GO BACK TO ASSIGNMENT 1
 

GO BACK TO GUIDE 2 INSTEAD


 
 
ANOTHER AWFUL EXAMPLE:
TOO FEW CATEGORIES (ONLY 2)
VERY UNDESCRIPTIVE CATEGORIES THAT DON'T MAKE SENSE
In this example using the Census' Current Population Survey from 2000, the variable years of education is recoded from 1 to 20 into just two categories: eighth grade or less, and ninth grade or more (including graduate school)

How much educationally do people with less than a high school degree have in common with people who have doctorates? Yet they are in the same category in this example.





SDA 1.2: Tables

CPS Internet and Computer Use Supplement (Aug 2000)
Jan 26, 2002 (Sat 08:53 PM Eastern Standard Time)
Variables
Role Name Label Range MD
Row educ(Recoded) Demographics-highest level of school completed 1-2
Weight rweight Relative weight -.00000004-4.34921174

See how lopsided the categories are? This can happen when you collapse 16 categories down to 2.
"High" includes anyone with more than eighth grade. Do you agree? Does this category system make sense?
 
 
Frequency Distribution
Cells contain:
-Column percent
-N of cases
Distribution
educ 1 low 6.6
6,215
2 high 93.4
88,605
COL TOTAL 100.0
94,821
Summary Statistics
Mean = 1.93 Std Dev = .25 Coef var = .13
Median = 2.00 Variance = .06 Min = 1.00
Mode = 2.00 Skewness = -3.51 Max = 2.00
Sum = 183,425.79 Kurtosis = 10.33 Range = 1.00
Inference about the mean:
Std Err = .00 CV(mean) = .00
Statistics exclude missing-data and out-of-range values.
Recode for 'educ'
1 = 31-34 "low"; 2 = 35-46 "high"
Text for 'educ'
What is the highest level of school you have completed or the
highest degree you have received?(peeduca)
Text for 'rweight'
Relative weight for CPS 2000.
The weight 'pwsswgt' is scaled so that the sum of the
weights is equal to the number of valid cases.
**

Created by COMPUTE version 1.2 
on Mar 02, 2001  (Fri 10:49 AM Eastern Standard Time)

rweight = weight * 121745 / 274301983.771

Input variable:

weight: Final weight(pwsswgt)
From study: D:\SDAweb\SDACPS2000
Allocation of cases (unweighted) 
Valid cases 95,296
Cases with invalid codes on 
row variable
39,690
Total cases 134,986
CSM, UC Berkeley


Compiled and presented by Susan Carol Losh
August 26 2004
Using Netscape Composer to upload.
Thanks to John Robinson and the other folks at University of Maryland who put these archives together.
And to the people at the University of Michigan's Inter-University Consortium for Political Science Research who have even more online archives!