5 ms·

It is naive postulation to dismiss the paper for not addressing a variable, when you have not even read the paper to know whether they addressed the variable.

by elif 12y ago

It is naive postulation to dismiss the paper for not addressing a variable, when you have not even read the paper to know whether they addressed the variable. Further, your cursory read of the abstract somehow missed the clause "after controlling for socioeconomic status"

michael_h 12y ago
I think the 'controlling for socioeconomic status" was just between the states, not an all encompassing control of economic status. Two states with a similar rise in economic activity and ADHD wouldn't fail afoul of the control. Maybe. This article might hold up, but my past experience with statistics in medical papers does not give me great hope. For a while, my job was to review the math in interesting papers (well, interesting to who ever flagged it) from medical journals. In the three years that I did this, I encountered exactly zero that didn't commit some grave statistical sin.
- elif 12y ago
  It would be a statistical sin to determine that after seeing N papers with an error, that the N+1th would certainly have an error. Extrapolating the specific error without even seeing the paper is closer to rhetoric than science.
  - michael_h 12y ago
    Of course you're right, which is exactly why I didn't write that. Again: the paper might be correct, but past experience has taught me to be skeptical.
    - throwawayaway 12y ago
      https://news.ycombinator.com/item?id=9156257 https://news.ycombinator.com/item?id=9156257 I was bothered by this study, and particularly this bit: "After adjusting for sociodemographics, other drug use and childhood depression, we found no significant associations between lifetime use of psychedelics and increased likelihood of past year serious psychological distress, mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety." It seems they test life time usage, yet only "past year" serious psychological distress, mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety. Can I get your opinion on whether that "past year" applies to just "serious psychological distress" or all the terms, including: "mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety". It would seem to me that "past year" applies to all of the terms. In my opinion the study misses the rather obvious problems that cause somebody to quit psychedelics permanently, allowing them to recover. Those that quit long ago would not report issues because they didn't happen in the immediate 12 month period previous. Also acquaintances of mine who have ended up permanently hospitalised, are obviously not going to be part of a survey like this. Do you agree or disagree with any of that?
      - michael_h 12y ago
        I took a quick look. From the paper: >Past year mental health indicators new to this study were suicide thoughts, suicide plan...[etc] So, it looks like 'past year' applies to all of the terms. I think the study was looking at the long term effects, so if people who quit long ago didn't report any issues, that would support the thesis that 'taking psychedelics doesn't affect long term mental health'. I'm generally not a fan of studies about illegal activity that use voluntary surveys, but this one doesn't look poorly run. However, once you start putting maths to psychology, things get weird very quickly and it's difficult to tell what's valid and what isn't.
        throwawayaway 12y ago
        Thank you for taking the time. I disagree, I don't think they mention "long term" mental health anywhere. They just say mental health without qualifying it. In fact they only mention "long-term" benefits. I find it really objectionable that it's titled: "Psychedelics not linked to mental health problems or suicidal behavior: A population study" while looking at whole lifetime usage and comparing with past year effects. If there was suicidal behaviour("successful" or otherwise) and mental health problems in the immediate aftermath of taking psychedelics over a lifetime, they don't really have any way of capturing that information unless it's happened in the last 12 months. We agree that the "past year" applies to all of the terms. Just seems really irresponsible to me, but I'm clearly biased.
