UNICEF Webinar: Quasi-experimental design and methods

What is the main difference between quasi-experiments and RCTs? How can I measure impact when establishing a control group is not an option?

Dr. Howard White of the International Initiative for Impact Evaluation (3ie), covers the basics of quasi-experiments.

Quasi-experimental research designs, like experimental designs, test causal hypotheses. The main difference is that the former lacks random assignment to a control group and assignment to conditions is by means of self-selection, administrative selection or both. There are different methods for creating a valid comparison group, such as regression discontinuity design (RDD) and propensity score matching (PSM). The webinar will cover the most common methods and answer questions on the topic.

Q&A

Do you have any good examples of using indirect methods to assess the impact of child grants on health/ education outcomes? For example in cases where there is no baseline - and using e.g. IV or RDD against threshold?

Are there any sample size requirements for any of the methods presented that would prevent biases?

The findings, interpretations and opinions expressed in the webinars are those of the presenters and do not necessarily reflect the policies or views of the United Nations Children’s Fund (UNICEF). The presenters are independent impact evaluation experts who were commissioned by UNICEF to prepare the webinars and use their own knowledge and judgement on key issues and to provide advice. The questions and comments reflected in the Q & A materials are based on those submitted by UNICEF staff as part of this capacity-building initiative. They do not necessarily reflect the policies or views of UNICEF.

The webinars were commissioned by UNICEF and UNICEF is entitled to all intellectual property and other proprietary rights which bear a direct relation to the contract under which this work was produced. The materials on this page are subject to a Creative Commons license CC BY-NC (Attribution-NonCommercial) and may be used and reproduced in line with the conditions of this licence.