ASOK15648U The Logic of Design-Based Research: Measuring Causal Effects in Social Sciences

Volume 2021/2022
Education

MA Research Methodology and Practice (MSC Curriculum 2015)

Course package (MSc 2015):

Welfare, inequality and mobility

Knowledge, organization and politics

Culture, lifestyle and everyday life

Credit students must be at master level.
Exchange students at both bachelor and master level can sign up for this course.

Content

In this course, you will become familiar with the methodological criteria for measuring causal effects. You will learn when comparing outcomes (e.g., school grades) between two groups (e.g., children of divorced and intact families) can be said to reflect a causal relationship (e.g., divorce causes poorer school grades). This course is primarily a design course, why focus will be on the line of thought in constructing sound and credible quantitative research designs. In other words, you will learn to think along the lines of what makes a reliable design when seeking to answer causal research questions. Even if you are not planning on conducting quantitative research yourself, this course offers you not only a key understanding of quantitative designs used in social science and policy evaluations around the globe but also the skills needed for assessing the scientific and societal value of existing empirical research. Such skills have a high payoff in terms of understanding many aspects of the jobs that sociologists typically enter upon graduation.

 

Maybe you have noticed how many research questions ask about causal relationships. Do financial crises lead to more divorce cases? Is child poverty damaging for children’s opportunities later in life? And what about the association between unemployment and crime—which is cause, which is effect? Also, married men have better health than unmarried men, but is this because of other differences between those men or is marriage the causal engine driving these differences (or is it a mix of the two)?

 

Answers to these types of questions are integral to rigorous sociology and evidence-based policy. Important decisions are based on them. Policies are made, laws are written, and public opinion is shaped in response to the answers to such questions. But even though most people have an opinion on these types of questions, remarkably few are able to answer them in scientifically sound ways. And even fewer know about the methodological conditions that need to be in place for answering such questions in a sound and reliable way.

 

In this course, you will learn about the methodological criteria when seeking to answer causal questions and the logic of design-based research. The course consists of three parts (which each come with a minor portfolio student assignment—when compiled, these minor assignments will make up the main bulk of the exam assignment). In the first part of the course, we introduce you to the assumptions required for identifying causal effects. Here, we go through the counterfactual model of causality and the golden standard for estimating causal effects known as randomized controlled trials (RCTs). We also introduce you to alternative strategies when randomization is not in control of the researcher (quasi-experiments). In the second part, we read empirical research articles within family sociology that use both classical and more advanced statistical approaches (such as fixed effects and propensity score matching), and we discuss the pros and cons of the different applied research designs for the studies under consideration. Thus, in this second part of the course, we focus on understanding the issue of selection and the difference between correlation and causality based on empirical sociological research. In the third part, we discuss further, alternative research designs such as difference-in-differences, regression discontinuity designs and instrumental variable estimation using empirical examples. We discuss how we can use these designs to get closer to estimating causal effects and under which assumptions. Considered together, these three parts offer you key insights into the mechanics of causal effects and reasoning, how they are applied in empirical sociological research, and how a range of different research designs enables one to recover causal effect in different (and often novel) ways.

Learning Outcome

Knowledge 

  • You can explain the principles of the counterfactual model of causality and the general challenge of estimating causal effects in the social sciences.
  • You know about potential pitfalls in causal estimation and can demonstrate understanding of the difference between correlation and causality.
  • You can explain the line of thinking in different types of design-based research and you know which questions can be answered with each type of design.

 

Skills

  • You can formulate and motivate sociological causal research questions.
  • You can discuss why and how we can deal with selection issues with a simple comparison between groups.
  • You can find or set up (by thought) a realistic natural experiment to answer your research question and explain why you can use this design to recover causal effects.
  • You can critically discuss the assumptions that need to be met in a specific research design, including issues related to external validity (i.e., generalizing to the wider population from select samples).

 

Competences

  • Upon course completion, you will have a broad set of tools for constructing design-based quantitative research.
  • The course serves as a steppingstone enabling you to participate in more advanced courses in econometrics and/or policy evaluation (in or outside sociology).
  • The course enables you to critically assess the scientific value of causal claims and statements flourishing in society and in much policy-related work. In other words, you will be able to evaluate existing empirical research which is an important competence in several work settings – even if you are not planning on performing quantitative analysis yourself.

As textbook we will use Dunning, T. (2012). Natural Experiments in the Social Sciences. Cambridge University Press.

 

We will also read several empirical research articles.

 

All course material is in English.

Basic skills in quantitative methods (e.g., regression models and statistical control) are advantageous but not required.

Note, that this course is not a hands-on statistical course, and we will not be working in STATA or use other statistical software. Instead, focus is on analytically understanding of different research designs.
Lectures, group discussions, and student presentations.
Note that this course is similar in its methodological content to the course entitled “Design-Based Research in Criminology,” which was previously offered by the department.
  • Category
  • Hours
  • Class Instruction
  • 28
  • Preparation
  • 148
  • Exam Preparation
  • 30
  • Total
  • 206
Continuous feedback during the course of the semester
Peer feedback (Students give each other feedback)
Credit
7,5 ECTS
Type of assessment
Portfolio, -
Portfolio
Individual or group. A portfolio assignment is defined as a series of short assignments during the course that address one or more set questions and feedback is offered during the course. All of the assignments are submitted together for assessment at the end of the course. The portfolio assignments must be no longer than 10 pages. For group assignments, an extra 5 pages is added per additional student. Further details for this exam form can be found in the Curriculum and in the General Guide to Examinations at KUnet.
Exam registration requirements

You must be registered for the course to take the exam.

Aid
All aids allowed
Marking scale
7-point grading scale
Censorship form
No external censorship
Exam period

Find more information on your study page at KUnet.

Exchange students and Danish full degree guest students please see the homepage of Sociology;
www.sociology.ku.dk under Education --> Exams

Re-exam

Written take-home essay with NEW formulated questions

Individual/group.

 

Criteria for exam assesment

Please see the learning outcome