Free Kick differential over the last 15 years

[Ghossein]

Right, so the review process is correct because you feel in your extremely biased opinion that it was correct? Okay gotcha.

You are asking for some sort of evidence to back up these claims. Have you somehow missed the table that prompted the posting of this thread?

Very strong statistical evidence to be found there.

it's more that i'm not a conspiracy theorist, but sure.

As I've already explained in great detail, simply adding up differentials is not "strong statistical evidence". I have even analysed it against ladder position to find an absolute lack of effect size. Unfortunately, most AFL fans don't understand what constitute statistical analysis, or would just rather pretend that a spreadsheet counts.

 

[Ghossein]

WCE +900 over 15 years. Next best +472.

A normal looking figure for WCE would be +600. Still a fair way in front but not a ridiculous amount. That is a difference of 300. Over 15 years this is 1 free kick a game. With the home and away differences this is about 0 away and +2 at home.

One free kick a game does not seem like a huge problem on a week to week basis.

Let's look at your guestimate of 1-2 dodgy high tackles. Assuming these go in your favour 60% of the time and to the opposition 40% that is a difference of 20%. 20% of 1-2 is 0.3. This would account for 90 of the 300 free kicks that take WC from a reasonable figure of +600 to an outlier of +900.

You previously said 88% of calls are correct. In an average of 40 free kicks a game approximately 5 decisions are incorrect. Win this 3-2 consistently and there's your outlier.

okay, and this means what exactly?

 

[Ghossein]

for anyone to believe this, as has already been pointed out, you would have to assume that the umpires have been ignoring this trend since 2003, and that this trend has existed across different generations of umpires

 

[Taylor]

I have even analysed it against ladder position to find an absolute lack of effect size

I discounted that comparison because ladder position isn't a reliable measure of performance across the season, let alone a single game.

If you can show evidence that teams ranked #1 or #2 play in the grand final more than 25% of the time (8 teams playing for 2 spots in the finals) then there would be a slight drift towards it.

 

[Ghossein]

I discounted that comparison because ladder position isn't a reliable measure of performance across the season, let alone a single game.

If you can show evidence that teams ranked #1 or #2 play in the grand final more than 25% of the time (8 teams playing for 2 spots in the finals) then there would be a slight drift towards it.

Yep and summing differential isn't a reliable measure of umpiring bias across the season, let alone a single game

you can't suddenly be concerned with methodology when it suits you

 

[Andronicus]

it's more that i'm not a conspiracy theorist, but sure.

As I've already explained in great detail, simply adding up differentials is not "strong statistical evidence". I have even analysed it against ladder position to find an absolute lack of effect size. Unfortunately, most AFL fans don't understand what constitute statistical analysis, or would just rather pretend that a spreadsheet counts.

Well as long as you've provided your own analysis that suits your agenda

 

[Ghossein]

Well as long as you've provided your own analysis that suits your agenda

right so i somehow manipulated how a correlation and regression analysis works have I?

 

[Andronicus]

right so i somehow manipulated how a correlation and regression analysis works have I?

You've been pretty happy to ignore a massive statistical anomaly so far.

 

[Ghossein]

You've been pretty happy to ignore a massive statistical anomaly so far.

I'm not ignoring it, I am pointing out all the obvious reasons as to why it is a poor analysis. the main one being simply summing data over a long period of time isn't actually an analysis. You can't just add s**t up on a spreadsheet and act like you're some kind of stats master extraordinaire. I honestly do not know why so many freo supporters have such a hard time getting their head around that.

If you would like to look at my working, here you go. Point out where the 'bias' lies.

 

[Taylor]

Yep and summing differential isn't a reliable measure of umpiring bias across the season, let alone a single game

you can't suddenly be concerned with methodology when it suits you

I think you're choosing the comparison.

The free kick differential across fifteen years compared to other teams in the same competition in the same period is a fair one. You could make the argument that selectively choosing those years might leave out areas of data that hugely impact the comparison.

The only conclusion you can draw from the collation of data and comparison against the other teams in that period is that West Coast has had a measurably positive result from umpiring.

Not what that positive result leads to.
Not what caused the positive result.
Not that there is any bias.

You can hypothesise that bad years were made less bad, that good years were made better, that West Coast has a home advantage - but you won't be able to quantify it off that data.

