Bad Science

Bad Science

1. Nocebo effect - people feel bad (experience something) because they're expecting to

2. Cannot base scientific evidence from 1 person as they may make mistakes i.e placebo or nocebo , even if they did have a real effect, this 1 person, not generalisable, must take lots of individuals from target population

3. Double blind trials ensure that the results are not due to anyone's expectations or biases

4. Randomisation - reduce any experimenter bias, must be mentioned that it has been done and how it has been done

5. Meta analysis can be useful for finding effects that many studies couldn't find simply due to having a too small sample size

6. Placebo - frowned , can be an easy way to get a positive result so is a cheat people use upon due to ethics as not letting someone ill have the chance to get better, many suggest to compare new treatment to the best pre-existing treatment

   • knowing you are given something to make you feel better or your parents knowing it should make you better equally cause a placebo effect
     • Hawthorne effect - simply being in a trial makes you perform better

7. Experimenter bias is major - giving drugs telling patients they believe they will work actually helps treatments, same with the opposite, even if they don't say anything mannerisms will slip out, again important to double-blind

8. Confounding variables - related to both elements of experiment and can confuse an apparently causal relationship

9. Cherry picking- picking data that suits your case

10. Using pps that are more likely to show positive results - i.e testing a drug on young people with 1 problem compared to old with lots, comparing the drug against a useless control - sugar pills dont say much, should be compared against the best currently available treatment,

    • Using different does between control and new is wrong, not asking about side effects or asking in sneaky ways
      • Ignoring people who dropped out that had bad experiences, removing outliers if they look bad or keeping them in if they're crazy but make it look good,
        • Stopping a trial at 4 1/2 months instead of 6 with great results due to worry they will change, torturing the data looking for a small reason it works
      • Assuming correlation = causation, placebo and baseline group unequal in struggles

11. Systematic reviews, as the name implies, typically involve a detailed and comprehensive plan and search strategy derived a priori, with the goal of reducing bias by identifying, appraising, and synthesizing all relevant studies on a particular topic.

    • A systematic review answers a specific research question by evaluating and summarizing all the studies on the topic. A meta-analysis goes on to use statistical methods to combine the data from the studies.

12. Inaccurate referencing

13. Regression to the mean - when things are at their extremes, they are likely to settle back downbasically another phrase for the phenomenon whereby, as alternative therapists like to say, all things have a natural cycle. Let’s say you have back pain. It comes and goes. You have good days and bad days, good weeks and bad weeks. When it’s at its very worst, it’s going to get better, because that’s the way things are with your back pain.Similarly, many illnesses have what is called a ‘natural history’: they are bad, and then they get better. As Voltaire said: ‘The art of medicine consists in amusing the patient while nature cures the disease.’ Let’s say you have a cold. It’s going to get better after a few days, but at the moment you feel miserable. It’s quite natural that when your symptoms are at their very worst, you will do things to try to get better. You might take a homoeopathic remedy. Then, when you get better you will assume whatever you did made you better - not always the case.

    • Publication bias - positive trials more likely to be published than negative, publishing several times to seem like there is lots of support - duplicate publication - , hiding harm and side effects, , advertising lies, science distortion in the media
      • Bias due to prior beliefs, prior data read or researched - death penalty example
        • Other biases - availability of info social influences/conformity, self bias, context and expectation bias,
          • Bad stats - picking the biggest number, presenting figures in a certain way