Statistics for everyone: ch.28 proportion of two groups

easier R than SPSS with Rcmdr : Contents

ch.28 proportion of two groups

The calculation of the samples size seems quite unusual outside of the field of medicine. That’s why SPSS has no menus at all, and there are very few books related to it. (The book I wrote is out of print, and I haven’t written it since.)

Let’s start with the menu above, which I think will be used the most.

Leave the rest by default, and put the incidence rate of each of the two groups (red square).

As a result, the calculation is that 376 people are needed in both groups.

In addition, the graph allows you to guess how many more people you need to increase the power.

Now let’s change the ratio to 2.

This shows the calculation that 285 and 570 people are needed, respectively. In the end, we need 855 people in total, so we need more.

In general, 1:1 recruitment requires the fewest numbers overall, but sometimes the ratio is different. For example, if the new treatment is thought to have less pain and fewer side effects, having more patients assigned to the new treatment will help to recruit patients as a whole. Conversely, if you expect a subject’s rejection of a new treatment, it may be a good idea to assign fewer patients to the new treatment. In addition, it’s a good idea to consider other situations in advance, such as time or cost.

easier R than SPSS with Rcmdr : Contents

=================================================