Download SAS Data Analysis: Descriptive Statistics and Confidence Intervals - Prof. Sung Eun Kim and more Exams Statistics in PDF only on Docsity! STAT 475 : Data Analysis with SAS, Note 6 Sung E. Kim, California State University-Long Beach, Dept of Math and Statistics____ 1 NOTE #6: Descriptive and Univariate Statistics I PROC MEANS; PROC MEANS <DATA=mydata> <list of statistics> <options>; VAR variable1 variable2, …; OUTPUT OUT=outdataname Statistics = variables; RUN; Mostly used Statistics in PROC MEANS: CLM Lower and Upper 95% confidence interval for mean LCLM/UCLM 95% Lower/Upper Confidence Limit for mean KURT Kurtosis MAX Maximum MEAN Average MEDIAN Median MIN Minimum N Number of observations without missing NMISS Number of observations with missings PROBT Probability of a greater absolute value for t-value P95 95th percentile (also available P1, P5, P10, P25, P50,P75, P90, P99) Q1 / Q3 25th / 75th percentile RANGE Range STD Standard deviation SUM Sum T t-test for Ho: mean = 0 VAR Variance /* Example 5-1 */ DATA Ex5_1; INPUT Class $ Gender $ Score; DATALINES; A Male 96 A Male 87 A Male 89 A Female 98 A Female 82 B Male 65 B Male 85 B Female 63 B Female 93 B Female 77 C Male 62 C Male 94 C Male 80 C Female 99 ; PROC MEANS DATA = ex5_1 ; BY Class; RUN; * To use BY statement data must be sorted by the variable; PROC SORT DATA = ex5_1 OUT=ex5_1sort; by Gender; run; STAT 475 : Data Analysis with SAS, Note 6 Sung E. Kim, California State University-Long Beach, Dept of Math and Statistics____ 2 PROC MEANS DATA = ex5_1sort MEAN N; BY Gender; RUN; PROC MEANS DATA = ex5_1 chartype; * this option will give you binary _TYPE_; Class Class Gender; *Class statement doesn't require sorted data; OUTPUT OUT = ex5_1out N = count Mean = meanscore; RUN; Proc Print DATA=ex5_1out; run; Proc Print DATA=ex5_1out (DROP = _FREQ_) ; Where _TYPE_ EQ '11'; *note _TYPE_ is a character variable; RUN; /* Example 5-2 */ DATA GNP; SET SASHELP.GNP; Year =year(date); quarter = qtr(date); PROC MEANS DATA = GNP chartype; VAR GNP CONSUMP INVEST EXPORTS GOVT; CLASS quarter; OUTPUT OUT = gnp_out /* (drop = _:) will remove all variable beginning with an underscore */ N (quarter) = count MEAN (GNP CONSUMP)= STD (GNP CONSUMP) = MAX (INVEST EXPORTS) = LCLM (GNP) = UCLM (GNP)= / autoname; run; PROC PRINT data = gnp_out heading=horizontal; RUN; DATA gnp_CI; SET gnp_out (KEEP=Quarter GNP_Mean GNP_StdDev count); DO i =1 to 5; IF _N_=i then DO; LL = GNP_Mean - TINV (.975, count-1) * GNP_StdDev/SQRT(count); UL = GNP_Mean + TINV (.975, count-1) * GNP_StdDev/SQRT(count); END; END; DROP i; RUN; Proc Print DATA = gnp_CI heading=horizontal; run; /* Example 5-3 */ /* This example simulate 100 random samples of each size 30 from N(MU, STD) and calculate 95% CI for mean for each sample */ DATA Rand_Norm; Count=100; N=30; MU=5; STD=2; seed=0; DO I=1 TO Count; DO K=1 TO N; X=MU+STD*RANNOR(seed); OUTPUT; END; END; RUN; PROC PRINT; RUN;
