Italians eat a lot of wheat

[RRM]

24 vs 26 is not that much of a difference.

Correct, that difference is insignificant.
That is in the 75th percentile.
The study says that the odds are 1.0 in that percentile, meaning: no statistically significant difference.

You are clearly exaggerating. If that was such a big deal as you claim it

That was the 75th percentile.
That is in people with moderately high anti-Gliadin levels.
Now lets take a look in people who had more anti-Gliadin in their blood.
Lets take a look in the top 5% (95th percentile) and the top 10% (90th percentile).
In the top 10% (90the percentile), the difference is 16% vs 10%, which is significant.
In the top 5% (95th percentile), the difference is even 11% vs 5%, which is more than twice as much.
That is significant.

Smoking affects everybody, whether you get cancer or not we can see it on your lungs.

No, you cannot see the effects of smoking in the lungs of each smoker.
There are also smokers with so-called clean lungs.
And non-smokers with not-so-clean lungs.
As with all diseases, there are many factors at play, and individual susceptibility is very different.
Something that causes disease in some, will not do so in others.
But, yes, on average, smokers have less-clean lungs than non-smokers.

Gluten does not affect everybody's brain as we could see it in the study.

The study does not show that.
What the study measured, is the association with psychosis.
This study did not measure other effects in the brain.
From this study it shows there is a significant association between levels of anti-Gliadin and risk of psychosis.

That wheat intake is a risk factor.

Japan has the highest number of autistic children and they also have the lowest rate of wheat consumption.

What did i write?
That wheat intake is a risk factor.
I did not write that it is the only risk factor.
I did not write it is the main risk factor.
I wrote it is a risk factor.

The same is true for smoking.
For example:
In Japan, they smoke more cigarettes than in (air-clean) Norway, but in Norway lung cancer-mortality in men is higher.

So, yes, smoking is a risk factor for lung cancer, but not the only factor.
Wheat is a risk factor for psychosis and autism, but not the only factor.

[Aytundra]

Japan has the highest number of autistic children and they also have the lowest rate of wheat consumption.

How do you know they have the lowest rate of wheat consumption?
Last time I looked, noodles are very popular there, and I don't think it is simply "rice" noodles, things like buckwheat noodles are a new fad. Noodles have to be made out of some grain...
packaged noodles are very popular, and cup packaged noodles are very popular.
Like those noodles you get when you fly to asia, and they serve those hot watered noodle cups on the plane.

And even if Japan's whole country's wheat consumption is lower than other countries,
you still have to look at the proportion of wheat in their daily diets.
As in: What proportion of their daily meal is taken up by noodles?

[Ducky]

How do you know they have the lowest rate of wheat consumption?

Look at the map I posted.

[Aytundra]

24 vs 26 is not that much of a difference. You are clearly exaggerating. If that was such a big deal as you claim it scientists would have repeated this experiment many times to get a clearer picture.

This study didnt prove that gluten caused unrepairable damage to the brain.

Ducky, I have a question for you:
Can you name the type of study that Karlsson, H. et al., performed?
And explain to me the scopes and limits of this type of study.

[Aytundra]

How do you know they have the lowest rate of wheat consumption?

Look at the map I posted.

I did and it was very pretty in colours. http://chartsbin.com/view/5267
But nothing substantial, doesn't even tell me a legend. A poor cartographer, geologist and statistician.

[Ducky]

In the top 5% (95th percentile), the difference is even 11% vs 5%, which is more than twice as much.
That is significant.

You are very dishonest.
I guess if it was 1% vs 2% you would say: "Look at this, this is a 100% difference!".
11% out of 211 that's not that much at all.

[Ducky]

How do you know they have the lowest rate of wheat consumption?

Look at the map I posted.

I did and it was very pretty in colours. http://chartsbin.com/view/5267
But nothing substantial, doesn't even tell me a legend. A poor cartographer, geologist and statistician.

Well then why dont you find a better one?

[Ducky]

Ducky, I have a question for you:
Can you name the type of study that Karlsson, H. et al., performed?
And explain to me the scopes and limits of this type of study.

I cannot. And your point is?

[Aytundra]

Ducky another question for you:
This is in general.

What is the difference between correlation and causation?

[Ducky]

What is the difference between correlation and causation?

You didnt even answer my first.
Are you that desperate that you have to pull this redherring nonsense on me?

[RRM]

In the top 5% (95th percentile), the difference is even 11% vs 5%, which is more than twice as much.
That is significant.

11% out of 211 that's not that much at all.

