Hi DCR Students -
Week 3 in DCR is important! The SIZE of your study sample turns out to be a critical factor for determining both IMPORTANCE and FEASIBILITY (in opposing directions). So it's really important for you to know how to plan your sample size, and it's a great skill for you to develop and use in your future work.
Doing a sample size calculation builds on all the work we've done so far in the course, particularly in defining the variables for your study (predictor/intervention and outcome). If you've been working under the assumption that "my study is different..." and not quite fitting into the 1-sentence summary format and other exercises we've done so far, you're going to have trouble with your sample size calculations. Even if you do it right, you may be shocked and dismayed by the size of the study you need! Some advice:
- Make sure you watch Steve Cummings's lecture, in 3 parts, now posted on the Syllabus. It's excellent.
- For any given sample size calculation, you'll need to choose only ONE predictor and ONE outcome variable. If you have more, ignore them for the moment.
- Dichotomous (2-level) variables are the easiest to handle in sample size calculations. If you have a 3+-level variable, or a difficult continuous variable (not normally distributed), try making it into a dichotomous variable and going from there.
- It can be hard to find the standard deviation for a continuous variable. It's OK to just guess; just say what you did so it's clear.
- Your default choice for alpha should be 0.05, 2-sided. Very rarely would you choose anything else.
- If you have 2 continuous variables, try dichotomizing ONE of them - you'll find it's easier to interpret the results that way.
Whatever you do, even if you get stuck, take a crack at actually using the sample size calculator and getting a NUMBER for your sample size, before you arrive in class on Tuesday! Simplify down your study as much as you need to.
As always in DCR, it's OK to change the study design of your main project in DCR at any time!
See you in Byer's Auditorium Tuesday at 9am for a conversation with Steve Cummings.