DOC

# Comparing Measures of Association Doing Simulations in SPSS for

By Lori Coleman,2014-01-10 22:18
13 views 0
Comparing Measures of Association Doing Simulations in SPSS for

Acadia University 2005 Atlantic DLI Workshop

2001 Census Individual Public Use Microdata File

Computing Exercise

The user documentation for the 2001 Census individual public use microdata file includes some examples of estimating counts, proportions and means employing the weights used to construct this sample. This computing exercise replicates these examples to demonstrate the use of this file’s weight variable in SPSS. A special subset of all off 801,055 cases for just the ten variables in the examples in Chapter III has been prepared for this exercise. Instructions will be provided in class about loading this file.

1. Open the user documentation for the 2001 individual PUMF in an Adobe

Reader session. The location of the file will be provided in class. From the

menu, select View / Go To / Page… and enter 178. Under the heading, 2.

Usual Estimators, find Example 1.

2. Next, begin an SPSS session from the icon on the desktop or by selecting

SPSS from Start / Program / SPSS.

3. Load the 2001 individual PUMF by selecting from the Data Editor menu, File /

Open / Data and follow the directions in class to the file:

Examples2001indiv.sav. Once the file has been read by SPSS, you should

see the ten variables for the first 20 or more cases displayed in the Data

Editor. If you don’t see this, ask for assistance from your instructor.

4. Example 1 in Chapter III says, “We want to estimate the total number of

women aged 25 years and over, living in Edmonton, and for whom the

highest level of schooling attained is a master’s degree or a doctorate.”

What are the names of the variables to complete the analysis in Example 1?

Variable Name Codes to identify cases

To obtain this estimate, two steps will be used. First, we’ll create a filter

variable representing the groups identified in Example 1 and then we’ll weight

the data and produce a frequencies listing to find the answer.

5. From the SPSS Data Editor, select from the menu Data / Select Cases. In

the “Select Cases” dialogue box, click the “If condition is satisfied” radio C. Humphrey 1

Acadia University 2005 Atlantic DLI Workshop

button and then the “If” button. In the open text field enter:

cmap eq 835 and any(hlosp,13,14) and agep ge 25

Click Continue and with the “Filtered” radio button on, click OK.

It will take a few seconds for SPSS to work through 800,000 cases.

6. Next, assign the weight variable. From the Data Editor menu, select Data /

Weight Cases… In the Weight Cases dialogue box, click the radio button for

“Weight cases by frequency variable” and move WEIGHTP to the text box.

Click OK and notice in the lower right frame of the Data Editor that SPSS

reports “Weight On”.

7. To obtain the estimate, select from the menu, Analyze / Descriptive Statistics

/ Frequencies. Select SEXP and click OK.

What is the weighted frequency for Females?

Does this figure match the answer in the user documentation?

8. Turn now to Example 2: “We want to estimate the proportion of individuals

living in the Montréal census metropolitan area (CMA) who are immigrants.”

What are the names of the variables to complete the analysis in Example 2?

Variable Name Codes to identify cases

9. From the Data Editor, we need to redefine the filter variable. Select Data /

Select Cases… and change the IF statement to:

CMAP eq 462

Click Continue and OK.

10. To obtain the proportion, select from the menu, Analyze / Descriptive

Statistics / Frequencies. Click RESET to clear SEXP from the variable list

and then select IMMPOPP and click OK.

What is the weighted percentage for Immigrant Status equal to “Permanent

residents: immigrants”?

C. Humphrey 2

Acadia University 2005 Atlantic DLI Workshop

Divide this percentage by 100 produces the proportion being sought in

Example 2. Does this figure match the answer in the user documentation?

11. Next is Example 3: “We want to estimate the following proportion: out of all

males aged 20 to 45 living in the Vancouver CMA, the proportion whose legal

marital status is divorced.”

What are the names of the variables to complete the analysis in Example 3?

Variable Name Codes to identify cases

12. From the Data Editor, we need to redefine the filter variable. Select Data /

Select Cases… and change the IF statement to:

CMAP eq 933 and SEXP eq 2 and (AGEP ge 20 and AGEP le 45)

Click Continue and OK.

13. To obtain the proportion, select from the menu, Analyze / Descriptive

Statistics / Frequencies. Click RESET to clear IMMPOPP from the variable

list and then select MARSTLP and click OK.

What is the weighted percentage for Marital Status equal to “Divorced”?

Divide this percentage by 100 produces the proportion being sought in

Example 3. Does this figure match the answer in the user documentation?

14. Example 4 is: “We want to estimate the total average income of women aged

15 years and over living in Ontario.”

C. Humphrey 3

Acadia University 2005 Atlantic DLI Workshop

Which variables are needed to provide this estimate?

Variable Name Codes to identify cases

15. From the Data Editor, we need to redefine the filter variable. Select Data /

Select Cases… and change the IF statement to:

PROVP eq 35 and SEXP eq 1 and AGEP ge 15

Click Continue and OK.

16. To calculate the average total income, select from the menu, Analyze /

Descriptive Statistics / Descriptives. Select TOTINCP. We also want the sum

of total income. Click Options and then Sum. Then Continue and OK.

What is the weighted average for total income?

What is the weighted sum?

Does this figure match the answer in the user documentation?

Does the numerator or denominator in the documentation match SPSS output?

Which do we believe is right?

17. Example 5 is to find the number of males in Canada with a salary greater than

0 but less than \$10,000.

What are the names of the variables to complete the analysis in Example 2?

Variable Name Codes to identify cases

18. From the Data Editor, we need to redefine the filter variable. Select Data /

Select Cases… and change the IF statement to:

C. Humphrey 4

Acadia University 2005 Atlantic DLI Workshop

WAGESP gt 0 and WAGESP lt 10000

Click Continue and OK.

19. To calculate the average total income, select from the menu, Analyze /

Descriptives Statistics / Frequencies. Click RESET to clear MARSTLP from

the variable list and then select SEXP and click OK.

What is the weighted frequency of males?

Does this match the frequency in the user documentation?

C. Humphrey 5

Report this document

For any questions or suggestions please email
cust-service@docsford.com