Hmmm. You dont understand the numbers here.
11% says nothing on itself.
Its by no means a qualifying number here.
The 11% only refers to how big the chosen percentile is.
They could have chosen a smaller or bigger percentile, which would result in a smaller or larger percentage.
You obviously never followed a course in statistics.
What matters, is the comparison, the ratio.
One percentage refers only to the size of the percentile.
You can make the percentage as big or small as you want.
So, you take any percentile, and then compare the percentage of case subjects in that percentile vs the percentage of controls in that percentile.
THEN you know the ratio.
Then you know whether case subjects or controls were over-represented,
Comparing the percentages tells you about the distribution.

Let me give an example:
School X has 1000 students.
School Y has 100 students.

Now we want to know which school has relatively more top-performers.
The stats show us that in the 95th percentile, there is 5% of the students from school X.
Is that good or bad?
Its totally meaningless.
It only refers to the size of the sample, and which sample it is (the top 5% group).
Only when we know the percentage of students from school Y in that percentile, will we know whether school X has a lot of top performers, compared to school Y, or not.
Do you understand?

Now, if school Y has 11% in the 95th percentile, it means it has more than twice as many top-performers as school X has.
Its the ratio of 11% vs 5% that counts.
Not the size of the %.
Now we still dont know what grades any of the students had, or whether either one of those schools are any good, compared to other schools.
The percentages dont tell you that.
The percentages only relate to the size of the sample.
And only comparing the percentages is what yields an actual result.
All we know is that school Y has more than twice as many top-performers as school Y has.

explain to me the scopes and limits of this type of study.

I cannot. And your point is?

You didnt even answer my first.

The point is that if you want to understand what the results of a study mean,
you need to know what they actually were investigating, and what not.
If you dont know what they are talking about, you cannot understand what they say.
If you know nothing about statistics, you cannot interpret the numbers that result from a statistical analysis.
Its totally okay.
Nobody knows everything.
Everbody makes mistakes.
Just acknowledge what you dont know.
Dont be ashamed of it.
Embrace it, and learn.
We all learn every single day.
Dont be afraid to not know something.
Not knowing something opens the door to learning something.
But dont pretend that you know, when you don't, because then you will embarrass yourself.
You talk to us as if you understand what those numbers mean, but you show us that you have no clue.
That attitude stops you from progressing.
Acknowledge everything that you dont know, and you will progress.

[Ducky]

OK I understand what you're saying but when you say "twice as many" you also have to take into consideration the twice as many of 'how many'?

1. Clearly 2% vs 4% is not the same as 40% vs 80%.

2. Also for that study to have any relevance it should have been repeated several times elsewhere. Apparently it wasn't convincing enough for that other scientists repeat it.
A 200 person study is not an overwhelming evidence for anything. It can be a base for a theory but not for anything more.

3. This study counted the antibodies and not the opioid peptides. So its indirect evidence. How do you know if every human body reacts the same way by producing the same number of antibodies?

No, I never learned statistics but its not that hard to understand, thanks for the explanation.

[RRM]

OK I understand what you're saying but when you say "twice as many" you also have to take into consideration the twice as many of 'how many'?

In statistics, its about analyzing what is significant.
When analyzing the raw data, they noticed that among the high-level anti-Gliadin study subjects, there were relatively many case controls.
Thats why they highlighted that section (the top).
If you analyze the top half (50th percentile), you will usually find not much difference, as distribution curves will largely overlap.
Going back to the analogy with the schools: when you analyse the mid-performers, both schools will normally perform not much different in that mid-section.
In diseases, the mid sections in statistics usually overlap, which is because most diseases are multifactorial.
This means that there are various factors at play in each disease.
That is also true where you might not expect it, as in lung cancer.
Hence that long could be denied that smoking causes lung cancer.
Only with proper statistics, you will find sound correlations.

And as they analyzed the 90th percentile, they found a significant difference; with a 1.7 odds ratio.
That is really significant.
This difference is even more significant in the 95th percentile.
Keep in mind that there are many factors at play regarding developing a psychosis; there is an entire list of possible causes.
http://en.wikipedia.org/wiki/Psychosis#Causes

to have any relevance it should have been repeated several times elsewhere.

Only if no other study would ever have noticed any association between wheat-gluten and psychosis.
Remember the meta-analysis?
There are at least 12 scientific studies from which you may conclude there is an association between wheat-gluten and psychosis.
So, this study fits right in.

How do you know if every human body reacts the same way by producing the same number of antibodies?

That is certainly not the case.
The study shows that there is a whole range of different levels of response in antibodies.
Every individual is different.

[Aytundra]

Thank you Ducky!
Thank you RRM!
Looks at the splatters on her brain's walls, and appreciates the picasso painting, painted by Ducky and RRM!
Thank you again! I learned a lot from both of you!
Smiles.

[Ducky]

Thank you Ducky!
Thank you RRM!
I learned a lot from both of you!
Smiles.

What have you learned from me?