Dolphins have scored 96 points this season, 15 more than any other AFC East team...

[J. David Wannyheimer]

What I said doesn't take away from the team scoring a rare number of points in two games. What I said was that it statistically skews the average in our favor because of the small sample size. I'm also implying that it's problematic to use the averages as they sit to compare us with other teams because of that skewing. Now if we score 30+ points in 6-8 games, then clearly I'm wrong. We'll see.

You specifically referred to the Oakland game as an outlier, when the truth is that NFL games should never be dismissed as outliers in a discussion of a team's overall performance. In a highly competitive league that plays a small number of games and in which scoring is highly variable from week to week, picking and choosing certain results as being more significant than others is quackery. And on top of that, when the team scored 33 points just three weeks prior, dismissing a 38 point performance against an objectively inferior opponent as an outlier is patently absurd.

Through four weeks, the Miami Dolphins have had two pretty good offensive performances, and two rather poor ones. In the next twelve games, Miami will almost certainly have more offensive performances that will be pretty good, and some that will be rather poor. However, the truth is that after four weeks, Miami has had just as many 'explosive' scoring performances (33+ points) as they've had in a full sixteen game season over any of the past 13 years. There's no projection or 'fluke' there. That's a simple fact: in four games, the Miami Dolphins have had as many dominant offensive performances as they typically have in a full sixteen game season.

You can either make the assumption that the Dolphins shot their wad in the first four games, or you can assume that the Miami Dolphins are likely to have their best offensive performance since 2001 (when they finished 8th in scoring) this season. Or you can assume nothing, which would be pretty silly on a discussion board. Personally, I assume that the Dolphins offense will continue to perform very well against teams that don't have outstanding defensive lines, and struggles against teams that do have outstanding defensive lines. This would fit the trend of the Joe Philbin/Ryan Tannehill era in Miami. However, I'm also going to assume that our good performances will be more productive, based on what I've seen in weeks one through four.

I think that's fair.

 

[Krush]

Defense really needs to step up and put Rodgers on his can, and the defensive back need to be in their receivers hip pockets, and none of this stuff in allowing Rodgers completing 12 yards passes in front of you and consider you have done a good job.

 

[Padfoot]

To me, this is evidence that a) we can contend this year and b) Tannehill is our starting QB

That being said, we have given up the most points in the division...

Thoughts from the Tanne-haters?

71 of those points came against piss poor defensive performances. And outside of the first half of the Raiders game, Tannehill's accuracy has looked awful.

I hope the Raiders game was more of a wake up call than a fluke but don't kid yourself that you can just say "oh look, we scored 96 points and that's all that matters, everything must be going well."

 

[JTC111]

You specifically referred to the Oakland game as an outlier, when the truth is that NFL games should never be dismissed as outliers in a discussion of a team's overall performance. In a highly competitive league that plays a small number of games and in which scoring is highly variable from week to week, picking and choosing certain results as being more significant than others is quackery. And on top of that, when the team scored 33 points just three weeks prior, dismissing a 38 point performance against an objectively inferior opponent as an outlier is patently absurd.

Statistically it is an outlier because it's outside of the range of what's normal for this team. All that means is that high scoring games for the Miami Dolphins represent a small percentage of their games. And since you brought it up, let's include the 33 points against NE. Right now, we have two 30+ games out of four. Those games skew us upward unless 50% of our games are 30+ games. I doubt very much that will be the case.

Some of you think I'm insulting the team somehow by pointing this out. I'm not. I'm simply saying that the numbers we're seeing right now may not be reliable. Again, we'll see.

 

[Sons Of Shula]

Statistically it is an outlier because it's outside of the range of what's normal for this team. All that means is that high scoring games for the Miami Dolphins represent a small percentage of their games. And since you brought it up, let's include the 33 points against NE. Right now, we have two 30+ games out of four. Those games skew us upward unless 50% of our games are 30+ games. I doubt very much that will be the case.

