Provider: edX (UBCx)
Tutors: Michael Law
If you have already read some of the posts I wrote about MOOCs, it shouldn’t come as a surprise that I tend to prefer edX over Coursera. And indeed, this is another course that left me much more satisfied than most of those I took on Coursera.
Interrupted time series analysis is a quasi-experimental design that can be used to assess the impact of an intervention, especially when randomized controlled trials are not feasible. Here I quote from one of the readings for the course, as it is very clearly explained:
“A time series—repeated observations of a particular event collected over time—is divided into 2 segments in the simplest case. The first segment comprises rates of the event before the intervention or policy, and the second segment is the rates after the intervention. “Segmented regression” is used to measure statistically the changes in level and slope in the postintervention period compared to the preintervention period.” (the full article can be retrieved here).
It is probably easier to understand with a graph:
The point of the analysis is to asses if an intervention had an impact in terms of level (i.e.: is the segment after the intervention starting at a lower or higher point compared to where the segment before finished) and trend (has the slope of the segment significantly changed) of a phenomenon. In this case, we are trying to see if email advertisement (represented by the dotted black line) had an impact on the rate of enrolment to the MOOC (it had indeed significant impact on the level, and a less apparent impact on the trend, although this might be due to other circumstances).
The course is well thought, with in depth and progressive explanations, good readings and some practice exercise. The tutor is active in the forum and usually provides a timely feedback.
It is important to note that you should be already somehow familiar with R and with linear regression analysis, otherwise you will probably struggle to follow it through and complete within the expected time limit.
The only true limit I found was related to the number of practical exercises: although each lesson come with some comprehension questions and small ungraded exercises, I had the impression of getting to the graded exercise without the opportunity to practice enough. This is quite a limit in consideration of the fact that finding in internet practice exercises on the topic is much more difficult than, say, regression analysis or statistical inference.
This being said, the course is solid, well planned and well taught, and I would recommend it to anyone working in policy evaluation and monitoring.