Minitab Sample Size Calculation: 1 Sample t vs Minimum Sample Size for Means

[JasonAREdwards]

Minitab Sample Size Calculation: 1 Sample t vs Minimum Sample Size for Means:
I am wondering if someone can help me out with some confusion that I have over sample size calculations.
1. Why is the sample size to calculate means so much smaller than the sample necessary to know if my mean is on target (i.e. a 1 sample t test; my mean is what it is)?
2. When using Minitab sample size estimator, it asks about confidence level, but it does not ask about power?

Using Minitab sample size estimator to calculate means I enter the following:
Standard Deviation: 0.069151
Estimate sample size
Margin of error 0.02
Minitab indicates that the sample size is 49
- What is the Power for this conclusion?

Using Minitab power and sample size for 1 sample t I enter the following:
Difference 0.02
Power Value: 0.9
Standard Deviation: 0.069151
Minitab indicates that the sample size is 128
Thank you in advance!

 

[Stijloor]

A Quick Bump!

Can someone help?

Thank you very much!!

 

[Miner]

Minitab Sample Size Calculation: 1 Sample t vs Minimum Sample Size for Means:
I am wondering if someone can help me out with some confusion that I have over sample size calculations.
1. Why is the sample size to calculate means so much smaller than the sample necessary to know if my mean is on target (i.e. a 1 sample t test; my mean is what it is)?
2. When using Minitab sample size estimator, it asks about confidence level, but it does not ask about power?

The reason for this difference is that in the first scenario, you are creating a confidence interval for the mean. A confidence interval is a range of values in which you would expect to find the true population mean. No test is being performed, therefore there is no power value.

In the second scenario, you are performing a hypothesis test. You are not only estimating the population mean but you are testing whether that estimate is different from a specified value. As soon as you add the test two things happen. Power now enters the picture, and as you are trying to obtain more information than just a simple confidence interval, more samples are required to provide this additional information.

 

[supreecha]

Power and Sample Size
1-Sample t Test

Testing mean = null (versus not = null)
Calculating power for mean = null + difference
Alpha = 0.05 Assumed standard deviation = 0.069151


Sample Target
Difference SizePower Actual Power
0.02 1280.9 0.901017


Power Curve for 1-Sample t Test

Power and Sample Size

1-Sample t Test

Testing mean = null (versus not = null)
Calculating power for mean = null + difference
Alpha = 0.05 Assumed standard deviation = 0.069151


Sample Target
Difference Size Power Actual Power
0.02 128 0.9 0.901017
0.02 96 0.8 0.800964
0.02 76 0.7 0.701608
0.02 61 0.6 0.603614
0.02 48 0.5 0.501164
0.02 37 0.4 0.402332
0.02 27 0.3 0.304498
0.02 17 0.2 0.201998
0.02 8 0.1 0.109658