Some of you think I'm insulting the team somehow by pointing this out. I'm not. I'm simply saying that the numbers we're seeing right now may not be reliable. Again, we'll see.

And the same can be said for the 10 and 15 points we scored in games 2 and 3. Considering we were still learning a new offense those point totals "skewed" our average and was a statistical "outlier".

Right now, we have two games in the teens out of four. Those games skew us downward unless 50% of our games are 15 or less games. I doubt very much that will be the case.

You'll see.

 

[JTC111]

And the same can be said for the 10 and 15 points we scored in games 2 and 3. Considering we were still learning a new offense those point totals "skewed" our average and was a statistical "outlier".

Right now, we have two games in the teens out of four. Those games skew us downward unless 50% of our games are 15 or less games. I doubt very much that will be the case.

You'll see.

Last year we averaged 19.8 ppg.
In 2012 we averaged 18 ppg.
Prior to Philbin, in 2011 we averaged 20.5 ppg.
2010, 17.1 ppg.
2009, 22.5 ppg.
2008, 21.6 ppg.
2007, 16.7 ppg.
2006, 16.25 ppg.
2005, 19.9 ppg.
2004, 17.2 ppg.
2003, 19.4 ppg.
2002, 23.6 ppg.

Historically, those low games are closer to what we average than those high games, so the high games skew us more. Right now, we're averaging 24 ppg. That's higher than any end of season average we've had in the previous 12 years. You expressed hope and optimism that "it rises another 5-7 points by seasons end," and it was your post that I was primarily responding to when I made my first post in this thread. I'd love to average 29 or 30 points too, but that's just not going to happen.

 

[J. David Wannyheimer]

Statistically it is an outlier because it's outside of the range of what's normal for this team.

Nah. The Q-Test is probably the simplest and easiest method you can use to determine whether or not a particular data point is an outlier, and 38 points last week clearly isn't.

Considering all games between 2012 and 2014 (the Philbin/Tannehill era), you've got 36 data points, which is a very decent sample. Your gap is 3 and your range is 38. That gives you a Q value of 0.08, rounding up to the hundredth. That means I'm over 99.99% confident that 38 points is not an outlier in my data set. In fact, it's probably closer to 99.999999999999% confidence -- I'm too lazy to go flip through the chart for it.

That's like, first semester statistics. You can argue, but you're just wrong. It's not a statistical outlier. It's only an outlier in your mind.

 

[JTC111]

Nah. The Q-Test is probably the simplest and easiest method you can use to determine whether or not a particular data point is an outlier, and 38 points last week clearly isn't.

Considering all games between 2012 and 2014 (the Philbin/Tannehill era), you've got 36 data points, which is a very decent sample. Your gap is 3 and your range is 38. That gives you a Q value of 0.08, rounding up to the hundredth. That means I'm over 99.99% confident that 38 points is not an outlier in my data set. In fact, it's probably closer to 99.999999999999% confidence -- I'm too lazy to go flip through the chart for it.

That's like, first semester statistics. You can argue, but you're just wrong. It's not a statistical outlier. It's only an outlier in your mind.

I'm pretty sure you didn't actually run the numbers. I didn't either. I'm assuming that the largest number of points that we've scored in years is an outlier. But if you want to actually run them and show me that it's not, go ahead. But I'd like to see the work. Short of that, we'll have to agree to disagree on that point.

 

[J. David Wannyheimer]

I just gave you the work. The Q-Test is extremely simple. Gap / Range. Easy.

I specifically chose that test because 1) it's the most common that's used for non-exhaustive analysis, 2) it can be done in mere seconds with nothing but a calculator, and 3) it's easy to explain.

Gap = the absolute difference between the point in question and the next closest point. Range = the difference between the highest value and the lowest value in the set. Q value = gap divided by range. Compare against the confidence interval table. If the Q value is above the number for your confidence interval, it's an outlier. If not, it's not. The value for an outlier at 99% confidence has to be over 0.5, so I didn't even have to look up the table.

