Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Rhinorhino on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Another Statistics Blunder???
- Thread starter Scott24x7
- Start date
AnotherHiggins
Technical User
The point about not mentioning a base line risk is well taken. But still, folks sure seem ready to judge this article based on a single sentence.
The article also says,
The biggest blunder I see in the article is grammatical, not statistical. "California adults who worked 11 hours or more each week". They obviously meant to say, "California adults who worked 11 hours or more [red][of overtime][/red] each week".
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
The article also says,
The Article said:
The biggest blunder I see in the article is grammatical, not statistical. "California adults who worked 11 hours or more each week". They obviously meant to say, "California adults who worked 11 hours or more [red][of overtime][/red] each week".
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
golom
SOURCE: Hypertension: Journal of the American Heart Association, online August 28, 2006
I don't know anything about it, but I am guessing it is a peer reviewed publication (though there are those that argue about whether or not peer-reviewed is still indicative of quality).
RE: Relative Risk - when you would use this unless you were not given the necessary caveats that should be included in every study? I'm having trouble imagining any scenario where a properly conducted study and properly caveated result should ever be subjected to 'Relative Risk'
anotherhiggins
Actually I think they had it right. All 24,000+ study participants worked 11+ hours. What the scentence says is that no one who worked fewer than 11 hours was included in the study (eliminiting the stress of being unemployed - although 11 hours would seem underemployed and that is almost as stressful).
I don't think it was intended to mean that all 24,000 participats worked overtime...
SOURCE: Hypertension: Journal of the American Heart Association, online August 28, 2006
I don't know anything about it, but I am guessing it is a peer reviewed publication (though there are those that argue about whether or not peer-reviewed is still indicative of quality).
RE: Relative Risk - when you would use this unless you were not given the necessary caveats that should be included in every study? I'm having trouble imagining any scenario where a properly conducted study and properly caveated result should ever be subjected to 'Relative Risk'
anotherhiggins
Actually I think they had it right. All 24,000+ study participants worked 11+ hours. What the scentence says is that no one who worked fewer than 11 hours was included in the study (eliminiting the stress of being unemployed - although 11 hours would seem underemployed and that is almost as stressful).
I don't think it was intended to mean that all 24,000 participats worked overtime...
AnotherHiggins
Technical User
Well that makes sense - if all 24,305 people worked OT, then they wouldn't have a sample of folks working less than 39 hours, would they?
Dang.
Where is that infernal edit button?!?
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
Dang.
Where is that infernal edit button?!?
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
- Thread starter
- #24
Well, personally, I think the length of this thread proves my point... I never said I didn't understand what they were presenting. I said I thought they were presenting it "wrong". And to jebenson's point, the 5 in 100 vs 6.45 in 100 is where the REAL proof in the pudding is, eh?
Benson - You're not from Indianapolis are you???
Best Regards,
Scott
"Everything should be made as simple as possible, and no simpler."![[hammer]](/data/assets/smilies/hammer.gif "[hammer] [hammer]")
Benson - You're not from Indianapolis are you???
Best Regards,
Scott
"Everything should be made as simple as possible, and no simpler."
Maniac...
I guess I took issue with the way you phrased it:
Its not wrong, and it doesn't feel wrong to me. But it certainly could be a whole lot clearer.
And if numbers are the issue:
Overall 30% of Americans have hypertension or are taking blood pressure lowering medication.
So even if we knock, say 10% off of that as the baseline for the 'average' worker in CA who works less than 40 hours a week, we would see an incidient rise from 20 out of 100 workers to 26 out of a 100 workers for those who worked more than 50 hours a week.
Maybe the article should have made it explicit what the risk is but it is a pretty well established fact in the US at least that HBP is one of the leading causes of death in the US.
Heart Disease, which I believe HBP is one element of, is #1, however HBP alone in 2002 was the primary cause of death for 277,000 Americans (facts.htm link above) which in the 2006 would make it the 3rd leading primary cause of death if taken by itself.
I guess I took issue with the way you phrased it:
Its not wrong, and it doesn't feel wrong to me. But it certainly could be a whole lot clearer.
