Many strands of research make use of survey data. Regardless of whether the surveyed population represents companies, policy-makers or members of administration, or the population at large, both academic and non-academic research as well as policy evaluation extensively use survey instruments. However, classical surveys do not necessarily establish strong causal links between characteristics of interest and observed effects. While a range of options are available to identify causal patterns ex post by means of econometric modelling that exploit semi-experimental features of certain surveys or events, an alternative available to the researcher is to build experiments directly into the survey design. Against this background, this course provides a general introduction to the logic, design and implementation of experimental survey designs, for a range of applications from academic research, to social and economic analysis, to analyses supporting decision-making and managerial practices, to policy making and evaluation. The final modules of the course will guide the students into applied analysis of survey results in econometric software. Finally, the course provides the PhD candidates with the opportunity to develop their own survey experiments tailored to the specific needs of their PhD dissertations.

.

Class Structure: Every session lasts 2 hours. The class is structured following a hybrid lecture-seminar format. The class begins with a lecture on the topic of the day. The second half of the class is devoted to preparatory work for the final paper, to discussions on how thematic and content knowledge discussed in the course can be applied to PhD theses, and to general discussions over the theme of the session. Every class is accompanied by a specific literature list made available in advance, as well as dedicated stata do files and instructions for practical use when needed. Evaluation: Students pursue a group or individual `mini-project?, selecting their own research question. This mini project requires students to identify a gap in the literature and a survey-testable hypothesis, and to develop a survey experimental design to test such hypothesis. PhD candidates whose theses have potentially a direct application of survey experimental research are encouraged to take contact with the lecturer, so as to tailor their examination requirements to the specific needs of their PhD projects. Students might opt-out from the co-authored paper and opt instead for single-authored pieces. The projects will be presented in the last sessions of the seminar and will have to pass group peer-review. Course structure: Session # 1: introduction to the course; understanding causality in survey research, foundations of survey research (population, sample, quotas, practical matters). Single- vs multiple- factorial designs Session # 2: vignette experiments, cueing experiments, and other single factorial designs Session # 3: multi-factorial designs: conjoint experiments Session # 4: conjoint experiments in practice: stata seminar Session # 5: research design presentations

In presenza

Presentazione orale - Presentazione report scritto

P.D.2-2 - Marzo