SAS Data Analysis Project

Introduction

This project involves the analysis of health data using SAS. The dataset includes variables related to demographics, health habits, and cholesterol levels.

Data Preparation

The data is read in from a file named "sas13.dat" using SAS infile statement. Then, new variables are created based on existing variables to facilitate analysis.

/* Read in data */
data DataHealth;
infile "sas13.dat";
input ID 1-4
SEX 6
AGE1 8-9
MARITAL1 12
CIGARET1 13
CHOLESTEROL1 15-18
WTKG1 20-23
HTCM1 24-27
CHOLESTEROL2 29-32;
run;

/* Create new variables */
data DataHealthNew;
set DataHealth;

/* Variable creation code */
...

run;

Descriptives & Frequencies

Descriptive statistics and frequency analysis are conducted to understand the distribution of variables.

/* Descriptives & frequencies */
proc means data=DataHealthNew maxdec=1; run;

proc freq data=DataHealthNew;
tables SEX CIGARET1 MARSTAT1 NEVMAR MARRIED WIDDIVSEP HICHOL1 HICHOL2;
run;

Scatterplot Matrix & Boxplots

Scatterplot matrix and boxplots are created to visualize relationships between variables.

/* Scatterplot Matrix & Boxplots */
proc sgscatter data=DataHealthNew (keep=CHOLESTEROL1 AGE1 BMI1);
matrix CHOLESTEROL1 AGE1 BMI1 /diagonal=(histogram normal);
run;

proc sgplot data=DataHealthNew (keep=CHOLESTEROL1 SEX);
vbox CHOLESTEROL1 /category=SEX;
run;

proc sgplot data=DataHealthNew (keep=CHOLESTEROL1 MARSTAT1);
vbox CHOLESTEROL1 /category=MARSTAT1;
run;

Statistical Tests

T-tests, contingency tables, correlation analysis, multiple linear regression, logistic regression, and comparison of means are performed to analyze relationships and make inferences.

/* T-test of CHOL1 between male and female */
proc ttest data=DataHealthNew (keep=CHOLESTEROL1 SEX) ci=equal;
class SEX;
var CHOLESTEROL1;
run;

/* Contingency Tables */
proc freq data=DataHealthNew (keep=AGEDECADE HICHOL1);
tables AGEDECADE * HICHOL1 / expected nocol chisq;
run;

/* Correlation, MLR, and Plots */
proc corr data=DataHealthNew (keep=CHOLESTEROL1 AGE1 BMI1) nomiss;
var CHOLESTEROL1 AGE1 BMI1;
run;

proc reg data=DataHealthNew;
model CHOLESTEROL1 = AGE1 FEMALE BMI1 MARRIED WIDDIVSEP;
run;

/* Compare the means for matched samples */
proc ttest data=DataHealthNew (keep=CHOLESTEROL1 CHOLESTEROL2);
paired CHOLESTEROL2 * CHOLESTEROL1;
run;

proc ttest data=DataHealthNew (keep=CHOLESTEROL1 CHOLESTEROL2 AGEDECADE);
where AGEDECADE = 5;
paired CHOLESTEROL2 * CHOLESTEROL1;
run;

/* Comparison of proportions for matched samples */
proc freq data=DataHealthNew (keep=HICHOL1 HICHOL2);
tables HICHOL1 * HICHOL2 /norow nocol agree;
run;

/* Logistic Regression */
proc logistic data=DataHealthNew;
class MARSTAT1 (ref='1') /param=reference;
model HICHOL1 (event='1') = FEMALE AGE1 BMI1 MARSTAT1;
run;

Conclusion

The analysis provides insights into the relationship between various factors and cholesterol levels in the dataset.