The Effect of Joel Embiid's Injury in the Playoffs

[Ferry Avenue]

In the table below we have Joel Embiid's regular season and postseason statistics. Rows labeled with a zero in the column labeled "Playoffs" are regular season statistics, whereas rows labeled with a "1" in that column are postseason statistics. 39 regular season games and 6 postseason games.



None of the above differences in Embiid's performance between regular season and postseason is statistically significant. Even the biggest difference between regular season and postseason -- in his FG% -- cannot be ruled out as being driven by random variation.

Embiid averaged 34.7 points per game in the regular season, with a standard deviation of 9.5 points. In the postseason he averaged 33 points per game with a standard deviation of 10.7 points. He averaged 11 rebounds per game in the regular season and 10.8 rebounds per game in the postseason.

While there is no question Embiid experienced an injury during the regular season, whether his injury hampered him in the postseason in a way that is measurable statistically is highly questionable.

If Embiid was indeed only "70 percent" or some such figure physically, it certainly didn't result in a significant difference in his measurable performance.

The question becomes why so much was made of the effect of his injury in the playoffs -- by him and others -- when it didn't result in any appreciable decrement in his performance. It could've just as easily been said that while Embiid did experience an injury, it didn't seem to be affecting his performance.

[eyeatoma]

Lol Joel finally performs well even when injured and people are saying, eh the injury was no big deal.

Use your eyes and watch the game.

He limped around a bunch, couldn't dunk, couldn't get up for blocks, could barely rebound. He had no lift in his automatic midrange jumper. But yeah, no discernable difference.

The off-season hasn't even started and we're reading stuff like this.

Sent from my SM-S928B using Tapatalk

[Murray_17]

The question becomes why so much was made of the effect of his injury in the playoffs -- by him and others -- when it didn't result in any appreciable decrement in his performance. It could've just as easily been said that while Embiid did experience an injury, it didn't seem to be affecting his performance.

Because the "no statistical difference" you're claiming here comes from the fact that 6 and 39 are too small of a sample, therefore the variance on both is too big and that leads to a case in which the difference between the means on both samples has to be GIGANTIC to catch a statistical difference on every category.

Embiid could be averaging between 6 and 8 more points on +10% FG in those 6 games than on the 39 and the statistics still would not catch a meaningful difference. This is why to measure performance on the playoffs you need to put together several seasons to have enough of a sample to say "yeah, this guy is for real".

The premise you're using here is flawed.

This is also one of those cases where the eye test helps, you can see Embiid's stats diminishing over quarters because his injuries prevent him from being on shape to play over 35-40 minutes, which is easy to verify when you see his jumpers lack lift, etc.

[Ferry Avenue]

The question becomes why so much was made of the effect of his injury in the playoffs -- by him and others -- when it didn't result in any appreciable decrement in his performance. It could've just as easily been said that while Embiid did experience an injury, it didn't seem to be affecting his performance.

Because the "no statistical difference" you're claiming here comes from the fact that 6 and 39 are too small of a sample, therefore the variance on both is too big and that leads to a case in which the difference between the means on both samples has to be GIGANTIC to catch a statistical difference on every category.

That's actually not true. A relatively minor adjustment would've resulted in a significant difference in his PPG between the regular season and postseason. Had he scored 35 points in game 2 (instead of 50) and 25 points in game 6 (instead of 39), the difference in his PPG between the regular season and postseason would've been statistically significant. But of course he didn't score 35 and 25 points in those games -- he scored a whopping 50 and 39, despite being "injured."

This is also one of those cases where the eye test helps, you can see Embiid's stats diminishing over quarters because his injuries prevent him from being on shape to play over 35-40 minutes, which is easy to verify when you see his jumpers lack lift, etc.

How does that compare to the regular season? Did his performance diminish across quarters then as well, despite that he largely wasn't injured? Even if his performance decrement across quarters was exclusive to the postseason, it could be argued that an increase in his minutes per game in the postseason was responsible for that, as opposed to his injury. Had he been asked to play 40+ minutes during the regular season we could've very well seen a similar decrement in his performance across quarters, even without an injury.

[Ferry Avenue]

Lol Joel finally performs well even when injured and people are saying, eh the injury was no big deal.

Use your eyes and watch the game.

He limped around a bunch, couldn't dunk, couldn't get up for blocks, could barely rebound. He had no lift in his automatic midrange jumper. But yeah, no discernable difference.

None of that is reflected statistically. If it was so easy to see, it should be easily reflected statistically as well.

[Murray_17]

That's actually not true. A relatively minor adjustment would've resulted in a significant difference in his PPG between the regular season and postseason. Had he scored 35 points in game 2 (instead of 50) and 25 points in game 6 (instead of 39), the difference in his PPG between the regular season and postseason would've been statistically significant. But of course he didn't score 35 and 25 points in those games -- he scored a whopping 50 and 39, despite being "injured."

You're proving my point, your entire argument is that 2 games with a 15 and 10-point alteration swing the entire result, there is no way to asses if that's a random swing because you don't have enough of a sample to say those two games are a statistical anomaly in the comparison you made.

How does that compare to the regular season? Did his performance diminish across quarters then as well, despite that he largely wasn't injured? Even if his performance decrement across quarters was exclusive to the postseason, it could be argued that an increase in his minutes per game in the postseason was responsible for that, as opposed to his injury. Had he been asked to play 40+ minutes during the regular season we could've very well seen a similar decrement in his performance across quarters, even without an injury.

If he wasn't injured then he can perform better across the quarters because he actually is in better shape. Not to mention the minutes per game on the playoffs increase.