And if numbers are the issue:
Overall 30% of Americans have hypertension or are taking blood pressure lowering medication.
So even if we knock, say 10% off of that as the baseline for the 'average' worker in CA who works less than 40 hours a week, we would see an incidient rise from 20 out of 100 workers to 26 out of a 100 workers for those who worked more than 50 hours a week.
Maybe the article should have made it explicit what the risk is but it is a pretty well established fact in the US at least that HBP is one of the leading causes of death in the US.
Heart Disease, which I believe HBP is one element of, is #1, however HBP alone in 2002 was the primary cause of death for 277,000 Americans (facts.htm link above) which in the 2006 would make it the 3rd leading primary cause of death if taken by itself.
Just another worthless statistic to fill periodical pages.
Does age play a factor?
Does length of commute factor?
Are hours worked directly proportional to blood pressures?
Does gender factor?
Does stress from home carry over to stress at work?
Heck yea, they all do plus another 50 things...
Bo
Kentucky phone support-
"Mash the Kentrol key and hit scape."
Does age play a factor?
Does length of commute factor?
Are hours worked directly proportional to blood pressures?
Does gender factor?
Does stress from home carry over to stress at work?
Heck yea, they all do plus another 50 things...
Bo
Kentucky phone support-
"Mash the Kentrol key and hit scape."
Bo
See anotherhiggins post from "30 Aug 06 12:19"
1) The article specifies they too account for all that
and
2) As a published study in a peer-reviewed journal that almost certainly means the study was a statistically sound study which means those factors you mentioned are held constant.
That means that for every age bracket, for every commute length, for every gender, for every level of stress at home, and for all other 50 things, you would see, with the standard statistical caveats, a 29% increase the occurence of high blood pressure.
See anotherhiggins post from "30 Aug 06 12:19"
1) The article specifies they too account for all that
and
2) As a published study in a peer-reviewed journal that almost certainly means the study was a statistically sound study which means those factors you mentioned are held constant.
That means that for every age bracket, for every commute length, for every gender, for every level of stress at home, and for all other 50 things, you would see, with the standard statistical caveats, a 29% increase the occurence of high blood pressure.
Benson - You're not from Indianapolis are you???
Nope...I'm from (and still in) Texas. [small]YeeHaw![/small]
I used to rock and roll every night and party every day. Then it was every other day. Now I'm lucky if I can find 30 minutes a week in which to get funky. - Homer Simpson
Arrrr, mateys! Ye needs ta be preparin' yerselves fer Talk Like a Pirate Day! Ye has a choice: talk like a pira
Nope...I'm from (and still in) Texas. [small]YeeHaw![/small]
I used to rock and roll every night and party every day. Then it was every other day. Now I'm lucky if I can find 30 minutes a week in which to get funky. - Homer Simpson
Arrrr, mateys! Ye needs ta be preparin' yerselves fer Talk Like a Pirate Day! Ye has a choice: talk like a pira
[/i]...nearly 100% of math students study math.[/i]
Yes, but do they give it 110%?
I used to rock and roll every night and party every day. Then it was every other day. Now I'm lucky if I can find 30 minutes a week in which to get funky. - Homer Simpson
Arrrr, mateys! Ye needs ta be preparin' yerselves fer Talk Like a Pirate Day! Ye has a choice: talk like a pira
Yes, but do they give it 110%?
I used to rock and roll every night and party every day. Then it was every other day. Now I'm lucky if I can find 30 minutes a week in which to get funky. - Homer Simpson
Arrrr, mateys! Ye needs ta be preparin' yerselves fer Talk Like a Pirate Day! Ye has a choice: talk like a pira
I just love the
78% of women agree that xyz (overpriced product) made them feel better
Study based on 93 women.
93 women out of several billion is hardly representative, but then when you selling this to an already gullable audience, who cares about facts?
Only the truly stupid believe they know everything.
Stu.. 2004
78% of women agree that xyz (overpriced product) made them feel better
Study based on 93 women.
93 women out of several billion is hardly representative, but then when you selling this to an already gullable audience, who cares about facts?
Only the truly stupid believe they know everything.
Stu.. 2004
Has anyone considered the age-old cause/effect issue in many of these statistics?
