Applied Design and Analysis

Syllabus, HS510a

Applied Design and Analysis

Spring 2017

Time: TuTh 5:30-6:50pm,

Location: Schneider Building, Room G-1

Instructor: Grant A. Ritter

Office: Heller Rm 268

Phone: 781-736-3872

Office Hours: 4:00pm-5:30pm TuTh

Email:

Text: Wooldridge, Jeffrey M. Introductory Econometrics: A Modern Approach 5th ed., Cengage Learning, ISBN 978-81-315-24-65-7

Prerequisite: Knowledge of basic statistics and use of statistical software (such as HS404 or its equivalent)

Course Objectives: Course continues a presentation of quantitative methods covering experimental design issues, statistical analyses, and other topics relevant to researchers in the social sciences.

Course Requirements: The course will include four problem sets to be solved using a statistical software package, plus a set of five writing assignments which together will form the framework for a proposed research project. As the Final, the student must combine the five writing assignments together and edit to produce a potential research proposal. The course is graded pass/fail. To pass the student must regularly attend class and turn in both problem sets and written assignments.

Outline of Topics (26 classes of one hour twenty minutes each):

Linear Regression Topics

Review of Probability; mean, variance, standard deviation; random variable, independence, correlation

Review of Statistics; population vs sample, sample mean, sample variance, The Central Limit Theorem

Causality

Designs for Social Science: experimental, quasi-experimental, or observational

Data Preparation and Preliminary Analyses

Bivariate Analyses

Linear regression models; OLS; reading Stata output

Interpretation of linear regression output

Inclusion of Interaction terms in linear regressions; interpretation of their estimates

Additional Diagnostics for Linear Regression Models: Goodness of fit, VIFs, tests on the residuals

The F-test for comparing nested linear regression models

Multiple Comparison Tests: Bonferroni, Dunnett, Tukey

The Chow Tests

The Linear Probability Model

Logistic Regression Topics

Introduction and Background for dichotomous dependent variable

Graphic representation of relevant empirical data

Modeling the ‘log odds’ – justification for using logit transformation

Fitting the model, Intro to maximum likelihood method of solution

Interpretation of the model estimates – the odds ratios, constructing confidence intervals

Calculating marginal effects in logistic regression

Further topics in model building – interaction terms, adding blocks of variables, comparing results

Interpreting the interaction term in logistic regression models

Assessing model fit and comparing among models - 2LLN versus AIC versus BIC, pseudo R-square

Pros and cons of logistic modeling versus linear probability modeling

Application to observational, cohort, and case-control study designs

Diagnostics: the ROC curve, concordance and discordance, Somer’s D statistic

Additional Social Science Topics

Poisson, Negative Binomial, ZIP, and ZINB models for counting measures

Mediators and Moderators

Difference in Difference Models

Matching and Propensity Score Matching