Statistics

Below is a comparison of the commands used to perform various statistical analyses in R, SAS, SPSS and Stata. For R functions that are not included in base R, the library() function loads the package that contains the function right before it is used. The variables gender and workshop are categorical factors and q1 to q4, pretest and posttest are considered continuous and normally distributed.

The practice data set is shown here. The programs and the data they use are also available for download here. Detailed step-by-step explanations are in the books along with the output of each analysis.

Analysis of Variance

R

myModel <- aov(posttest ~ workshop,
  data = mydata100)
summary(myModel)
pairwise.t.test(posttest, workshop)
TukeyHSD(myModel, "workshop")
plot(TukeyHSD(myModel, "workshop"))

SAS

PROC GLM;
  CLASS workshop;
  MODEL posttest = workshop;
  MEANS workshop / TUKEY;
  RUN;

SPSS

UNIANOVA posttest  BY workshop
  /POSTHOC = workshop ( TUKEY )
  /PRINT = ETASQ HOMOGENEITY
  /DESIGN = workshop.

Stata

anova posttest workshop

Correlate, Pearson

cor( mydata[3:6],
  method = "pearson",
     use = "pairwise")
cor.test(mydata$q1,
  mydata$q2, use = "pairwise")

# Again, adjusting p-values for multiple testing.
library("Rcmdr")
rcorr.adjust( mydata[3:6] )

SAS

PROC CORR;
  VAR q1-q4;
  RUN;

SPSS

CORRELATIONS
  /VARIABLES=q1 TO q4.

Stata

correlate q*

Correlate, Spearman

R

cor( mydata[3:6],
  method = "spearman",
  use = "pairwise")
cor.test(mydata$q1,
  mydata$q2,
  use = "pairwise")

# Again, adjusting p-values for multiple testing.
library("Rcmdr")
rcorr.adjust(mydata[3:6], type = "spearman")

SAS

PROC CORR SPEARMAN;
  VAR q1-q4;
  RUN;

SPSS

NONPAR CORR
  /VARIABLES=q1 to q4
  /PRINT=SPEARMAN.

Stata

spearman q*

Crosstabulation & Chi-squared

R

myWG <- table(workshop, gender)
chisq.test(myWG)

library("gmodels")
CrossTable(workshop, gender,
  chisq = TRUE,
  format = "SAS")

SAS

PROC FREQ;
  TABLES workshop*gender / CHISQ;
  RUN;

SPSS

CROSSTABS
  /TABLES=workshop BY gender
  /FORMAT= AVALUE TABLES
  /STATISTIC=CHISQ
  /CELLS= COUNT ROW
  /COUNT ROUND CELL

Stata

tab gender workshop, row col exact

Descriptive Statistics

R

summary(mydata)

library("Hmisc")
describe(mydata)

SAS

PROC MEANS;
VAR q1--posttest;

PROC UNIVARIATE; VAR q1--posttest;

SPSS

DESCRIPTIVES VARIABLES=q1 to posttest
  /STATISTICS=MEAN STDDEV VARIANCE
   MIN MAX SEMEAN.

EXAMINE VARIABLES=q1 to posttest
  /PLOT BOXPLOT STEMLEAF NPPLOT
  /COMPARE GROUP
  /STATISTICS DESCRIPTIVES EXTREME
  /MISSING PAIRWISE.

Stata

summary q*

summary q*, detail

Frequencies

R

summary(mydata)

library("Deducer")
frequencies(mydata)

SAS

PROC FREQ;
  TABLES workshop--q4;
  RUN;

SPSS

FREQUENCIES VARIABLES= workshop TO q4.

Stata

tab1 workshop gender q*

Kruskal-Wallis

R

kruskal.test(posttest ~
  workshop)

pairwise.wilcox.test(posttest,
  workshop)

SAS

PROC npar1way;
  CLASS workshop;
  VAR posttest;

SPSS

NPAR TESTS
  /K-W=posttest BY
   workshop(1 3).

Stata

kwallis q1, by(gender)

Linear Regression

R

myModel <- lm(q4 ~ q1 + q2 + q3, data = mydata100)
summary(myModel)
plot(myModel)

SAS

PROC REG;
  MODEL q4 = q1-q3;

SPSS

REGRESSION
  /MISSING LISTWISE
  /STATISTICS COEFF OUTS R ANOVA
  /CRITERIA=PIN(.05) POUT(.10)
  /NOORIGIN
  /DEPENDENT q4
  /METHOD=ENTER q1 q2 q3.

Stata

regress q4 q1-q3
lvr2plot

Sign Test

R

library("PASWR")
SIGN.test(posttest, pretest,
  conf.level = .95)

SAS

myDiff=posttest-pretest;
PROC UNIVARIATE;
  VAR myDiff;
  RUN;

SPSS

NPTESTS
  /RELATED TEST(q1 q2) SIGN
  /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
  /CRITERIA ALPHA=0.05
   CILEVEL=95.

Stata

bitest posttest > pretest

t-Test, Indepdendent

R

t.test(q1 ~ gender,
  data = mydata100)

SAS

PROC TTEST;
  CLASS gender;
  VAR q1;
  RUN;

SPSS

T-TEST
  GROUPS = gender('m' 'f')
  /VARIABLES = q1.

Stata

ttest gender=q1, unpair unequ

t-Test, Paired

R


	t.test(posttest, pretest,
  paired = TRUE)

SAS

PROC TTEST;
  PAIRED pretest*posttest;
  RUN;

SPSS

T-TEST
  PAIRS=pretest WITH
  posttest (PAIRED).

Stata

anova posttest workshop

Variance Test

R

t.test(posttest, pretest,
  paired = TRUE)

SAS

It’s built into other procedures, such as GLM.

SPSS

It’s built into other procedures, such as GLM.

Stata

robvar posttest, by(gender)

* Or...
sdtest posttest = gender

Wilcoxon Rank Sum (Mann-Whitney)

R

wilcox.test(q1 ~ gender,
  data = mydata100)

SAS

PROC NPAR1WAY;
  CLASS gender;
  VAR q1;
  RUN;

SPSS

NPTESTS
  /RELATED TEST(pretest posttest) SIGN WILCOXON.

Stata

ranksum posttest, by(gender)

Wilcoxon Signed Rank (Paired)

R

wilcox.test(posttest, pretest, paired = TRUE)

SAS

myDiff=posttest-pretest;
PROC UNIVARIATE;
  VAR myDiff;
  RUN;

SPSS

NPTESTS
  /RELATED TEST(q1 q2) WILCOXON
  /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
  /CRITERIA ALPHA=0.05 CILEVEL=95.

Stata

signrank q1 = gender

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s