  - throwawayaway 12y ago
    no it wouldn't. it would be a good guess based on his experience. not a statistical sin at all, the events are related, the guys doing the work make mistakes all the time and the odds they get it right this time are low.
- elif 12y ago
  "Hierarchical and multivariate regression analyses were conducted to examine the relationship between artificial water fluoridation prevalence and ADHD prevalence after controlling for natural water fluoridation prevalence and SES, and SES respectively." ... "Moreover, results did not appear to be confounded by socioeconomic status because they remained consistent after controlling for this variable." ... "Moreover, artificial water fluoridation prevalence [compared with SES] even appeared to be the more robust predictor." Even if you have a problem with the stat analysis (i don't), The fact that it's a stronger predictor than SES is quite damning regardless of controls.
civilian 12y ago
But socioeconomic status might not be the confounding variable. It could be something more like "civilization" or "the amount of pro-health resources in the community". Fluoride and child doctors are both things that contribute to health, and different communities make different decisions about how much importance they want to place on childhood health. I also suspect that fluoridation does not track with socioeconomic status. I mean, look at how Portland Oregon's got rid of fluoridation. Nevertheless, reading the article, it does look like they have rat studies that show that fluoride in rats causes mental impariment.
- warfangle 12y ago
  I wonder if this is the rat study: http://www.sciencedirect.com/science/article/pii/S0378427408000647 http://www.sciencedirect.com/science/article/pii/S0378427408... Things to note: they soaked cultures of rat hippocampus neurons in various concentrations of sodium fluoride. They didn't make the rats drink fluorinated water. It's difficult for fluoride to cross the blood brain barrier.
  - civilian 12y ago
    From the article > Fluoride can readily cross the placenta, accumulate in the infant brain and easily exert neurotoxic effects, such as decreasing norepinephrine in the parietal and occipital lobes, decreasing serotonin in the parietal lobe and increasing serotonin in the frontal and occipital lobes [42-45]. So maybe it passes the blood-brain barrier more easily in infants or in utero? Also if you search for "rats' in the original article, http://www.ehjournal.net/content/pdf/s12940-015-0003-1.pdf http://www.ehjournal.net/content/pdf/s12940-015-0003-1.pdf then you'll see them referencing studies that has rats drink or be injected with fluoride. They weren't referencing the study you linked.
  - DennisP 12y ago
    You don't have to wonder. You can follow the link right there in the Newsweek article, and find it's a different study. Here it is for your convenience: http://www.ncbi.nlm.nih.gov/pubmed/7760776 http://www.ncbi.nlm.nih.gov/pubmed/7760776 From the abstract: "For prenatal exposures, dams received injections (SC) of 0.13 mg/kg NaF or saline on gestational days 14-18 or 17-19. Weanlings received drinking water containing 0, 75, 100, or 125 ppm F for 6 or 20 weeks, and 3 month-old adults received water containing 100 ppm F for 6 weeks."
rcthompson 12y ago
I didn't miss that clause. Controlling for SES is not trivial. And they only controlled for it at the state level. I'm just saying that the variables they measured seem like the kind of ones that are strongly correlated with economic development, so looking for a week association between the variables themselves is like looking for a needle in a haystack. Even if you eliminate 90% of the haystack (i.e. control for most, but not all, of SES effects), finding the needle doesn't get much easier. Or looking at it another way, my point is that since most correlations like this turn out to both have economic development as a common cause rather than being themselves causally related, I require extraordinary evidence to the contrary before I believe it.

