Thread: Science just making shit up now?

ADHD, Learning Issues May Be Linked to Secondhand Smoke - ADHD Center - Everyday Health

"Children exposed to secondhand smoke in their homes face a higher risk of developing attention-deficit/hyperactivity disorder, other behavioral problems and learning disorders, a new study finds."



How do you start with this and then have all this shit afterwards?


"The research doesn't definitively prove that tobacco smoke can harm children's brains"

"It's difficult to confirm whether secondhand smoke causes children's health problems"

"After adjusting their numbers to improve their validity from a statistical point of view"

"The researchers estimated"

Thats not science. Thats a POS earning a grant over the course of a year with a pr statement of an antismoking nature that i could of done while riding the coach. Precisely the kind of shit that irritates me. You have no point to make other then an opinion full of second guesses and altered numbers? WTF

While the title of this is a bit sensationalist (It's not science that's making this up), I agree with you completely on this. This study is a joke. I didn't see mention of any peer reviews at all. Most likely because this person's peers would laugh at this. It's just someones chance to get their name on the internet.

Originally Posted by deathgodusmc

ADHD, Learning Issues May Be Linked to Secondhand Smoke - ADHD Center - Everyday Health

"Children exposed to secondhand smoke in their homes face a higher risk of developing attention-deficit/hyperactivity disorder, other behavioral problems and learning disorders, a new study finds."



How do you start with this and then have all this shit afterwards?


"The research doesn't definitively prove that tobacco smoke can harm children's brains"

"It's difficult to confirm whether secondhand smoke causes children's health problems"

"After adjusting their numbers to improve their validity from a statistical point of view"

"The researchers estimated"

Thats not science. Thats a POS earning a grant over the course of a year with a pr statement of an antismoking nature that i could of done while riding the coach. Precisely the kind of shit that irritates me. You have no point to make other then an opinion full of second guesses and altered numbers? WTF

Of course it's not science. It's a blurb (too short to be called an article) in a web-based "Health" magazine. It's not written to be useful as science. It refers to a study and article published in the journal "Pediatrics" (or the Journal of Pediatrics, it's not clear, and I didn't look it up). If you're curious about how much science went on, the recent edition of "Pediatrics" might be a good place to start. Blurbs in web magazines aren't peer reviewed (at least, not in the same way).

Assuming that the quotes you grabbed were reasonable things to have written in the blurb (and not just indicative of the abysmal state of science journalism), I'm still not put off.

"The research doesn't definitively prove that tobacco smoke can harm children's brains"
"It's difficult to confirm whether secondhand smoke causes children's health problems"

That doesn't mean it's bad research. Causation is hard. Correlation is easier, and a good first step. Finding no correlation is a good indicator that there's nothing going on. Finding significant correlation is a good indicator that further study might be in order.

... and yeah, finding correlation can be easy. It happens all over the place, often for no reason that anyone can figure out. Finding an R-squared close to 1 can indicate that you've found something amazing, and it can also indicate that your model is stupid.

"After adjusting their numbers to improve their validity from a statistical point of view"
"The researchers estimated"

I can't tell what the blurb means by "adjusting", but if you're running a statistical model then adjusting your model - properly - is not only a reasonable thing to do, it's necessary. You might have to make adjustments to take into account all sorts of things. Most of these adjustments lower the perceived power of your results. It's also possible to adjust your numbers dishonestly (throwing out unfavourable data is a basic example). It's also possible to screw it up without being dishonest.

This study was based on a dataset gathered with a telephone survey. Maybe (and I'm basically guessing out my ass now) the authors weren't confident that everyone who was asked "So, do you blow cigarette smoke into the faces of your precious children?" answered the question truthfully. But people have been surveyed about their smoking habits for decades. There may be another study in the field that has figured out how often people lie about their smoking habits. You can use that to adjust your own model.

That's just an example. I have no idea what their study did. The sentence you quote; "After adjusting their numbers to improve their validity from a statistical point of view" is - to me - an indicator that whoever wrote the blurb had no clue at all about what any of that meant. In lots of studies like this, the statistical point of view is the ONLY point of view. It's a statistical study. Of course it's an estimate. It's not like the researchers are blowing smoke into a petri dish filled with brain cells and then looking under a microscope for how many normal cells have mutated into dementia cells.

So I think this is more a case of bad journalism.

If you're looking for bad science on the web, maybe try this:

Bad Science

... by Ben Goldacre. He has a column in the Guardian.

At a guess, you might enjoy this one:

Kids who spot bullshit, and the adults who get upset about it

Cheers,


AetheLove