Exploring rnorm

pnorm_explore.R

# regarding: https://www.reddit.com/r/Rlanguage/comments/d0ytw7/issues_of_bias_with_rnorm/

# set the RNG seed so the results are reporducible
set.seed(12345)
# Generate 100 target means beteen 0 and 100 from the uniform distribution
generated_means <- runif(100, 0, 100)

# Generate 100 standard deviation values also from uniform distribution
generated_sdev <- runif(100, 1, 20)

# generate number of samples to create for each set
generated_sample_size <- runif(100, 1000, 30000)

# Generate 10000 samples from each target mean and target standard deviation
samples_vary_size <- mapply(FUN=rnorm, generated_sample_size, generated_means, 10, SIMPLIFY=FALSE)
samples_vary_sdev <- mapply(FUN=rnorm, 10000, generated_means, generated_sdev, SIMPLIFY=FALSE)

# Calculate the means of each sample
vary_size_means <- sapply(samples_vary_size, mean)
vary_sdev_means <- sapply(samples_vary_sdev, mean)

# Calculate differences betwen target_means
size_differences <- abs(generated_means - vary_size_means)
sdev_differences <- abs(generated_means - vary_sdev_means)

# plot differences between means as function of what was varied
plot(generated_sdev, sdev_differences, ylab="differences between generated and sampled mean")
plot(generated_sample_size, size_differences, ylab="differences between generated and sampled mean")

# return to usual operation
set.seed(NULL)

# t-test for difference of means for samples and generated means
vary_size_results <- mapply(FUN=t.test, x=samples_vary_size, mu=generated_means, SIMPLIFY = F)
vary_size_p_values <- sapply(vary_size_results, FUN=getElement, name="p.value")

vary_sdev_results <- mapply(FUN=t.test, x=samples_vary_sdev, mu=generated_means, SIMPLIFY = F)
vary_sdev_p_values <- sapply(vary_sdev_results, FUN=getElement, name="p.value")

# approximatedly 5 of 100 expected positives just by chance
sum(vary_sdev_p_values < 0.05)
sum(vary_size_p_values < 0.05)