Those who work overtime--maybe they do so *by choice* because they're Type A's, and Type A's have been proven in other studys to have higher blood pressure--overtime notwithstanding.
This means that it is not necessarily true that if a type B works overtime, then he will then have higher blood pressure as a direct result of working more overtime.
100% of all people who have eaten carrots have died or will be dead. Carrots, therefore, are 100% deadly. Stop eating carrots and you'll live forever.
--Jim
Those who work overtime--maybe they do so *by choice* because they're Type A's, and Type A's have been proven in other studys to have higher blood pressure--overtime notwithstanding.
This means that it is not necessarily true that if a type B works overtime, then he will then have higher blood pressure as a direct result of working more overtime.
100% of all people who have eaten carrots have died or will be dead. Carrots, therefore, are 100% deadly. Stop eating carrots and you'll live forever.
--Jim
AnotherHiggins
Technical User
Columb,
Type A vs. Type B is a well known distinction here in the states and is nothing new. Basically, Type A folks can be considered 'uptight' while Type B folks are more 'go-with-the-flow'.
More here
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
Type A vs. Type B is a well known distinction here in the states and is nothing new. Basically, Type A folks can be considered 'uptight' while Type B folks are more 'go-with-the-flow'.
More here
[tt]_____
[blue]-John[/blue][/tt]
[tab][red]The plural of anecdote is not data[/red]
Help us help you. Please read FAQ181-2886 before posting.
Stu
Actually on something as simple and straight-forward as a yes/no question (do you feel better with product versus without), 93 women could be a representative sample.
Its not trying to make a causal medical link which would require a greater degree of certainty, it is mearly asking a yes/no question. If the sample was methodologically sound to and inclusive, then you could start making claims based off of that survey. Of course the survey has to be sound... (and I'm not even going to pretend I know that one).
***
One of the most mis-understood things about statistics is you do NOT need a large sample to start drawing fairly accurate conclusions if you are only looking at one variable. As few as 20 respondents can start producing fairly accurate results if the sample is representative.
Representative does not require that you have 50 of every possible combination, it mearly means that you randomly selected a signficant portion of the population at random.
The more variables that are included (say if you are trying to determine what % of caucasion women of spanish decent who are between 18-24 with children, a library card and at least 4 credit cards) the larger the sample needs to be because you are introducing more factors.
But on a simple yes or no question you would need very few respondants to gain a fair degree of accuracy.
***
The gallop poll (I think it's gallop) is very accurate in predicting the presidental election outcome, and if you look at the numbers, there are often fewer than 1,000 respondants.
You don't need a large sample, you need a representative sample and it doesn't have to be large to be representative.
Actually on something as simple and straight-forward as a yes/no question (do you feel better with product versus without), 93 women could be a representative sample.
Its not trying to make a causal medical link which would require a greater degree of certainty, it is mearly asking a yes/no question. If the sample was methodologically sound to and inclusive, then you could start making claims based off of that survey. Of course the survey has to be sound... (and I'm not even going to pretend I know that one).
***
One of the most mis-understood things about statistics is you do NOT need a large sample to start drawing fairly accurate conclusions if you are only looking at one variable. As few as 20 respondents can start producing fairly accurate results if the sample is representative.
Representative does not require that you have 50 of every possible combination, it mearly means that you randomly selected a signficant portion of the population at random.
The more variables that are included (say if you are trying to determine what % of caucasion women of spanish decent who are between 18-24 with children, a library card and at least 4 credit cards) the larger the sample needs to be because you are introducing more factors.
But on a simple yes or no question you would need very few respondants to gain a fair degree of accuracy.
***
The gallop poll (I think it's gallop) is very accurate in predicting the presidental election outcome, and if you look at the numbers, there are often fewer than 1,000 respondants.
You don't need a large sample, you need a representative sample and it doesn't have to be large to be representative.
I wonder how many of these 'surveys' take a sample we'll call it 2000 for matter of argument, then ask their questions and get 95 people of those 2000 who 'like' the product. Well if they take those 95, plus 5 more people, and trim their sample down to those 100 people, 95% of them like the product.
I just wonder if that happens.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- News
- Locked
- Question