IBM SPSS SamplePower 3 Easy Interface Manual

SELECT AN INTERFACE
When the program opens, select one of these options
The option at the left takes you to the classic interface. This is intended for people who are familiar with power analysis It includes a wide variety of procedures, including tests for means, proportions, correlations, survival studies, among others It includes the new module for cluster randomized trials
The option at the right takes you to the new, easy interface This is intended for people who are new to power analysis It will find sample size for six of the most common statistical tests It will walk you through the analysis step-by-step
THE NEW INTERFACE A 30 SECOND GUIDE

1. FINDING THE RIGHT PROCEDURE
The program will ask you two questions, which it will use to select the correct procedure.
2. A WORKED EXAMPLE
Once inside the procedure, select Show a worked example. The program will let you select from one of several examples, and will use this example to explain the process of finding the appropriate sample size.
3. HELP ME SET UP MY OWN STUDY

Select Help me set up my own study The program will walk you through the following Set up the screen to match the study Enter the data Find the sample size Generate a detailed report
THATS
IT!
FINDING THE RIGHT PROCEDURE

The program asks two questions, which it uses to locate the correct procedure
QUESTION 1: ONE GROUP OR TWO?
One group Your goal is to estimate a value in one group. For example, you want to estimate the proportion of people who support a candidate, or the mean rating assigned to a movie, or the mean score on a test. Two groups Your goal is to compare values in two groups. For example you want to see if men are more likely (or less likely) than women to support a specific candidate, to see if men rate a movie as being better (or worse) than women, or to see if men score higher (or lower) than women on a test.
QUESTION 2: WHAT IS THE DATA TYPE?
Categorical People will be choosing from a limited number of responses that do not follow a natural sequence. For example, Which candidate do you support?. (We cannot say that Smith is more than Jones) Ordinal People will be choosing from a limited number of responses that do follow a natural sequence. For example, Rate a movie on a scale of 1 to 4. (We can say that 4 is a higher rating than 2) Continuous Use this when people will be assigned a score (such as a test score) that follows a natural sequence.
ONE GROUP CATEGORICAL DATA

Suppose we are planning a study where people will be asked which candidate they support. Possible responses are Smith, Jones, and Undecided. Our goal is to report the proportion who say that they support Smith, and we want to report this proportion with an error margin of plus/minus 3 percentage points. What sample size do we need? Select One group and Categorical to navigate to this screen
1. 2. 3. 4. 5. 6. 7.
Click Design to set the number of categories Enter the proportion expected to support each candidate Select the response that we want to focus on in the analysis (Smith) Enter the percent missing data (0%) Enter the desired margin of error (3%) The program shows the required sample size as 1,025 Click Report to generate a detailed report
ONE GROUP ORDINAL DATA

Suppose we are planning a study where people will be asked to report their agreement with the statement that they are satisfied with your companys service. Possible responses are Disagree strongly, Disagree, Neutral, Agree, and Agree strongly. These responses will be assigned scores of 1 to 5, and we will compute the mean response. For example, a mean of 4.0 would correspond to a mean of Agree while a mean of 3.5 would correspond to a mean midway between Neutral and Agree. Our goal is to report the mean with an error margin of plus/minus 2 percentage points. If the mean is 4.0, we want to be able to say that the true mean falls in the range of 3.8 to 4.2. What sample size do we need? Select One group and Ordinal to navigate to this screen
1. 2. 3. 4. 5. 6.
Click Design to set the number of categories Enter the proportion expected to give each response Enter the percent missing data (0%) Enter the desired margin of error (2%) The program shows the required sample size as 142 Click Report to generate a detailed report
ONE GROUP CONTINUOUS DATA

Suppose we are planning a study where students will be scored on a test, where possible scores range from 200 to 800. The test is expected to have a standard deviation of 100 points in this population. Our goal is to report the mean with an error margin of plus/minus 10 points. If the sample mean is 500, we want to be able to say that the true mean falls in the range of 490 to 510. What sample size do we need?
1. 2. 3. 4. 5. 6. 7.
Click Design to set the range for the scale Enter the expected mean (500) Enter the expected standard deviation (100) Enter the percent missing data (0%) Enter the desired margin of error (10 points) The program shows the required sample size as 387 Click Report to generate a detailed report
TWO GROUPS CATEGORICAL DATA

Suppose we are planning a study where a sample of men and sample of women will be asked which candidate they support. Possible responses are Smith, Jones, and Undecided. Our goal is to report if men are more likely (or less likely) than women to support Smith. What sample size do we need? Select Two groups and Categorical to navigate to this screen
1. 2. 3. 4. 5. 6. 7.
Click Design to set the number of categories For men, enter the proportion expected to support each candidate For women, enter the proportion expected to support each candidate Select the response that we want to focus on in the analysis (Smith) Enter the percent missing data The program shows the required sample size as 388 per group Click Report to generate a detailed report
TWO GROUPS ORDINAL DATA