If the argument, as you suggest, is that the increase in minutes affected his performance then there is an actual statistical impact, contrary to your premise.

Your final point is also not a conclusion necessarily in line with your opening, the most probable reason Embiid cannot play 40+ plus minutes per game is because that increases the risk of an injury, which goes back to the origin of the point you started trying to refute.

The point, at the end, is that none of this can be prove by your original argument because the lack of a statistically significative difference is affected by the small sample so it's not a proper comparison, even if the result was an actual statistical difference.

[ExplosionsInDaSky]

Oh my GOD! King Troll at it again! Can't stand this guy!

[eyeatoma]

Lol Joel finally performs well even when injured and people are saying, eh the injury was no big deal.

Use your eyes and watch the game.

He limped around a bunch, couldn't dunk, couldn't get up for blocks, could barely rebound. He had no lift in his automatic midrange jumper. But yeah, no discernable difference.

None of that is reflected statistically. If it was so easy to see, it should be easily reflected statistically as well.

No it wouldn't.

[Ferry Avenue]

That's actually not true. A relatively minor adjustment would've resulted in a significant difference in his PPG between the regular season and postseason. Had he scored 35 points in game 2 (instead of 50) and 25 points in game 6 (instead of 39), the difference in his PPG between the regular season and postseason would've been statistically significant. But of course he didn't score 35 and 25 points in those games -- he scored a whopping 50 and 39, despite being "injured."

You're proving my point, your entire argument is that 2 games with a 15 and 10-point alteration swing the entire result, there is no way to asses if that's a random swing because you don't have enough of a sample to say those two games are a statistical anomaly in the comparison you made.

How does that compare to the regular season? Did his performance diminish across quarters then as well, despite that he largely wasn't injured? Even if his performance decrement across quarters was exclusive to the postseason, it could be argued that an increase in his minutes per game in the postseason was responsible for that, as opposed to his injury. Had he been asked to play 40+ minutes during the regular season we could've very well seen a similar decrement in his performance across quarters, even without an injury.

If he wasn't injured then he can perform better across the quarters because he actually is in better shape. Not to mention the minutes per game on the playoffs increase.

If the argument, as you suggest, is that the increase in minutes affected his performance then there is an actual statistical impact, contrary to your premise.

Your final point is also not a conclusion necessarily in line with your opening, the most probable reason Embiid cannot play 40+ plus minutes per game is because that increases the risk of an injury, which goes back to the origin of the point you started trying to refute.

The point, at the end, is that none of this can be prove by your original argument because the lack of a statistically significative difference is affected by the small sample so it's not a proper comparison, even if the result was an actual statistical difference.

You're not understanding how standard deviation is used in these computations.

[Ferry Avenue]

Lol Joel finally performs well even when injured and people are saying, eh the injury was no big deal.

Use your eyes and watch the game.

He limped around a bunch, couldn't dunk, couldn't get up for blocks, could barely rebound. He had no lift in his automatic midrange jumper. But yeah, no discernable difference.

None of that is reflected statistically. If it was so easy to see, it should be easily reflected statistically as well.

No it wouldn't.

He "can't get up for blocks and can barely rebound," yet his measurable postseason performance in those areas is no different from his regular season performance. I'd say your eyes are deceiving you.

[GoSixersBro]

Dude took Stanford's advice and literally did not watch the games. Stat nerds are thee worst.

[Snotbubbles]

Lies, damn lies, and statistics.

[Murray_17]

You're not understanding how standard deviation is used in these computations.

Yes, i understand, i have a master in statistics dude.

edit: also, a confidence interval is not a complex "computation" and it's something you learn in a cursory introduction to statistics, don't make it sound like it's cryptic or something.

[Ferry Avenue]

Lies, damn lies, and statistics.

Points are statistics and points win games.

If statistics are worse than lies, did the Sixers really beat the Knicks?

[Ferry Avenue]

You're not understanding how standard deviation is used in these computations.

Yes, i understand, i have a master in statistics dude.

edit: also, a confidence interval is not a complex "computation" and it's something you learn in a cursory introduction to statistics, don't make it sound like it's cryptic or something.

You should know that the null hypothesis that Embiid’s injury had no statistically significant effect on his performance can’t be rejected in this case. And that’s the only point being made here.

[LeTimmAy]

One stat that is missing and explains everything: MPG 33 vs 41.

[Monix]

One stat that is missing and explains everything: MPG 33 vs 41.

yep, could've just looked at basketballreference per36

[Sixerscan]

You're not understanding how standard deviation is used in these computations.

Yes, i understand, i have a master in statistics dude.

edit: also, a confidence interval is not a complex "computation" and it's something you learn in a cursory introduction to statistics, don't make it sound like it's cryptic or something.

But the null hypothesis

[Ferry Avenue]

One stat that is missing and explains everything: MPG 33 vs 41.

Still no statistically significant effect for points per minute. Regular season average was 1.04 points per minute (SD = 0.27). Playoffs average was 0.81 points per minute (SD = 0.29). P-value = 0.11. Random variation can't be ruled out.

Again, it wouldn't be difficult to obtain a statistically significant effect here. Simply drop points in game 2 from 50 to 35 and points in game 6 from 39 to 25 and you'd have it. That's the kind of difference one might expect due to the injury, but that's not what happened.

[Ferry Avenue]

You're not understanding how standard deviation is used in these computations.

Yes, i understand, i have a master in statistics dude.

edit: also, a confidence interval is not a complex "computation" and it's something you learn in a cursory introduction to statistics, don't make it sound like it's cryptic or something.

But the null hypothesis

Right, the thing that's been rejected when you take a prescription medication for example, which lets you know it's been proven to work and you're not being used as a guinea pig.