raoulvanoosten

@raoulvanoosten@ecoevo.social
  • Transportation Scientist and Methodological Adviser at NDC Nederland
  • PhD in Evolutionary Ecology
  • Coping with long-term illness
  • Progressive metal, video and board games, survival/obstacle running

This profile is from a federated server and may be incomplete. Browse more on the original instance.

raoulvanoosten, to statistics

I managed to simulate data and use those to calculate power for an upcoming experiment. However, power is highly variable because there is quite some variation in the real data. In some cases I need only 16 blocks for 90% power, whereas in others not even 20 blocks is enough. How would you proceed?

I used the simr package with a generalized linear model to calculate power.

raoulvanoosten,

@DataAngler excellent idea, but a bit above my current programming skills. The simr package is easy to use, but I realize now this way uses double simulations, which is not simr's intended use. I did it this way because I needed to add a new variable to my data. Maybe simr has better ways of doing that.

raoulvanoosten,

@DataAngler thanks I'll check it out

raoulvanoosten, to statistics

The minimum effect is my power threshold so they cancel each other out. How can I do this? Preferably with linear models in r (I like emmeans and simr.

@lakens you wrote that more power is needed for minimum effects compared to null tests, so you might know.

I have asked here but gotten no response https://stats.stackexchange.com/questions/621178/power-analysis-for-minimum-effect-tests-and-good-enough-range-hypotheses

raoulvanoosten,

@lakens exactly. In my example in the forum post, I say my control is 100 mg, the minimum effect I expect of a treatment is 105 mg. So the minimum effect is 5. For a null hypothesis, I calculated the sample size (6 samples with 90% power). But I want to test the hypothesis that the effect is at least 5. So when I simulate 100 and 105, they are basically both 100

raoulvanoosten,

@lakens no wonder it has kept me up for weeks... So what is there to do?

raoulvanoosten,

@lakens I want to test the hypothesis trt - ctl > 5

raoulvanoosten,

@lakens not exactly. I expect my treatment to have a positive effect, and to be falsifiable I consider at least 5 a meaningful effect. So I test the hypotheses >5 and <5. But before I conduct the experiment, I would like to know how many samples I need for either hypothesis to be falsified. Indeed, if the true effect is exactly 5 I cannot falsify either hypothesis, which is where my issue with the power lies.

raoulvanoosten,

@lakens would the most sensible solution be to test for effects larger than zero and calculate power accordingly (for an effect of 5), together with a test for equivalence (at <5)? Otherwise I could/would calculate power for an arbitrary effect between 0 and 5.

raoulvanoosten,

@lakens so thát's a non-nil null 💡. Indeed I shifted the problem, and the core of the matter is that a null can never be falsified. Thanks, problem "solved" 🙏.

raoulvanoosten,

@lakens you mean "it is impossible to verify the null hypothesis", not "falsify", right?

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • kavyap
  • DreamBathrooms
  • thenastyranch
  • ngwrru68w68
  • cisconetworking
  • magazineikmin
  • Youngstown
  • InstantRegret
  • rosin
  • slotface
  • khanakhh
  • mdbf
  • Durango
  • megavids
  • modclub
  • tacticalgear
  • GTA5RPClips
  • normalnudes
  • osvaldo12
  • everett
  • anitta
  • ethstaker
  • tester
  • Leos
  • cubers
  • provamag3
  • lostlight
  • All magazines
    •