by elif 12y ago

michael_h 12y ago

I think the 'controlling for socioeconomic status" was just between the states, not an all encompassing control of economic status. Two states with a similar rise in economic activity and ADHD wouldn't fail afoul of the control. Maybe. This article might hold up, but my past experience with statistics in medical papers does not give me great hope. For a while, my job was to review the math in interesting papers (well, interesting to who ever flagged it) from medical journals. In the three years that I did this, I encountered exactly zero that didn't commit some grave statistical sin.

elif 12y ago
It would be a statistical sin to determine that after seeing N papers with an error, that the N+1th would certainly have an error. Extrapolating the specific error without even seeing the paper is closer to rhetoric than science.
- michael_h 12y ago
  Of course you're right, which is exactly why I didn't write that. Again: the paper might be correct, but past experience has taught me to be skeptical.
  - throwawayaway 12y ago
    https://news.ycombinator.com/item?id=9156257 https://news.ycombinator.com/item?id=9156257 I was bothered by this study, and particularly this bit: "After adjusting for sociodemographics, other drug use and childhood depression, we found no significant associations between lifetime use of psychedelics and increased likelihood of past year serious psychological distress, mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety." It seems they test life time usage, yet only "past year" serious psychological distress, mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety. Can I get your opinion on whether that "past year" applies to just "serious psychological distress" or all the terms, including: "mental health treatment, suicidal thoughts, suicidal plans and suicide attempt, depression and anxiety". It would seem to me that "past year" applies to all of the terms. In my opinion the study misses the rather obvious problems that cause somebody to quit psychedelics permanently, allowing them to recover. Those that quit long ago would not report issues because they didn't happen in the immediate 12 month period previous. Also acquaintances of mine who have ended up permanently hospitalised, are obviously not going to be part of a survey like this. Do you agree or disagree with any of that?
    - michael_h 12y ago
      I took a quick look. From the paper: >Past year mental health indicators new to this study were suicide thoughts, suicide plan...[etc] So, it looks like 'past year' applies to all of the terms. I think the study was looking at the long term effects, so if people who quit long ago didn't report any issues, that would support the thesis that 'taking psychedelics doesn't affect long term mental health'. I'm generally not a fan of studies about illegal activity that use voluntary surveys, but this one doesn't look poorly run. However, once you start putting maths to psychology, things get weird very quickly and it's difficult to tell what's valid and what isn't.
      - throwawayaway 12y ago
        Thank you for taking the time. I disagree, I don't think they mention "long term" mental health anywhere. They just say mental health without qualifying it. In fact they only mention "long-term" benefits. I find it really objectionable that it's titled: "Psychedelics not linked to mental health problems or suicidal behavior: A population study" while looking at whole lifetime usage and comparing with past year effects. If there was suicidal behaviour("successful" or otherwise) and mental health problems in the immediate aftermath of taking psychedelics over a lifetime, they don't really have any way of capturing that information unless it's happened in the last 12 months. We agree that the "past year" applies to all of the terms. Just seems really irresponsible to me, but I'm clearly biased.
- throwawayaway 12y ago
  no it wouldn't. it would be a good guess based on his experience. not a statistical sin at all, the events are related, the guys doing the work make mistakes all the time and the odds they get it right this time are low.
elif 12y ago
"Hierarchical and multivariate regression analyses were conducted to examine the relationship between artificial water fluoridation prevalence and ADHD prevalence after controlling for natural water fluoridation prevalence and SES, and SES respectively." ... "Moreover, results did not appear to be confounded by socioeconomic status because they remained consistent after controlling for this variable." ... "Moreover, artificial water fluoridation prevalence [compared with SES] even appeared to be the more robust predictor." Even if you have a problem with the stat analysis (i don't), The fact that it's a stronger predictor than SES is quite damning regardless of controls.

civilian 12y ago

But socioeconomic status might not be the confounding variable. It could be something more like "civilization" or "the amount of pro-health resources in the community". Fluoride and child doctors are both things that contribute to health, and different communities make different decisions about how much importance they want to place on childhood health. I also suspect that fluoridation does not track with socioeconomic status. I mean, look at how Portland Oregon's got rid of fluoridation. Nevertheless, reading the article, it does look like they have rat studies that show that fluoride in rats causes mental impariment.

warfangle 12y ago
I wonder if this is the rat study: http://www.sciencedirect.com/science/article/pii/S0378427408000647 http://www.sciencedirect.com/science/article/pii/S0378427408... Things to note: they soaked cultures of rat hippocampus neurons in various concentrations of sodium fluoride. They didn't make the rats drink fluorinated water. It's difficult for fluoride to cross the blood brain barrier.
- civilian 12y ago
  From the article > Fluoride can readily cross the placenta, accumulate in the infant brain and easily exert neurotoxic effects, such as decreasing norepinephrine in the parietal and occipital lobes, decreasing serotonin in the parietal lobe and increasing serotonin in the frontal and occipital lobes [42-45]. So maybe it passes the blood-brain barrier more easily in infants or in utero? Also if you search for "rats' in the original article, http://www.ehjournal.net/content/pdf/s12940-015-0003-1.pdf http://www.ehjournal.net/content/pdf/s12940-015-0003-1.pdf then you'll see them referencing studies that has rats drink or be injected with fluoride. They weren't referencing the study you linked.
- DennisP 12y ago
  You don't have to wonder. You can follow the link right there in the Newsweek article, and find it's a different study. Here it is for your convenience: http://www.ncbi.nlm.nih.gov/pubmed/7760776 http://www.ncbi.nlm.nih.gov/pubmed/7760776 From the abstract: "For prenatal exposures, dams received injections (SC) of 0.13 mg/kg NaF or saline on gestational days 14-18 or 17-19. Weanlings received drinking water containing 0, 75, 100, or 125 ppm F for 6 or 20 weeks, and 3 month-old adults received water containing 100 ppm F for 6 weeks."

rcthompson 12y ago

I didn't miss that clause. Controlling for SES is not trivial. And they only controlled for it at the state level. I'm just saying that the variables they measured seem like the kind of ones that are strongly correlated with economic development, so looking for a week association between the variables themselves is like looking for a needle in a haystack. Even if you eliminate 90% of the haystack (i.e. control for most, but not all, of SES effects), finding the needle doesn't get much easier. Or looking at it another way, my point is that since most correlations like this turn out to both have economic development as a common cause rather than being themselves causally related, I require extraordinary evidence to the contrary before I believe it.