The sample size calculator determines the minimum number
of contacts to include in your sample, based on what you consider
to be an acceptable margin of error. Results are reported at a 95%
confidence level.
About this task
Determining the appropriate sample size is important when
your goal is to make inferences about a group of people based on a
sample. In general, a larger sample size produces a smaller margin
of error. Use the sample size calculator to either compute the sample
size needed for a particular error bound, or to determine the error
bound for different sample sizes.
Procedure
- On the Sample tab of the Sample
process configuration dialog, click Sample Size Calculator.
The Sample Size Calculator opens.
- For Response Rate Estimate, enter
your best guess for the Minimum and Maximum response
rates that you expect from your marketing campaign.
These
two values must be percentages between 0% and 100%. The lower the
expected response rate, the larger the sample size must be to achieve
the same level of accuracy for the measured response rate.
- If you are not using a predictive model, select No
Model under Modeling Estimate.
- If you are using a predictive model, select Model
Performance, then enter percentages for the File
Depth and Cumulative Gain.
To obtain these values:
- Open the Evaluate reports in IBM®
SPSS® Modeler Advantage Enterprise Marketing Management
Edition, and open
the Testing partition.
- Select the Gains tab, and display the information
as a Table.
- Use a value from the first column (Segment) of the table as the File
Depth, to indicate the percentage of customers that you
intend to contact.
- Use the corresponding value from the last column (Cumulative Gain)
of the table as the Cumulative Gain.
The
calculator uses this information to determine the number of samples
that you need to use, based on the expected response rate and modeling
performance.
- Use either approach:
- To determine the minimum sample size based on the margin of error
that you are willing to accept: Enter a value of 0% to 100% in the Error
Bound (+ or -) field to indicate the percentage margin
of error that you are willing to accept for this sample. Then click Compute
Sample Size. The Min. Sample Size field
indicates the smallest sample that will meet the specified error bound.
A smaller Error Bound percentage requires a larger sample size. Conversely,
a larger Error Bound requires a smaller sample size. For example,
a 3% Error Bound requires a larger sample size than if you allow for
a 10% Error Bound.
- To determine the margin of error that will result from a specific
sample size: Enter a value in the Min. Sample Size field
to indicate the sample size that you plan to use, then click Compute
Error Bound. Based on the results, you can decide whether
to increase or decrease the sample size. Larger sample sizes result
in smaller Error Bounds. If the resulting Error Bound is too high,
use a larger sample size.
- After you determine the optimal sample size:
- Copy the value from the Min. Sample Size field.
- Click Done to close the calculator.
- Confirm that Specify Size By # Records is
selected.
- Paste the value into the Max. Size field
in the Sample process box.