Suppose we are planning a study where a sample of men and sample of women will be asked to report their agreement with the statement that they are satisfied with your companys service. Possible responses are Disagree strongly, Disagree, Neutral, Agree, and Agree strongly. These responses will be assigned scores of 1 to 5, and we will compute the mean response. For example, a mean of 4.0 would correspond to a mean of Agree while a mean of 3.5 would correspond to a mean midway between Neutral and Agree. Our goal is to report if men are more satisfied with the service (or less satisfied) than women. What sample size do we need? Select Two groups and Ordinal to navigate to this screen
1. 2. 3. 4. 5. 6.
Click Design to set the number of categories For men, enter the proportion expected to give each response For women, enter the proportion expected to give each response Enter the percent missing data The program shows the required sample size as 151 per group Click Report to generate a detailed report
TWO GROUPS CONTINUOUS DATA

Suppose we are planning a study where people will be assigned at random to one of two training conditions (Control and Tutored), and then asked to take a test. Our goal is to report that one group did better than the other (that the difference is statistically significant). We expect that the mean scores for the two groups will be 500 vs. 550, with a standard deviation of 100 for each group. What sample size do we need? Select Two groups and Continuous to navigate to this screen
1. 2. 3. 4. 5. 6. 7.
Click Design to set the range for the scale Enter the expected mean for the tutored group (550) Enter the expected mean for the control group (500) Enter the common standard deviation (100) Enter the percent missing data (0%) The program shows the required sample size as 64 per group Click Report to generate a detailed report
MISSING DATA
When we are dealing with ordinal or continuous data, those who fail to respond are always included in the percent missing (which is entered at the upper right of the screen). When we are working with categorical data, by contrast, we need to consider several options. In particular, we need to be clear about the difference between Undecided and Missing. Suppose we expect that for every 100 people surveyed, we expect the following Case 1 One question we might ask is Of the people who responded, what proportion support Smith? In this case, the relevant proportion would be Smith 40 Jones 40 Undecided 15 No response (hang up the phone) 5
Smith 40 = Smith+Jones+Undecided 40+40+15

In this case, Undecided would be listed as one of the categories. Missing would be captured under percent missing. Case 2 Another question we might ask is Of the people who have selected a candidate, what proportion support Smith? In this case, the relevant proportion would be
Smith = 40 Smith+Jones 40+40

In this case, the relevant sample includes only those who have selected a candidate. Therefore, both No response and Undecided should be included under percent missing (which would be 20%). Case 3 Another question we might ask is Of all the people approached for the survey what proportion support Smith? In this case, the relevant proportion would be
Smith 40 = Smith+Jones+Undecided+No response 40+40+15+5
In this case, the relevant sample includes everyone who was approached for the survey. Therefore, both No response and Undecided should be included as valid categories, and percent missing is zero. Case 4 In some cases, we might be more interested in the Undecided than in Smith, since we want to know what proportion of the votes are in play. If the question is Of those who responded, what proportion is Undecided? the relevant proportion would be
Undecided 15 = Smith+Jones+Undecided 40+40+15

In this case (as in Case 1), Undecided would be listed as one of the categories. Missing would be captured under percent missing.

IBM SPSS SamplePower 3 Easy Interface Manual

Uploaded by

Copyright:

Available Formats

IBM SPSS SamplePower 3 Easy Interface Manual

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IBM SPSS SamplePower 3 Easy Interface Manual

Uploaded by

Copyright:

Available Formats

SELECT AN INTERFACE

When the program opens, select one of these options

THE NEW INTERFACE A 30 SECOND GUIDE

3. HELP ME SET UP MY OWN STUDY

FINDING THE RIGHT PROCEDURE

QUESTION 1: ONE GROUP OR TWO?

QUESTION 2: WHAT IS THE DATA TYPE?

ONE GROUP CATEGORICAL DATA

ONE GROUP ORDINAL DATA

ONE GROUP CONTINUOUS DATA

TWO GROUPS CATEGORICAL DATA

TWO GROUPS ORDINAL DATA

TWO GROUPS CONTINUOUS DATA

Smith 40 = Smith+Jones+Undecided 40+40+15

Smith = 40 Smith+Jones 40+40

Smith 40 = Smith+Jones+Undecided+No response 40+40+15+5

Undecided 15 = Smith+Jones+Undecided 40+40+15

You might also like