REVIEW GUIDE FOR PA METHODS
Date: 10/23//08:
Subject to updating.
Applies to both PA 765 and PA 766.
The following topics are covered in the methods test item data bank. Topics listed are general topics. Individual test items may include reference to more detailed aspects of general topics.
Sample items are available at http://www2.chass.ncsu.edu/garson/pa765/sampleexam.htm.
RESEARCH DESIGN
purpose of residual analysis
standardizing data
levels of measurement
normal curve & confidence levels
types of ordinal scales
types of validity
threats to validity
types of reliability
reliability analysis
testing for normality
testing for homoscedasticity
testing for linearity
testing for unidimensionality
Cronbach's alpha
data transformation
data screening
skew, kurtosis
MAR, MCAR
data imputation
SIGNIFICANCE
assumptions of significance testing
significance v. power
Type I error
confidence intervals
chi-square tests
t-tests
POWER ANALYSIS
power
Type II error
how sample size is estimated with SamplePower
power analysis vs. precision analysis
CORRELATION/ PARTIAL CORRELATION
strict monotonicity
assumptions of correlation
attenuation
types of explanation & suppression
control variables/control effects
canonical correlation
REGRESSION
assumptions of regression
model specification
standardized and unstandardized b
R-square
significance of b and of model
SPSS curve estimation procedure
WLS regression
ANOVA FAMILY
assumptions of anova
homogeneity of variance
Levene's test
factors vs. covariates
within- and between-group designs
FACTOR ANALYSIS & RELATED
types of extraction
orthogonal vs. oblique rotation
communality
eigenvalues
factor loadings
confimatory factor analysis in SEM
cluster analysis
what is canonical correlation
MULTIDIMENSIONAL SCALING
distance measures
stress
testing ordinality vs. metricity
CLUSTER ANALYSIS
Euclidean distance
cluster vs. factor analysis
UPGMA
hierarchical vs. k-means vs. 2-step clustering
LOG-LINEAR ANALYSIS
assumptions
difference from logistic regression
sampling adequacy
likelihood ratio tests
parsimonious model
types of models
link function
effect size measure
odds and odds ratios
LOGISTIC REGRESSION
binomial vs. multinomial
assumptions
linearity in the logit
logits
odds
likelihood ratio test
Box-Tidwell test
significance
maximum likelihood
classification table
PARTIAL LEAST SQUARES REGRESSION
PLS regression vs. pls path analysis
PRESS statistic
factor weights and loadings
Variable importance in projection (VIP)
PROBIT REGRESSION
ORDINAL REGRESSION
NONLINEAR REGRESSION
GENERALIZED LINEAR MODEL AND GEE
distribution assumptions
link functions
common models
Poisson regression
offset variables
comparing models
COX REGRESSION AND EHA
status variable
proportional hazard assumption
dfBeta
LINEAR MIXED MODELS
random effects models
repeated measures models
HIERARCHICAL LINEAR MODELING
linear mixed models and multilevel models
SPSS v. HLM software
STRUCTURAL EQUATION MODELING
assumptions
comparison with OLS regression
comparison with path analysis
measured vs. latent variables
ordinal data and Bayesian estimation
overidentification
model development approaches
path significance tests
likelihood ratio test
modification indexes
critical ratios
assigning a metric
confirmatory factor analysis
measurement vs structural model
multiplication rule for compound paths
goodness of fit measures
multigroup models
checking cross-group invariance
mixture analysis