It's not my fault you have no understanding of basic statistics, yet insist on eyeballing some arithmetic, declaring one game to be a statistical anomaly, and then trying to use that as a springboard for arguments painted with a broad brush. That's the kind of thinking that led to shouright declaring Davone Bess a playmaker based on a single number derived from analytics.

The fact that you asked to 'see the work' pretty much says it all. It's a ****ing Q-Test. The work consists of basic division.

 

[JTC111]

I just gave you the work. The Q-Test is extremely simple. Gap / Range. Easy.

I specifically chose that test because 1) it's the most common that's used for non-exhaustive analysis, 2) it can be done in mere seconds with nothing but a calculator, and 3) it's easy to explain.

Gap = the absolute difference between the point in question and the next closest point. Range = the difference between the highest value and the lowest value in the set. Q value = gap divided by range. Compare against the confidence interval table. If the Q value is above the number for your confidence interval, it's an outlier. If not, it's not. The value for an outlier at 99% confidence has to be over 0.5, so I didn't even have to look up the table.

It's not my fault you have no understanding of basic statistics, yet insist on eyeballing some arithmetic, declaring one game to be a statistical anomaly, and then trying to use that as a springboard for arguments painted with a broad brush. That's the kind of thinking that led to shouright declaring Davone Bess a playmaker based on a single number derived from analytics.

The fact that you asked to 'see the work' pretty much says it all. It's a ****ing Q-Test. The work consists of basic division.

As we both know, there's really no standard definition for "outlier," so I was hoping you'd run the numbers and show me how far 38 points deviates from the norm. But in the end we're simply going to wind up arguing about whether that number is enough to justify calling it an outlier. In other words, lacking a standard mathematical definition of the word, "outlier" can often become a matter of opinion in all but the most extreme cases. But I'm pretty comfortable saying that doubling our ppg average for the previous year is an outlier.

 

[J. David Wannyheimer]

As we both know, there's really no standard definition for "outlier," so I was hoping you'd run the numbers and show me how far 38 points deviates from the norm. But in the end we're simply going to wind up arguing about whether that number is enough to justify calling it an outlier. In other words, lacking a standard mathematical definition of the word, "outlier" can often become a matter of opinion in all but the most extreme cases. But I'm pretty comfortable saying that doubling our ppg average for the previous year is an outlier.

Nope. You're just wrong. I've demonstrated that you're wrong. It's not even close to an outlier. It's not even CLOSE. The fact that I used the most common method used to determine whether or not a data point is an outlier and determined with 99.9999999% confidence that it's not an outlier, versus the fact that you don't even know what that method is and are still arguing with me, really doesn't help your argument at all.

Again, it's not even CLOSE. Get a grip.

 

[JTC111]

Nope. You're just wrong. I've demonstrated that you're wrong. It's not even close to an outlier. It's not even CLOSE. The fact that I used the most common method used to determine whether or not a data point is an outlier and determined with 99.9999999% confidence that it's not an outlier, versus the fact that you don't even know what that method is and are still arguing with me, really doesn't help your argument at all.

Again, it's not even CLOSE. Get a grip.

So you know a standard definition of "outlier"? And I don't mean a definition some professor gave you, or a definition preferred by a random textbook. I'm asking for a mathematically accepted standard that applies all the time. As far as I know, no such definition exists.

 

[J. David Wannyheimer]

So you know a standard definition of "outlier"? And I don't mean a definition some professor gave you, or a definition preferred by a random textbook. I'm asking for a mathematically accepted standard that applies all the time. As far as I know, no such definition exists.

You are completely ignorant of the topic and it's painfully obvious at this point. If you'd like a second opinion, go ask Locke for one. He's familiar with statistical methodology and I'm sure he'd be more than happy to waste 20 minutes of his life trying to explain how to differentiate one's ass from a hole in the ground for you.

The fact that you don't even understand the difference between a mathematical proof and analytical processes is just