Good Source To Learn About How To Design Experiments?

It feels like bio hacking consists of two steps:


 


1. Generate an idea about what to test (will stop eating gluten make me feel better) and


2. Design an experiment to validate it.


 


There are tons of really great ideas here in the forum. But Ive not seen much information on how to design the experiments or on the whole philosophy behind it. Bigger studies always gets butchered by science pros on their method etc. Sometimes I read the research conclusions, totally buy into it, just to have some science nerd the next day with no knowledge in the particular area tested, butcher the method and prove why the results cant be trusted.


 


Where and what should you read to improve your scientific method (not improve your ability to generate hypothesizes to test) as a bio hacker? Is there some easy beginner info on this out there you can recommend?


 


Whats even the name of this field, "scientific method"?


 


 


Comments

  • RekaReka ✭✭✭
    edited May 2014

    You have a hypothesis such as avoiding gluten makes your palms less hairy. This is a causation. Or, making it more simple: gluten makes your palms hairy. :-)


    You want to test the effect of one thing which is called the independent variable: avoiding gluten.


    The hair on your palms is the dependent variable because you assume that it depends on your gluten intake.


     


    You may have noticed that gluten avoidance and less hair goes together but you can't say for sure which one causes the other, and there may be some other background variable which makes them correlate. So you have to make changes in your independent variable while making very sure that you don't change anything else which could confound your results.


     


    The studies get criticised most often because they observe a correlation and draw the conclusion that one variable caused the other, which you can't say based on just the correlation. (Example: You can observe that in places where the asphalt on the streets tends to melt people tend to get more strokes. But the melting asphalt doesn't cause the strokes, it's the extreme heat that causes both.)


    The second reason for criticism is that they don't take care of confounding variables. Like the studies where some crappy heavily processed meat or fat is used (often in the form of a hamburger and fries which also gives you tons of processed carbs ans gluten) to prove that red meat and / or fat kills you.


    This happens in almost every "meat eater" study where the meat eaters are people who eat fast food, smoke and stress out, and the vegetarians are health conscious folks who would never touch an alcoholic drink in their lives and don't use chemicals.


     


    Check this out for more:


    https://explorable.com/research-methodology


     


     


    Edit: Oh I forgot the main point! So you cut out gluten without making any other changes to your diet. (You have to take care of what you replace it with, if anything.) Then you have to measure palm hairiness, giving enough time to the change in your independent variable to have its effect. If you notice your palms are less hairy you can even go back and increase your gluten intake while keeping everything else the same, to see if they become hairier.


    This is how you can prove the "gluten causes hairy palms" hypothesis.


    It doesn't get easier... It's you who gets better.

     

    Is your social worker in that horse?

     

    Success has a price, not a secret.

  • ACH85ACH85 ✭✭

    Nice, Reka.


     


    I'll add the whole "double-blind placebo controlled" thing.


     


    Placebo effect is when your expectation or bias causes an outcome. It's weird, but people have recovered from serious illness when they think they're taking an advanced new drug, but it's really just a sugar pill, so the placebo effect can be powerful.


     


    Also, it's hard to know if an independent variable is actually causing an outcome, or if it would have happened anyway. If you're already losing weight and I give you a diet pill, it wouldn't be honest of me to use you as my pill's weight-loss case study. Therefore, experiments often have "controls" which is a group that is monitored but does not receive treatment. But because of the placebo effect, you STILL might not know if the independent variable is causing an outcome. To address that problem, you split your experimental group into two groups: one receiving the independent variable, and one receiving a placebo variable, and you don't tell participants which group they are in. This is most commonly two groups that each get a pill, but group A receives a pill with a drug in it, and group B receives a sugar pill. This is "placebo controlled" and the participants are "blind," meaning they don't know if they are receiving a treatment. 


     


    But if the researchers expect an outcome, they could STILL influence the experiment by selectively noticing outcomes, or failing to notice outcomes in the placebo group. Therefore, even the researchers can't know which group is the experimental group and which is the control. They might know that group A gets a red pill and group B gets a blue pill, but they have a colleague decide which pills contain the drug and which are just sugar. This makes the experiment "double-blind." 


     


    But we are experimenting on ourselves, so the control has to be "whatever I was doing before I made a change" and you basically have to take a hard look at whether you are expecting a specific outcome. If you have a strong expectation, you might still be able to create a double-blind experiment by having a friend help. For example if you want to see if Brain Octane is better than MCT oil, you could design an experiment, but then have a friend put each oil into identical bottles labeled A and B and not tell you which was which until the end of your experiment. 


  • edited June 2014

    Great info Reka and ACH85!


     


    The link was exactly what I was looking for. It explains a bit more about the method.


     


    I liked the idea of asking a friend to label the mct/octance bottles to try to protect your n1 experiment against the placebo effect. Nice obvious little hack.


     


    I cant help to wonder how much of the normal studies still have placebo effects skewing the results. I mean, there must be a lot of studies where people can feel they have swalloed something else than a sugar pill. I can feel a difference when I eat a turmeric capsul very easily.

Sign In or Register to comment.