Statistics

Bad Numbers, Bad Statistics

I am only about two years behind on my emails. Every email folks send me is worthy of a full post, but obviously I am so far behind I’ll never catch up. But it’s all juicy material and readers need to see it.

Bad numbers

From reader Weston, a link to a paper about conspiracies. More quantifying of the unquantifiable.

Thought I’d elect my own bad science paper for the year, on if the following conspiracies weren’t true, the math shows that they would have probably been debunked by now. The 4 conspiracies chosen are the moon landing, global warming, harmful vaccinations, and the cure for cancer. That’s a lively bunch.

The author, of course, has all sorts of great citations at the beginning to understand how a conspiracy works. And then the author starts by using a failure model to say that a conspiracy will become debunked if one person talks.

Going a bit further, the paper assumes that there was a single event for which the truth about the situation is known. A moon landing is a good example of this. The moon landing, one time, either happened or it didn’t. Climate change is not a single event. Vaccinations are not single events. So not only do I think his premise is flawed, but he’s already undercut himself with his examples; He’s making an enormous amount of assumptions, and then applying to situations that do not match his assumptions. Oh, and no test data. Which is the kicker.

There’s even some wee-p values!

LINK

As an avid reader, thought you’d enjoy
-Weston

You racist

From Warren, a peer-reviewed paper on the supposed racial differences in pain assessment and treatment recommendations. The academy has race on the brain, the racists.

Subject: Candidate for Bad Science award, racial bias in prescribing pain meds

Found news releases of this article floating around on Facebook. The basic point touted in the news articles is that research claims that racial bias is one reason why blacks are treated disparately when given pain medicine. It deals with white laypeople and medical professionals (or trainees) endorsing false statements about biological differences between blacks and whites, and how that influenced their ability to rate the pain and provide recommendations for prescribing pain medicines in (just) two hypothetical cases, one white and one black patient.

Of course none of the news articles link to the original article, but it’s PNAS and behind a paywall. I would consider it a candidate for your Bad Science award.

Right off the bat, they create a “composite” of false beliefs regarding blacks by averaging together people’s survey responses for each of the false questions (which was on an ordinal scale) and then used continuous statistics with it. The differences between the groups (without any sense of uncertainty) also is suspect. There may be other major flaws with it, but I’m only a fledgling.

Here is the link to the full article: LINK

Best,
Warren

Don’t worry, be happy

Lastly, a Blog Challenge. This is from Christopher. The link is to the article “Does A ‘Happiness Gene’ Exist?”:

I’m at the end of my rope. The only way I can cope with this latest “science” nightmare is a Briggs-style deconstruction. Please consider.

LINK

The challenge is critique the paper (which you’ll have to get) in the way in which you have been taught. Then write a Guest Post which I’ll happily post.

Categories: Statistics

8 replies »

  1. I wonder if everyone involved in a conspiracy knows it. This climate hoax is perpetrated by people who are wrong headed but (one likes to believe) sincere.
    I like to believe that the oncologic community, MDs and researchers, do want to be ‘good guys’. Of course the current 3 step trial system is a dismal failure; Torquemada would have loved it, but is the harm these blind nit-wits do a conspiracy ?
    It’s always a good idea to follow the money, and there are many billions of dollars going to pointless clinical trials, but are the people running these torture camps motivated only by venality? I’m inclined to think it’s more groupthink than willing evil.

  2. The paper is pay-walled and I’m not about to waste and money on it, but from the linked article there are a few red-flags regarding it’s value.

    1. A bazillion co-authors. So many in fact that they left out and misidentified a few. One has to wonder why so many piled on. It’s been suggested that scientific value is inversely correlated to the number of authors of a paper (not proven, but anecdotally supported).
    2. Well, it’s a meta-analysis so everybody involved with the original studies seems to be credited. Briggs doesn’t like meta-analyses for the extra unidentifiable uncertainty they haul into the situation.
    3. Suspect genes are associated with “happiness” or positive feelings of well-being. Trigger weasel word there. Everything in the world is associated somehow. Wee-Ps must be involved to cement the deal.
    4. Unlikely that these emotions have been well-defined but nonetheless dubiously quantified.
    5. “The researchers have no plans to stop their search.” Rampant confirmation bias will be lurking.

  3. Conspiracy theories are based on confusing consequences with intentions.

  4. There is a difference in pain response with different races, sexes, ages, physical types, stages in life and in particular, level of anxiety or depression at a given time. Race is just one rather less interesting or useful way of determining or appreciating a person’s pain.

    All signals and clues must be absorbed and assumed at some speed and understood when diagnosing and understanding people’s pain.

    It is not done, as some grumps think, purposes for dismissal of importance or for box ticking and ignoring people. It is useful to get a result, however that might be achieved. Pain scales are a useful if cynical tool and that is my opinion.

    “So that I can tick my box, If no Pain is lying on a deserted beach pain free and 10 is having your leg scooped off without anaesthetic with a spoon, how bad is it?”

    When they say “11.5”
    It tells you quite a bit about the person you’re working with, along with how they say it.

    When you’ve asked for twenty years, patterns emerge. Race is only one and it doesn’t stand out much, but an overall ‘type’ is recognised by any clinician with experience.
    It matters in situations of mysterious, complex or chronic pain. Otherwise it is a guide as to how to handle a given individual.

    None of this can be learned in a book or with numbers.
    Nor run through a computer programme.

    Asian (in England that means Indian subcontinent) women of a certain age and men to some extent of the same ethnic group tend to ‘display’ pain outwardly more than other ethnic groups. I think this is cultural though. More westernised ladies don’t display the same behaviour. The behaviour isn’t really a true measure of their pain. It’s very complex!

    As to the silly claims about prescribing medicine based on ethnic groups, there are really a very few options regarding type and dose of analgesia or pain medication. Those decisions are not especially difficult to make or even to reverse assuming a safe does is prescribed which is a minimal for any prescribing Dr. Once diagnosis is clear, patients govern how much analgesia they require up to the limits of what is safe and assuming they don’t overuse. So the paper’s rubbish from that point of view. It assumes a mystery or difficulty over that aspect of the field.

    This is the kind of reason it’s obvious the paper’s not worth reading.
    Chanel the money back to patient care.
    “Research” has too much public money and money talks.

  5. Channel! Not Chanel, but the perfumes might be good analgesia.
    I think I got desert right. The voice thing read channel with a French accent but only after I posted it!
    Naughty blog.

Leave a Reply

Your email address will not be published. Required fields are marked *