Applied Multivariate Statistics
This course aims to introduce the students to concepts and techniques of summarising,
visualising the geometry, and analysing such dependent, multivariate data.
The Applied Multivariate Analysis course is a second-year elective course for MA Economics students. The course builds off foundational knowledge and skills covered in Quantitative and Qualitative Research Methods and the Econometrics of Impact Evaluation core courses taken within the first year of the MA programme and provides a deeper and more extensive engagement with statistics for students who are interested in doing extensive empirical work or in pursuing further studies in economics.
Data in most disciplines such as, economics, sociology, and political science, are
multidimensional in nature. For example, in field surveys to assess the main dimensions of
challenges faced in accessing a government programme, different data pertaining to
demographic details, socio-economic profile, preferences etc. would be collected for
respondents. In financial markets, stock market assets, observed simultaneously exhibit
joint movement, which then affects the overall index of the market. To understand
consumption behaviour of individuals, numerous variables ranging from income to
expenditure on a diverse basket of goods are collected to understand the variations of
consumption across caste, gender, class, and geography. In each of the above examples, one can think of each measured variable as an additional dimension, and therefore many
variables correspond to data sitting in a high dimensional space.
Such variables also tend to be highly dependent. The most basic phenomenon of dependence is that of correlation – the tendency of quantities to vary together: tall parents tend to have tall children. Understanding and extracting structures of dependence in higher dimensions requires sophisticated mathematical machinery.