Analytical techniques enable researchers to examine complex relationships between variables. There are three basic types of analytical techniques:

Regression analysis assumes that the dependent, or outcome, variable is directly affected by one or more independent variables. There are four important types of regression analyses:

**Ordinary least squares (OLS) regression**

- Used to determine the relationship between a dependent variable and one or more independent variables
- Used when the dependent variable is continuous. For example, if the dependent variable was family child care expenses, measured in dollars, OLS regression would be used

**Logistic regression**

Used when the dependent variable is
dichotomous, or has only two potential outcomes. For example, logistic regression
would be used to examine whether a family uses child care subsidies

Visit the following websites for more information about OLS and logistic regression:

**Hierarchical linear modeling**

- Used when data are nested. Nested data occur when several individuals belong to the same group under study. For example, in child care research, many children are cared for by the same child care provider and many child care providers work within the same state. The children are nested in the child care provider and the child care provider is nested in the state
- Allows researchers to determine the effects of characteristics for each level of nested data, child care providers and states, on the outcome variables

**Duration models**

Used to estimate the length of a status
or process. For example, in child care policy research, duration models have
been used to estimate the length of time that families receive child care subsidies.

Visit the following website for more information about duration models:

- Box-Steffensmeier, J. and Zorn, C.J.W (2002), "Duration models for repeated events," The Journal of Politics 64(4).

Glossary terms related to regression analysis:

Adjusted R-Squared

Alpha Level

Coefficient of Determination

Degrees of Freedom

Dependent Variable

Effect Size

Error Term

Heteroskedastic

Independent Variable

Indirect Effect

Interaction Effect

Intercept

Least Squares

Linear Regression

Logistic Regression

Main Effect

Multicollinearity

Ordinary Least Squares Estimation

Outlier

Parameter

Predictor Variable

Regression Analysis

Regression Coefficient

Regression Equation

R-Squared

Significance Level

Simple Linear Regression

Slope

Standard Error

Standardized Variables

Statistical Significance

T-test

Type I Error

Type II Error

Z-test

Grouping methods are techniques for classifying observations into meaningful categories.
One grouping method, **discriminant analysis**, identifies characteristics that distinguish
between groups. For example, a researcher could use discriminant analysis to determine which characteristics
identify families that seek child care subsidies and which identify families that do not.

Visit the following website for more information about discriminant analysis:

The second grouping method, **cluster analysis**, is used to classify similar individuals
together. For example, cluster analysis would be used to group together families who hold
similar views of child care.

Visit the following websites for more information about cluster analysis:

Glossary terms related to grouping methods:

Cluster Analysis

Discriminant Analysis

Exploratory Study

This meeting centered on innovative methods for conducting subgroup analysis and discussions of guidelines for interpretation and reporting of subgroup analyses in prevention and intervention research.

Multiple equation modeling, which is an extension of regression, is used to examine the causal pathways from independent variables to the dependent variable. There are two main types of multiple equation models:

- Path analysis
- Structural equation modeling

**Path analysis**

Allows researchers to examine multiple direct and indirect causes
of a dependent, or outcome, variable.

- A path diagram is created that identifies the routes between the independent and dependent variables
- The paths can run directly from an independent variable to a dependent variable, or they can run indirectly from an independent variable, through an intermediary variable, to the dependent variable
- The entire model is tested to determine the relative importance of each causal pathway

**Structural equation modeling**

Expands path analysis by allowing for multiple indicators
of unobserved (or latent) variables in the model.

Visit the following websites for more information about multiple equation models:

- Electronic Textbook: StatSoft
- Structural Equation Modeling (David A. Kenny)

Glossary terms related to multiple equation models: