Power Analysis

[cjkoch]

Environment (for bugs)

• JASP version:
• OS name and version:
• Analysis: Graphing with CI or SE and effect sizes are important but it might also be helpful to include power analysis to support hypothesis tests.

[boutinb]

We should do that indeed. We set that in our roadmap.

[EJWagenmakers]

Power analyses are a little tricky. There is G* power and this program already does a lot. Perhaps we can mimic and expand its functionality but it would be a big job. As an aside, power is useful in the design stage, not in the inference stage (unless you report p< alpha instead of the exact p-value observed). See http://www.ejwagenmakers.com/2015/WagenmakersEtAlAPowerFallacy2015.pdf

Cheers, E.J.

---

Eric-Jan Wagenmakers Department of Psychological Methods, room G 0.29 University of Amsterdam, Nieuwe Achtergracht 129B Letter: PO Box 15906, 1001 NK Amsterdam Parcel: Valckenierstraat 59, 1018 XE Amsterdam

Web: ejwagenmakers.com Book: bayesmodels.com Stats: jasp-stats.org Email: EJ.Wagenmakers@gmail.com Cell: (+31) 6 45626624

“Man follows only phantoms.” Pierre-Simon Laplace, last words

---

On Thu, Apr 13, 2017 at 11:39 AM, bruno boutin notifications@github.com wrote:

We should do that indeed. We set that in our roadmap.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jasp-stats/jasp-desktop/issues/1703#issuecomment-293841997, or mute the thread https://github.com/notifications/unsubscribe-auth/AINFh2kMOj6RXMu6rbx8-qaA4DvKnyxKks5rve1cgaJpZM4Mt1r7 .

[Glitchy-Tozier]

I think this feature would be a big step in differentiating JASP from other software, like SPSS.

Also having power analysis present would really help Linux users that don't want to learn R or mess around with WINE or virtual machines just to do power analysis. (Which would make Jasp especially attractive for them)

[tomtomme]

I would like to see Power-Analysis included in JASP too (even if GPower works fine with wine on linux with no hassle)! Yes, its a huge task, but GPower development is stalled since some years and is missing some important tests. Its also english only. I would add the german translation to JASP-Power :D

And there are R-packages for power-analyises that may make it easier to integrate into JASP? Like:

power-calculator-website with open source R-Code http://powerandsamplesize.com

Package | Purpose pwr | for a variety of models webpower | for a variety of models power and ES | for a variety of models MBESS | for a variety of models PANGEA | Power ANalysis for GEneral Anova designs crossedpower | Power Analysis with Crossed Random Effects SIMR, pamm, clusterPower, nlmeU| for linear mixed models simglm | to calculate the power through simulation asypow | Power calculations via asymptotic likelihood ratio methods longpower | for longitudinal data PwrGSD | for group sequential designs pamm | for random effects in mixed models powerSurvEpi | for survival analysis in epidemiological studies powerMediation | for mediation effects in linear, logistic, Poisson, and cox regression semPower | for structural equation models (SEM) powerpkg | for the affected sib pair and the TDT (transmission disequilibrium test) design powerGWASinteraction | for interactions for GWAS gap | for case-cohort designs ssize.fdr | for microarray experiments powerEQTL | for eQTL Analysis PowerTOS | for various study designs used in equivalence studies. Superpower | for Factorial ANOVA designs

[Glitchy-Tozier]

I can't contribute much R knowledge but I definitely can help with the translating to German process.

[boutinb]

@tomtomme Maybe it could be use-case for our workshop? Not the whole power analysis, but some easy part.

[EJWagenmakers]

We want this in JASP at some point in the future, but we want to do it well and it is a pretty major extension imo.

[tomtomme]

I think this can be closed, now that the power-module is there. Obviously it misses ANOVAs and Regressions, but still the "foot is in the door" (german saying thats possibly not really translatable :D)