MacLean' on twitter, or go through the @Senators mentions/comments.

However, an issue arises - how do we know we aren't just succumbing to confirmation bias + groupthink, and thus, generating reasons (hypotheses) that are biased negatively to MacLean while also just piling on because everyone else is doing it? Short answer: we don't know until we critically think about the situation. Thankfully, we have some resources available to examine some of the accusations facing MacLean, and can use these resources + statistics to critically think about the situation at hand. The points I'll talk about from here on out are observations that I've noticed re: MacLean and lineup utilization, and thus, I'm going to go through the resources I have available to see whether they hold up or not.

Observation 1. MacLean seems to switch lines really frequently. Is this true?

Cory Conacher made an interesting comment after he was picked up by Buffalo:

Conacher on Sens: "The lines switched a lot here."

— Hockey Central (@SNHockeyCentral) March 5, 2014

A quick look at Left Wing Lock lets us know that the main combinations MacLean lines are the MacArthur - Turris - Ryan (together 17% of the time at EV), and Greening - Smith - Neil (together 12.82% of the time) - which confirms what most of us have picked up as well. Those frequency %'s may seem low, but I highly encourage you all to look at the other teams in the tool, and you'll see that this is the case for almost ALL of the teams. Most coaches love to keep duos together, and rotate around the 3^rd guy. MacArthur - Turris, Michalek - Spezza, Smith - Neil all seem to be duos MacLean likes.

What Conacher may be getting at is the extent to which MacLean "puts his lines in a blender" in-game, which is harder to look at, but by my eye, it doesn't seem like a unique thing among NHL coaches.

One interesting comment I found re: the importance of having consistent linemates comes from Justin Bourne (a must-follow @jtbourne who writes over at the Backhand Shelf):

There were two moments in last night's NHL Revealed that reminded me of one "chemistry" difficulty - knowing what hand your linemates are. I know that sounds like a super easy thing, but it's funny how you fall into a mental groove when you know, say, both your linemates are left shots. You're occasionally passing it to a jersey without knowing which linemate it is, and it's nice to have that default so you know which side of their body you should pass to.

At one point there's just a terribly botched pass and Getzlaf says to Crosby (jokingly), "Wait, you're not right-handed? You're not Perry?" He really did pass it to the wrong side of Sid's body.

It's not just knowing which hand guys are either - some guys like passes on their forehand even when it's on their backhand side (meaning you pass it behind their back foot), some guys, as we saw Dustin Brown in the show, make it clear they want it "Backhand, backhand" whenever possible.

So that's the one little glitch that comes with line shuffling - it gives the players one extra, annoying thing to think about.

Although this was with respect to the Olympics, a 2-week tournament where players come together from different teams, the handedness comment could still apply to changing linemates in-game.

Observation 2. The Smith line seems to play a TON, especially in losses. Is this true?

To examine this, I went through the even-strength TOI, in order to try to remove the effect of special teams (even though MacLean has stated that it does interrupt with shift order), for the last 10 Sens wins and losses. I also included the ranking of each player on the team in brackets, as well as the total season EV TOI per game as a comparison (from NHL.com).

[if you wanted to see the specific games + full spread of TOI's, click here]

The tricky thing with this type of analysis is that there are so many variables in determining a win and a loss. For example, some of these wins/losses are decided by the shootout, whereas others are blowouts. Some involve the Sens dominating possession but getting unlucky + losing, while some involve the Sens lacking in possession but getting a few breaks and winning. Regardless, I still think this data is useful at a basic level to see if there is any difference in ice-time between wins and losses, and for the purpose of my observation - it works just fine.

As we can see over the past 10 wins and losses, the trio of Greening, Smith and Neil actually play a little more when the Sens win! The players who play more in losses than wins (third column) are Zibanejad (!), Conacher, Spezza, and Michalek. Unsurprisingly, in games that the Sens have won, MacArthur and Ryan have played more minutes.

I added in each player's individual ranking to try and find some context, as well as for the purpose of comparison. As we can see, regardless of wins/losses, Neil, Conacher, and Condra play the least amount of minutes, and Turris, Spezza, and Smith play the most. From this small sample, ice time for Zibanejad and Ryan seem to vary the most if the Sens win or lose, with MacArthur and Michalek in the middle somewhere.

In sum, it looks like my initial observations were a bit biased. Though this analysis isn't conclusive, it looks like MacLean's line juggling seems to be a league-wide NHL coach trait (sorry, Cory), and that the Zack Smith line plays around the same number of minutes in both wins and losses. Now, you can argue that Ryan + MacArthur should have more EV TOI than Spezza and Michalek, and that Conacher + Condra should also be allotted more instead of Greening - Smith - Neil, if we're judging by performance and possession, but we'll leave that for another post.

Although confirmation bias + groupthink are cognitive biases that happens unbeknownst to the individual sometimes, there are steps to take to avoid it. Firstly, do some individual research - often, it means a lot more if you see something with your own eyes rather than take someone's word for it (which is a reason I try to present all my data + methodologies so that you can see for yourself and replicate what I did). Secondly, play "devil's advocate". This is where you take the opposing viewpoint for discussions sake in order to try and see both sides. For example, even though I think Cowen is terrible, take the stance that he is good and try to find supporting evidence, which in turn, will either confirm or disconfirm your original hypothesis that Cowen is terrible. Communities such as this one really help us combat these biases, as the discussion here is pretty great and spans multiple viewpoints. Though it's extremely easy to agree with the group, going against the grain once in a while (such as Mark's defense of Cowen in 5 thoughts) helps to push the limits of our critical thinking abilities.

Thanks for reading!