This is my new favourite video on youtube! I've bene struggling to get these concepts right for so long - you've put them in such a clear and logical way! Thank you so much!!
Hi Courtney, I recently shared your video with a statistics class that I'm currently taking because your explanation of the concepts is very good! However, I was told by my professor that in 6 minutes 13 seconds of the video the effect size of ANOVA use eta was indicated, but it should be eta squared. Again, your explanation of the concepts impacted my understanding to the degree that I shared the video with my class! Therefore, great job and thank you for the clarity of the concepts!!!
I've been doing Monte Carlo methods to estimate power of studies that are already published but did not report power. I don't see that as a waste of time because it helps me evaluate whether the study was likely to have missed an effect that other studies detected due to low power.
R square and Cohen's d has the opposite spectrum when they come to indicating effect size, where R square is negatively related to effect size, and Cohen's d is quite positively related.
There might be a small mistake in one of the slides. At <a href="#" class="seekto" data-time="360">6:00</a>, first it's written as Cohen's "d" for small and then switches to (Pearson's ) "r" for medium and large. Thank you! Have my "Like"
When calculating the effect size do you use the number of participants enrolled or the intent to treat population numbers? For example there were 53 patients randomized to receive treatment X however two of those patients were exluded because of an incomplete baseline assessent. Do I use the number of patients initially enrolled (53) or the number of patients (51) after exclusion?
I don't get it. How do you compute power BEFORE you do the study? I cannot compute p-values for example without my data (since I can't compute the mean). Therefore, how do I actually compute the power before the study starts (before I collect data)? Is it just with simulated data and then using that and estimates of the means, stds, guess a sample size that is needed?