I want to test for effects of morph (two levels) and treatment(two levels) on 517 observations of flight option, controlling for species and individual. For that, I am assuming that option is continuous, although represented by only 3 possible outcomes.

My issue is: after reading the previous posts on power calculation, it strikes to me that a dataset with 517 observations renders so low statistical power. It might be a problem in how I am coding the test, but what could be wrong?

Here is the dataset: https://www.dropbox.com/s/quuvbjuc5o2g4y5/data%20stack.csv?dl=0

And the code

df1=read.csv("data stack.csv")
powerSim(m2,fixed("morph+treatment","z"), nsim=100)#(96.38, 100.0)

Simulating: |============================Power for predictor 'morph+treatment', (95% confidence interval):======================| 0.00% ( 0.00, 3.62)

Should I just not believe the strong detected fixed effects?

Linear mixed model fit by REML ['lmerMod']
Formula: option ~ morph + treatment + (1 | species) + (1 | individual)
Data: df1

REML criterion at convergence: 881.4

Scaled residuals: 
Min      1Q  Median      3Q     Max 
-3.1989 -0.5434 -0.0808  0.6476  2.7047 

Random effects:
Groups     Name        Variance Std.Dev.
individual (Intercept) 0.01314  0.1146  
species    (Intercept) 0.03776  0.1943  
Residual               0.29372  0.5420  
Number of obs: 517, groups:  individual, 198; species, 10

Fixed effects:
            Estimate Std. Error t value
 (Intercept)      1.60719    0.09803  16.395
 morphsnake       1.14357    0.13359   8.560
 treatmentsand   -0.36989    0.04850  -7.626

 Correlation of Fixed Effects:
           (Intr) mrphs-
 morphsnk-lk -0.695       
 treatmntsnd -0.232  0.013

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.