This course has a modular structure. There are seven modules, each of which consists of four lectures of which up to two can be computer sessions.
To complete the course, students have to follow Module 1, plus three of the other six modules. Of those three other modules, one might be obliged for students participating in a specific master specialization.
For each module one assignment has to be made that counts for a restricted part of the final grade. These assignments are in most cases group work by two or three students.
The modules are spread over the week in such a way that more than one module can be followed in the same week. The class is spread across two blocks.
Some of the lectures may deal with the mathematical derivation of the estimators in order to enhance the understanding of these estimators. We will also teach you to how to use R and work with R syntax files for implementing the various estimators.
COURSE PROGRAMME
WEEK 1, Block 1
Module 1: Regression analysis and using R
In this module, the course is introduced and student's knowledge of the Ordinary Least Squares (OLS) regression model is refreshed. The first lecture focuses on the application of OLS in cross-national research. Because the number of countries that can be used in cross-national analyses is restricted, the results are relatively strongly influenced by deviations of the OLS assumptions and by data issues. Residual analysis is discussed as an instrument for checking the assumptions, and additional correction methods are given. In the second lecture, the use of dummy variables, handling nonlinearity, and interaction analysis are discussed.
In the computer sessions, the students work with the statistical program R, including the use of R syntax files. Students have two apply the knowledge gained in the module in a regression analysis assignment.
Literature to be announced.
WEEKS 2-5, Block 1
Module 2: Time series analysis
Economists often work with data collected for several points in time. This repeated characteristic of the data has major advantages: it is both useful to understand complex relationships and to forecast future outcomes. However, time-series data generally comes with methodological pitfalls, which are in violation with the standard OLS analysis. The following violations will be discussed: autocorrelation, non-stationarity and time-varying volatility. Furthermore, the focus is on techniques to deal with such dynamic data (e.g. VEC, ARCH).
During the computer sessions students apply the obtained knowledge in a time series assignment.
Literature to be announced.
Module 3: Experimental methods
In this module, students learn the methodological basis of incentivized experiments in economics. In the first week, two interactive lectures will be given on the experimental method and experimental design, including introduction to statistical methods relevant for experimental data (nonparametric tests). The advantages and disadvantages of experimental data will be discussed in contrast with other sources of data (happenstance, natural experiments, and questionnaires). Students will actively participate by discussing pre-class reading assignments. In week two, students of each Master’s track will analyze a track-specific academic articles, applying the method of experimental economics. In two workgroup meetings, students will be challenged to identify the contribution of the experimental method to a research question within the scope of their Master’s track, and respond critically to the contribution assessments made by their peers. At the end of the module, students will have become acquainted with basic aspects of the experimental method, they will be able to identify the position of the method vis-a-vis other empirical methods, and they will understand the contributions of the experimental data to the research questions addressed in their Master’s track.
Literature to be announced.
WEEKS 6-7, Block 1 and WEEKS 1-2, Block 2
Module 4: Treatment effects & Endogeneity
In many economic problems, as is the case with supply and demand, variables are simultaneously determined. Consequently, the explanatory (independent) variables are related to the error term in the equation of interest. This correlation may lead to biased estimates. Various techniques have been suggested to correct for the resulting bias. The by far most frequently used method is the application of instrumental variables (IV) in two-stage least squares (2sls) or three stage least squares regressions (3sls). The first week’s lecture of this module is dedicated to these instrumental variables methods.
Often one wants to know whether a certain policy has had the expected effect. For example, whether an IMF loan has led to an inflow of new private capital, or whether schooling has increased the participants’ change of getting work. In medical terms one could say that one is interested in the question whether the treatment had any affect. However, the persons who got the treatment were ill and thus do not constitute a random sample from the population. In order to get a correct estimate for the treatment effect, one has to correct for the fact that the treated differed from the non-treated. In the second week of this module, methods are discussed for estimating these treatment effects.
During the computer sessions two assignments have to be made in which the student’s apply the obtained knowledge in a combined instrumental variables and treatment effect assignment.
Literature to be announced.
Module 5: Qualitative methods
The purpose of this module is to teach students how to conduct qualitative research, specifically case study research, that will lead to high-quality master theses. Students will be introduced to case study research by comparing positivistic and interpretive approaches to qualitative research. Topics covered includes case selection, data collection and data analysis. This course module will be a mixture of theoretical and philosophical insights, assignments and class participation. You must complete three assignments in pairs.
Literature to be announced.
Module 5x: Qualitative methods for political science students
In this module, students will become acquainted with the promises and challenges of qualitative research. In a wide range of social science disciplines empirical analyses based on qualitative data and methods are an important source of information. In the four sessions, students will learn the basic tenets of a number of key aspects of qualitative research. Since qualitative research is characterized by a lower a number of cases, case selection is a delicate important first step – this will be discussed in the first session. The second session will discuss the importance of conceptualization and the popular method of process-tracing. In the third session, we will discuss the function of interview-based research and debate the utility of the QCA method. In the fourth session, qualitative and quantitative content analysis methods will be explored and discussed.
Literature to be announced.
WEEKS 3-6, Block 2
Module 6: Categorical data analysis
There are many instances in which the outcome that you want to explain or predict is a categorical variable. For example, firms might adopt a certain accounting practice or not (binary variable), we might be interested in how many patents a firm obtains (count variable) or in which country a firm decides to locate/invest (multinomial variable). For each of these cases, an OLS regression is unsuitable and we have to use categorical data analysis. In this module you will learn to work with the most common types of categorical data analyses. We will deal with the following regression models: Logistic regression, Ordered logistic regression, Tobit regression, and Poisson regression. For each of these models specific attention will be paid to how the results can be interpreted in meaningful ways through effect size analyses.
Literature to be announced.
Module 7: Panel data and multilevel analysis
Week 1: Multilevel: Research is often conducted with data acquired from different levels. For instance; a comparison between firms across Europe. Firms reside within countries, but also firms have their own sector. Combining all firms into an analysis is prone to errors of interpretation, such as drawing conclusions based on spurious effects, ecological fallacy and atomistic fallacy. The use of multilevel models is often required. The technique is designed to study effects of independent variables at different levels of analysis (e.g. individuals within organizations, organizations within countries, etc.) on a variety of economic outcomes.
Week 2: Panel Data: Panel data contains information collected for the same individuals (countries, organizations etc.) at several points in time. For analysing such data, special panel models have been developed. Among the models discussed in this lecture are seemingly unrelated regression, fixed effects models and random effects models.
What if you have data that is collected at multiple moments in time and uses different levels? Do we need to worry? Does a multilevel panel dataset exist? Is it better than using a blunt instrument? All will be revealed in the season finale.
Literature to be announced.
WEEK 7, Block 2
Module 1, Part 2: Summary of the Course
This second part of Module 1 gives an overview over the key concepts learned in the course, and provides some summarizing examples.
Literature to be announced.
|