Html Version of SAS output from exemplar 5 ANALYSIS
.
To see programme that was used for it view the SAS program file.


Links in this page
Standard errors and design effects
Regression models
back to top uparrow

First some exploratory analyses

Cannabis
q85a Frequency Percent Cumulative
Frequency		 Cumulative
Percent
never used 210 58.66 210 58.66
tried once or twice 90 25.14 300 83.80
use daily 12 3.35 312 87.15
use weekly 8 2.23 320 89.39
used in last month 11 3.07 331 92.46
used more than a month ago 27 7.54 358 100.00

amphetamines
q85b Frequency Percent Cumulative
Frequency		 Cumulative
Percent
never used 281 83.14 281 83.14
tried once or twice 32 9.47 313 92.60
use daily 1 0.30 314 92.90
use weekly 1 0.30 315 93.20
used in last month 5 1.48 320 94.67
used more than a month ago 18 5.33 338 100.00

In general, would you say your health is....
genhelf Frequency Percent Cumulative
Frequency		 Cumulative
Percent
excellent 72 19.94 72 19.94
very good 139 38.50 211 58.45
good 120 33.24 331 91.69
fair 24 6.65 355 98.34
poor 6 1.66 361 100.00

Now survey nmeans with right kind of weights

The SURVEYMEANS Procedure

Data Summary
Number of Observations 361
Sum of Weights 435.174782
 
Statistics
Variable Label N Mean Std Error of Mean Lower 95%
CL for Mean		 Upper 95%
CL for Mean
sinc
sacc
genhelf		 data zone income deprivation score
data zone access deprivation score
In general, would you say your health is....		 361
361
361		 18.711681
-0.067572
2.363313		 0.694596
0.032883
0.052741		 17.345705
-0.132240
2.259594		 20.077657
-0.002904
2.467031
The MEANS Procedure

Mean squared weight gives value design effect would have if variables unrelated to weight

Variable Label N Mean Std Dev Minimum Maximum
weight
wtsq		 weight
 		 361
361		 1.0007474
1.0974063		 0.3101247
0.7729044		 0.6304000
0.3974041		 2.5298242
6.4000104

Now actual design factors for these variables (See SAS code for how calculated - a bit tiresome. design effect for income deprivation is slightly larger than the calculation from the mean squared weight, as we might expect since we know response depended on this..

 
Obs dfsinc dfsacc dfgenhelf desinc desacc degenhelf
1 1.14547 1.01905 1.06833 1.31211 1.03846 1.14133
back to top uparrow
The SURVEYREG Procedure
Regression Analysis for Dependent Variable genhelf

Data Summary
Number of Observations 338
Sum of Weights 338.60539
Weighted Mean of genhelf 2.34349
Weighted Sum of genhelf 793.51678
 
Fit Statistics
R-square 0.09035
Root MSE 0.8884
Denominator DF 337
 
Class Level Information
Class Variable Label Levels Values
living composition of household 5 couple couple w kids other sing parent single
q85a cannabis 6 never used tried once or tw use daily use weekly used in last mon used more than a
q85b amphetamines 6 never used tried once or tw use daily use weekly used in last mon used more than a
 
ANOVA for Dependent Variable genhelf
Source DF Sum of Squares Mean Square F Value Pr > F
Model 11 25.5985 2.327137 2.94 0.0010
Error 326 257.7389 0.790610    
Corrected Total 337 283.3374      
 
Tests of Model Effects
Effect Num DF F Value Pr > F
Model 11 22.44 <.0001
Intercept 1 361.12 <.0001
sinc 1 6.09 0.0141
q85a 5 3.11 0.0093
q85b 5 15.71 <.0001

  The denominator degrees of freedom for the F tests is 337.

