Contingency
A frequency/count
tables
Association: Pearson product moment
correlation
Both are continuous
Prediction: predict one variable using the
value of the other: simple linear correlation
Dependent
or some method of curve estimation
Variable IS:
Two dependent
variables
At least one is discrete/
Spearman s rho or Kendall s tau-b
categorical
A Measure
Association: a partial or multiple
correlation procedure
All are continuous
Prediction: predict one of the
variables using the value of the other
More than two
variables: multiple regression
dependent variables
At least one is
Non-parametric tests
discrete/categorical
© DataStep Development 2004
Descrete/ Non-parametric
categorical tests
Recode continuous variables into
categories
One or more
Analysis of Co-variance
independent variables
(ANCOVA)
are continuous
Independent variable represents
two independent groups:
Independent samples t-test
See Measuring Associations
Dependent
Variables are:
Independent variable represents
Only two values
paired cases: paired samples t-
One dependent
test
variable
Independent variable represents
before/after measure on same
One independent
case: paired samples t-test
variable
More than two One-Way Analysis of Variance
All independent variables values (One-Way ANOVA)
are categorical
Continuous
(parametric tests)
More than one
General Univariate Analysis of
independent
Variance (ANOVA)
variable
All independent variables Multivariate Analysis of
are categorical Variance (MANOVA)
Two or more
dependent
variables
One or more independent Multivariate Analysis of
variables are continuous Co-variance (MANCOVA)
© DataStep Development 2004