There is a graph of free kick differential showing the home and away of each team that would allow you to see that west coast has an advantage at home, superior to other home advantages. Their away advantage being so strong, even greater than a few clubs home advantage, should discount a bias by home town umpiring being the sole cause of the massive differential difference between West Coast and the 2nd ranked team.

 

[Taylor]

I'm not ignoring it, I am pointing out all the obvious reasons as to why it is a poor analysis. the main one being simply summing data over a long period of time isn't actually an analysis. You can't just add s**t up on a spreadsheet and act like you're some kind of stats master extraordinaire. I honestly do not know why so many freo supporters have such a hard time getting their head around that.

If you would like to look at my working, here you go. Point out where the 'bias' lies.

It's pretty hard to discredit raw data, but you're giving it a good shake.

Presenting a spreadsheet with a significant statistical anomaly is allowed to be the sole purpose of the spreadsheet. People will draw their own conclusions from it, many will be confirmation bias driven but the integrity of the data isn't at question.

 

[Ghossein]

I think you're choosing the comparison.

The free kick differential across fifteen years compared to other teams in the same competition in the same period is a fair one. You could make the argument that selectively choosing those years might leave out areas of data that hugely impact the comparison.

The only conclusion you can draw from the collation of data and comparison against the other teams in that period is that West Coast has had a measurably positive result from umpiring.

Not what that positive result leads to.
Not what caused the positive result.
Not that there is any bias.

Correct, all you can draw from the raw data is that a differential exists. Nothing more.

You can hypothesise that bad years were made less bad, that good years were made better, that West Coast has a home advantage - but you won't be able to quantify it off that data.

except that this isn't true, as we can quite clearly see by looking at the raw data that certain teams actually had a more favourable differential during poor season than they received during decent seasons. hence why the correlation is 0.015.

There is a graph of free kick differential showing the home and away of each team that would allow you to see that west coast has an advantage at home, superior to other home advantages. Their away advantage being so strong, even greater than a few clubs home advantage, should discount a bias by home town umpiring being the sole cause of the massive differential difference between West Coast and the 2nd ranked team.

okay?

It's pretty hard to discredit raw data, but you're giving it a good shake.

I am not discrediting raw data by conducting further analyses. Raw data isn't an analysis. No one who has the most basic clue about stats would dare to simply look at a chart of raw data and attempt to extrapolate some result from it. Why this is controversial, or biased, or an attempt to 'discredit' the data is absurd.

Presenting a spreadsheet with a significant statistical anomaly is allowed to be the sole purpose of the spreadsheet. People will draw their own conclusions from it, many will be confirmation bias driven but the integrity of the data isn't at question.

I honestly don't see how conformation bias plays into it here, I've been quite open and honest about how I collected the data and the methods I used. People can claim that the methodology might be flawed (which is partially true), but to claim that it's somehow tainted because I support west coast doesn't make any sense.

 

[Taylor]

I honestly don't see how conformation bias plays into it here, I've been quite open and honest about how I collected the data and the methods I used. People can claim that the methodology might be flawed (which is partially true), but to claim that it's somehow tainted because I support west coast doesn't make any sense.

It isn't your confirmation bias I'm talking about, it's the one you're worried about from others.

 

[Andronicus]

I'm not ignoring it, I am pointing out all the obvious reasons as to why it is a poor analysis. the main one being simply summing data over a long period of time isn't actually an analysis. You can't just add s**t up on a spreadsheet and act like you're some kind of stats master extraordinaire. I honestly do not know why so many freo supporters have such a hard time getting their head around that.

If you would like to look at my working, here you go. Point out where the 'bias' lies.

Perfectly relevant if for some strange reason someone decides there needs to be an added column called ladder position that they want to work into the argument.

 

[Tony_Clifton]

If you would like to look at my working, here you go. Point out where the 'bias' lies.

Here you go:

The primary question is whether the outlier figure is legitimate or are WCE getting a more favourable run from the umpires?

Your work did not address this at all.
Your work answered the question whether the differential is correlated with ladder position.
It showed no correlation.

I think you wanted your work to imply that having a positive count has no bearing on the outcome of games.
You imply that the lack of correlation renders the primary question meaningless.

So rather than attempt to answer the primary question you created an alternative question to answer, found a result that proved favourable and extrapolated conclusions from there. Hence, biased lies. The trick is if you don't want to answer the question create an alternative question and answer that. Been watching much question time late at night on the ABC?