 

Estimated Regression Coefficients
Parameter Estimate Standard Error t Value Pr > |t|
Intercept 2.3111636 0.25331000 9.12 <.0001
sinc 0.0134122 0.00543692 2.47 0.0141
q85a never used -0.1489457 0.17464000 -0.85 0.3943
q85a tried once or tw -0.1176441 0.17512565 -0.67 0.5022
q85a use daily 0.1516325 0.23036159 0.66 0.5108
q85a use weekly 0.1475498 0.30360484 0.49 0.6273
q85a used in last mon 0.8386110 0.29410484 2.85 0.0046
q85a used more than a 0.0000000 0.00000000 . .
q85b never used -0.1427034 0.20778187 -0.69 0.4927
q85b tried once or tw -0.3035456 0.21716450 -1.40 0.1631
q85b use daily 0.6240744 0.21000573 2.97 0.0032
q85b use weekly 1.0047980 0.31863538 3.15 0.0018
q85b used in last mon 0.4212229 0.34530883 1.22 0.2234
q85b used more than a 0.0000000 0.00000000 . .

  The denominator degrees of freedom for the t tests is 337.
Matrix X'WX is singular and a generalized inverse was used to solve the normal equations. Estimates are not unique.

 
The SURVEYREG Procedure
Regression Analysis for Dependent Variable genhelf

Data Summary
Number of Observations 358
Sum of Weights 357.73923
Weighted Mean of genhelf 2.34130
Weighted Sum of genhelf 837.57366
 
Fit Statistics
R-square 0.02247
Root MSE 0.9033
Denominator DF 357
 
ANOVA for Dependent Variable genhelf
Source DF Sum of Squares Mean Square F Value Pr > F
Model 1 6.6722 6.672154 8.18 0.0045
Error 356 290.2655 0.815352    
Corrected Total 357 296.9376      
 
Tests of Model Effects
Effect Num DF F Value Pr > F
Model 1 9.18 0.0026
Intercept 1 1619.80 <.0001
canscore 1 9.18 0.0026

  The denominator degrees of freedom for the F tests is 357.

 

Estimated Regression Coefficients
Parameter Estimate Standard Error t Value Pr > |t|
Intercept 2.28375766 0.05674393 40.25 <.0001
canscore 0.45467341 0.15003568 3.03 0.0026

  The denominator degrees of freedom for the t tests is 357.

 
The GLM Procedure

Number of observations 361

  Due to missing values, only 358 observations can be used in this analysis.

 
The GLM Procedure
Dependent Variable: genhelf In general, would you say your health is....

Weight: gweight weight adding to population totals

Source DF Sum of Squares Mean Square F Value Pr > F
Model 1 544.02811 544.02811 8.18 0.0045
Error 356 23667.40531 66.48148    
Corrected Total 357 24211.43342      
 
R-Square Coeff Var Root MSE genhelf Mean
0.022470 348.2522 8.153617 2.341297
 
Source DF Type I SS Mean Square F Value Pr > F
canscore 1 544.0281087 544.0281087 8.18 0.0045
 
Source DF Type III SS Mean Square F Value Pr > F
canscore 1 544.0281087 544.0281087 8.18 0.0045
 
Parameter Estimate Standard Error t Value Pr > |t|
Intercept 2.283757665 0.05180502 44.08 <.0001
canscore 0.454673407 0.15894220 2.86 0.0045
 
The GLM Procedure

Number of observations 361

  Due to missing values, only 338 observations can be used in this analysis.

 

The GLM Procedure
Dependent Variable: genhelf In general, would you say your health is....

Source DF Sum of Squares Mean Square F Value Pr > F
Model 3 18.5858474 6.1952825 8.01 <.0001
Error 334 258.2336793 0.7731547    
Corrected Total 337 276.8195266      
 
R-Square Coeff Var Root MSE genhelf Mean
0.067141 38.24978 0.879292 2.298817
 
Source DF Type I SS Mean Square F Value Pr > F
sinc 1 6.59991804 6.59991804 8.54 0.0037
canscore 1 9.79435975 9.79435975 12.67 0.0004
ampscore 1 2.19156958 2.19156958 2.83 0.0932
 
Source DF Type III SS Mean Square F Value Pr > F
sinc 1 5.98174624 5.98174624 7.74 0.0057
canscore 1 6.62583762 6.62583762 8.57 0.0037
ampscore 1 2.19156958 2.19156958 2.83 0.0932
 
Parameter Estimate Standard Error t Value Pr > |t|
Intercept 2.021651909 0.08727740 23.16 <.0001
sinc 0.012601011 0.00453028 2.78 0.0057
canscore 0.492456944 0.16822134 2.93 0.0037
ampscore 0.472872905 0.28086654 1.68 0.0932