Comments on: BYU-Utah Winter Strategy Conference

By: Andy

Andy — Wed, 05 Mar 2008 15:22:26 +0000

It is very difficult to study Markov chain topic. Not many good reference textbooks to study Markov chain.

I use Markov Chains and Stochastic Stability to study. This is good reference textbook.

Do you have any other good Markov Chains related textbooks recommend?

Regards,

Andy

By: teppof

teppof — Tue, 04 Mar 2008 01:17:55 +0000

Ty: Jay did give you lots of props as the bayesian brain in the project.

By: Ty Mackey

Ty Mackey — Mon, 03 Mar 2008 22:08:11 +0000

I’ll throw in my applause along with Bo for Bayesian hierarchical models, since Jay was discussing our diversification paper at the conference.

I wouldn’t say the field hasn’t adopted them because we’re methodologically sloppy or lazy, but instead because we take our methodological cues from economics. They don’t have the inherent interest in heterogeneity that strategy scholars do, and so their methodology doesn’t need to incorporate firm heterogeneity.

And while Bayesian methods aren’t as convenient as prepackaged programs like Stata or SAS, they really have become a lot easier to implement in recent years. Greg Allenby’s book, “Bayesian Statistics and Marketing” is a great place to start along with his bayesm package in R. Hansen, Perry, and Reese’s (2004) and Hahn and Doh’s (2006) recent papers (both in SMJ) have given the field of strategy a great introduction to the potential value of Bayesian hierarchical models for our field.

By: the strategy in strategy « orgtheory.net

the strategy in strategy « orgtheory.net — Mon, 03 Mar 2008 16:27:22 +0000

[...] that followed a presentation by Nicholai Foss (who posted his own thoughts on the conference here), an organizational economist from Copenhagen and blogger from our evil twin-site. Nicholai’s [...]

By: Bo

Bo — Mon, 03 Mar 2008 02:10:26 +0000

I too applaud the increasing attention to multi-level methods in management research. Indeed, these methods are common in fields such as health and education etc., in which data are often collected in a nested or hierarchical fashion. Strategic management issues are no different, however, for simplicity people often try to model data as (conditionally) at a fairly high level of aggregation; for instance by pretending that all the subjects in a sampling are drawn homogeneously from a single population. In fact, heterogeneity is often the rule
rather than the exception, and frequently the available predictor variables do not explain this heterogeneity sufficiently. With recent computational advances in Markov chain Monte Carlo (MCMC) methods it’s becoming increasingly straightforward to at least describe such heterogeneity with mixture models that employ latent variables (unobserved predictors) in a hierarchical structure.

Some have argued that these methods are better suited to certain types of studies in certain disciplines (i.e. studying students within classes within schools), however, new softwares (such as Mplus) have the capability to test hierarchical models where the dependent variable is at a higher level, enabling us to investigate micro foundational (e.g. individual level) influences on macro level phenomena.

Now, the question is: are we as management scholars poorly educated methodologically or simply lazy people looking for the convenient easy solution? one could make the argument that both are true and perhaps correlated; Since social science rarely is a matter of life or death (although some people like Nicolai may disagree) or even has measureable (tangible) outcomes, such as increasing test scores on SATs, there seems to be a natual tendency to do “methodologically sloppy” work. At the end of the day, what drives our research (in many instances) is the ability to publish (i.e. convince our peers that our model is an adequate representation of reality) rather than the ability to influence/impact society (at least directly). Hence, why learn more advanced (costly) methods that may yield better (as in more accurate) results when rational economic behavior tells us to focus on doing just enough to get published regardless of accuracy?

By: Warren Miller

Warren Miller — Sun, 02 Mar 2008 15:02:45 +0000

Thank you, Nicolai. Sounds like a great conference. I’m w/Joel Baum on “extreme events.” In a non-strategy context, an excellent book on the subject is “The Black Swan.” Highly recommended.

By: Hakan Ener

Hakan Ener — Sun, 02 Mar 2008 03:16:28 +0000

Great advice from the panel on research methods. However, there must be a good reason why methods such as hierarchical Bayesian analysis or Extreme Statistics are not used: it is simply easier to convince readers using a technique they are familiar with. Even “new methods” such as variance decomposition (used in explaining the drivers of firm performance) in the strategy literature since 20 years ago have not diffused because readers found it difficult to learn and be convinced through a new type of analysis. Until these methods are part of the standard PhD program, there will be a very limited niche for research that uses them.