 

[Ghossein]

Here you go:

The primary question is whether the outlier figure is legitimate or are WCE getting a more favourable run from the umpires?

Your work did not address this at all.
Your work answered the question whether the differential is correlated with ladder position.
It showed no correlation.

I think you wanted your work to imply that having a positive count has no bearing on the outcome of games.
You imply that the lack of correlation renders the primary question meaningless.

So rather than attempt to answer the primary question you created an alternative question to answer, found a result that proved favourable and extrapolated conclusions from there. Hence, biased lies. The trick is if you don't want to answer the question create an alternative question and answer that. Been watching much question time late at night on the ABC?

My correlation demonstrates that the free kick differential doesn't have an impact on team performance over the season. That's all it shows. Many were actually arguing it, it's not an alternative question, it's just that not everyone in this thread is arguing the exact same thing you are arguing.

I have already answered your primary question. We know that umpires review every match to look for incorrect paid or missed frees, and we know that umpires are right close to 90% of the time. In order for the argument that west coast have been getting a favourable run from the umpires, you would have to first argue that the umpire's review process is flawed, and that the umpiring administration has failed to do something about a clear outlier that has existed since 2003. Personally I choose to take the Occam's Razor approach and think that the free kick differential being legitimate is a far simpler and more likely explanation than a 15 year running conspiracy.

 

[Tony_Clifton]

You made the following post about 250 posts into the thread.

this is why I want to conduct an analysis and see if the differential actually has an impact on the team's performance that year. if it doesn't then the thread can be shut and we can stop complaining about this problem.

Prior to this there was 1 post that made a joke about the bulldogs winning the premiership due to free kicks. So 1 person was arguing something relevant to your correlation analysis.

Over 200 posts were devoted to discussing the reasons for the outlier.

Do you still believe it is not an alternative question?

oh and where you said "If it doesn't then the thread can be shut and we can stop complaining about this problem."
Are you not clearly saying that a lack of correlation renders the outlier discussion meaningless?
From this it follows that it doesn't matter if the free kicks are dubious or not.

 

[Ghossein]

You made the following post about 250 posts into the thread.


Prior to this there was 1 post that made a joke about the bulldogs winning the premiership due to free kicks. So 1 person was arguing something relevant to your correlation analysis.

Over 200 posts were devoted to discussing the reasons for the outlier.

Do you still believe it is not an alternative question?

oh and where you said "If it doesn't then the thread can be shut and we can stop complaining about this problem."
Are you not clearly saying that a lack of correlation renders the outlier discussion meaningless?
From this it follows that it doesn't matter if the free kicks are dubious or not.

ah yeah you're right no AFL fan has ever claimed that the umpires caused a particular team to do well

it hasn't been a consistent theme over the past few years to look at a differential and say that's the reason X team won/lost

we havent at all had a big discussion about the impact of high contact free kicks and winning

i'm just making all this s**t up

 

[Andronicus]

I know if they didn't get the terrible call against North a few years ago they would've lost. Just like they wouldve lost a final against Port.

There are 2 examples from 2 calls. Now, consider how many influential calls may have been made in the other 500 they had over the next best teams and we start to get a different picture.

 

[Tony_Clifton]

ah yeah you're right no AFL fan has ever claimed that the umpires caused a particular team to do well

It is correct that in other threads this has been a topic but not in this thread.

I refer you back to what you wrote previously,
"If it (ladder correlation) doesn't then the thread can be shut and we can stop complaining about this problem."

So the premise of your correlation analysis was not to answer questions from other threads but to disprove the relevance of the 200+posts posts attempting to analyze the reasons for the outlier.

So are you now trying to shift the goal posts after you were the one who set them in the first place?

Are you also saying that free kick differential has no impact on winning a particular game? Because your analysis did not address this either.

The things you have been doing is the equivalent of using analysis of a particular study to distract from the actual topic. Akin to climate deniers focusing on one study of polar bears and arguing that it disproves everything. Decieving, redirecting and confusing the focus of the topic at hand.

If you actually looked into what thos thread is about you might study what constitues an outlier. It is a bit vague and not scientific but generally its 1.5 times the interquartile range above the 75 percentile. I know you don't likegraphs but how about boxplots. When I had a look, according to generally accepted methods the 900 figure doesn't qualify as an outlier. So when I've been referring to the outlier I was wrong. Interesting figure might be more relevant. That's a technicality you could hang your hat on because i know you love technicalities. Doean't stop the fact that the large figure requirea further examination.

