CCK08 – Cointegration, Freakonomics and SNA

This is a post that was just waiting to be written, given the dormant econometrician in me! I am referring here to:

1. Cointegration modeling– an approach that models the behaviour of two or more non-stationary (i.e. with values that do not return to a fixed value or fluctuate around a trend) variables across time, cointegration grabbed my interest hugely in my grad courses in Econometrics at the Delhi School of Economics. For a long time, even non-stationary variables were modeled using standard correlation/regression techniques, producing what Clive Granger called spurious regression. Granger won the Nobel Prize in 2003 for defining a technique called Cointegration that allowed modeling of non-stationary variables to determine their relationships.
2. Freakonomics -A book by the same name by Steven Levitt (an economist) and Stephen Dubner (an award winning author and journalist) took the world by storm when it was released in 2005. They took a “mountain of data” and a “simple, unasked question” such as “Why drug dealers still lve with their moms?” or “What do schoolteachers and sumo wrestlers have in common?” (the latter is an important one because it looks at how policy can affect teacher motivation and the assessment process)
3. Social Network Analysis (SNA) – The analysis of social networks along parameters such as Network Centrality / Centralization, Small-World Networks, Cluster Analysis, Network Density, Prestige / Influence, Structural Equivalence, Network Neighborhood, External / Internal Ratio, Weighted Average Path Length, Shortest Paths & Path Distribution. A brilliant product built by Valdis Krebs, inFlow, helps analyze social networks in the context of a variety of phenomena – team building, leadership development, discovering key opinion leaders, discovering communities of practice etc. 

All these three approaches are grounded in statistical analysis (I think inFlow would probably benefit from doing dynamic time series analysis, if it already does not do so) and attack the problem of relating variables to other variables. Their application in a connective world increases exponentially and maybe can help answer a few interesting questions around the critical success factors affecting connective learning.