nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Sample Size and Power for Two (K) Means with Unequal variance

Hello,

Got a query regarding the DOE --> "Sample Size and Power".

Q1: Can it be used to calculate the sample size and power for two (or k) groups with an unequal variance? (below is sample data for 2 groups)

if Yes, then, how should I put in the "Std Dev"?

if No, is there an alternate solution for it? Thank you

Q2: There is also the "Power" option under "Fit Y by X", but it seems to use different parameters(see picture below), how to link these two platforms?

Column 1	Column 2
13.41	-25.79
21.3	-3.87
10.26	4.14
10.67	-17.02
8.68	36.3
9.46	-33.83
6.8	22.37
6	-76.82
8.53	-4.42
12.56	-42.96
4.17	-0.62
13.92	-2.74
11.62	-19.87
9.71	-0.19
6.56	-20.43
7.53	-3.19
9.58	25.3
12.99	-10.98
9.13	17.26
8.4	-17
12.02	54.65
8.31	41.79
10.95	30.24
7.74	-5.17
6.83	-13.58
8.9	-0.02
10.84	7.84
10.86	4.81
9.27	9.19
10.47	-23.64

Stay Healthy!

Accepted answers

Thank you Jian, your answer helps me to understand the delta setting ^-^
Meanwhile, I also found the jmp addin which was developed long time ago by Mark to dealing with an unequal variance situation.

All comments

jiancao

1. To use the tool you’d have to assume that variances are equal.
2. Delta in Fit Y by X is not the same as the Difference to Detect in DOE’s Sample Size and Power, but there is some relationship:Delta= Difference to Detect / Sqrt(2k), where k is number of samples. So, the Difference to Detect=2*Delta for two-sample means.

Thank you Jian, your answer helps me to understand the delta setting ^-^
Meanwhile, I also found the jmp addin which was developed long time ago by Mark to dealing with an unequal variance situation.

mia_stephens

FYI - There is a new platform in JMP 16, Sample Size Explorers, which can handle this. This platform will be in JMP 16 EA 2 (which will be available end of April).

Thank you Mia. I'd like to give it a try when available.
Have a wonderful week ^-^

jiancao

Mia,
Which one, Sample Size and Power for Two-sample Means with Unequal Variances, or Sample Size and Power for Testing Two Variances, will be added? Or both?

mia_stephens

Both - this is what it looks like now.

Power for Two Independent Sample Means includes an option for unequal standard deviations. As a bonus, there's also an option for unequal sample sizes.

Again, this will be featured in EA 2. We're actively recruiting for the EA program if you've got interested customers.