 

[Ghossein]

It is correct that in other threads this has been a topic but not in this thread.

I refer you back to what you wrote previously,
"If it (ladder correlation) doesn't then the thread can be shut and we can stop complaining about this problem."

So the premise of your correlation analysis was not to answer questions from other threads but to disprove the relevance of the 200+posts posts attempting to analyze the reasons for the outlier.

So are you now trying to shift the goal posts after you were the one who set them in the first place?

Are you also saying that free kick differential has no impact on winning a particular game? Because your analysis did not address this either.

The things you have been doing is the equivalent of using analysis of a particular study to distract from the actual topic. Akin to climate deniers focusing on one study of polar bears and arguing that it disproves everything. Decieving, redirecting and confusing the focus of the topic at hand.

If you actually looked into what thos thread is about you might study what constitues an outlier. It is a bit vague and not scientific but generally its 1.5 times the interquartile range above the 75 percentile. I know you don't likegraphs but how about boxplots. When I had a look, according to generally accepted methods the 900 figure doesn't qualify as an outlier. So when I've been referring to the outlier I was wrong. Interesting figure might be more relevant. That's a technicality you could hang your hat on because i know you love technicalities. Doean't stop the fact that the large figure requirea further examination.

i'm not going to go through all 24 pages of this thread to find the people that claimed that, i've done enough data searching on this topic . i'd love to examine how differentials have impacts on particular games but we don't have that information

i don't mind box plots but i'll tell you what I like more, using averages instead of sums as to not amplify the effect of outliers. imagine a 10 round season where the differentials are as follows:

+1, -3, +5, -4, +19, +1, +1, -3, -4, +2

now which do you think is a more accurate representation of the overall differential, the sum (+15), or the average (+1.5)?

it's honestly mind boggling how much is wrong with this discussion, be it the raw data, the way people analyse the raw data, the arguments about how home advantage works, or the arguments over whether or not the umpires just ignored an obvious outlier for 15 years. we've started this whole discussion from a spreadsheet where the initial method used to assess differential was inherently flawed.

 

[Tony_Clifton]

i don't mind box plots but i'll tell you what I like more, using averages instead of sums as to not amplify the effect of outliers. imagine a 10 round season where the differentials are as follows:

+1, -3, +5, -4, +19, +1, +1, -3, -4, +2

now which do you think is a more accurate representation of the overall differential, the sum (+15), or the average (+1.5)?

My answer is that the way you have done it they are exactly the same.

I think you got your maths wrong. Using your example for the whole season you just divide by 23. Doesn't change the proportions. It does nothing in regards to the 19 figure. 1 becomes 0.1, -3 becomes -0.3 and 19 becomes 1.9. The 1.9 still retains its impact. If X=y then x/10 =y/10

If you can't comprehend your own argument on ratios why should we trust you with correlation. Having a power saw doesn't make you a carpenter.

I think what you are getting at is if you win the free kick count 25-20 in one game then the average would be +5/45. Continue doing this for all games. Do you think this makes more sense?

 

[Tony_Clifton]

i'm not going to go through all 24 pages of this thread to find the people that claimed that.

Don't worry mate, I've done that for you. Before you jumped on board there was 1 post that claimed that.

 

[Tony_Clifton]

we've started this whole discussion from a spreadsheet where the initial method used to assess differential was inherently flawed.

If you believed this then you are arguing that the input data for your correlation was garbage. Its a bit trite but garbage in, garbage out. Why do the correlation in the first place?

Shouldn't your initial argument be in regards to the averaging thing? Or do you just change each time you are shut down?

 

[Ghossein]

If you believed this then you are arguing that the input data for your correlation was garbage. Its a bit trite but garbage in, garbage out. Why do the correlation in the first place?

Shouldn't your initial argument be in regards to the averaging thing? Or do you just change each time you are shut down?

it was, i was just tailoring it to the measures the thread seems to have a fetish for. if i was doing it legitimately i’d assess it for each match, try to create a measure of how “incorrect” the umpires are, then measure against the score margin

i don’t have the data or the time unfortunately, and it doesn’t seem that anyone actually cares. the thread was started after a contentious match and has since gone completely dead.