New paper suggests the field of strategy is vulnerable to a replication crisis. Publication standards need to change to require data disclosure and facilitate replication.
- Bergh, D. D., Sharp, B. M., Aguinis, H., & Li, M. (2017). Is there a credibility crisis in strategic management research? Evidence on the reproducibility of study findings. Strategic Organization, 15(3), 423–436. http://doi.org/10.1177/1476127017701076
Authors attempted to replicate 88 studies published in Strategic Management Journal. About 70% of articles did not provide enough data to replicate findings.
Of the 30% of articles that provided enough data to be retested, about 33% included statistically significant hypotheses that did not replicate. Far more significant results were found insignificant than vice versa.
Dov Eden’s review of field experiments in organizations is out at Annual Review of Organizational Psychology and Organizational Behavior (link).
Field experimentation, although rare, is the sterling-gold standard of orga-
nizational research methods. It yields the best internally valid and general-
izable ﬁndings compared to more fallible methods. Reviewers in many psy-
chology specialties, including organizational psychology, synthesize largely
nonexperimental research, warn of causal ambiguity, and call for experi-
mental replication. These calls go mostly unheeded. Practical application
is a raison d’ˆetre for much organizational research. With the emergence of
evidence-based management, ﬁeld experiments enable us to deliver the most
actionable tools to practitioners. This review explicates the role of experi-
mental control and randomization and enumerates some of the factors that
mitigate ﬁeld experimentation. It describes, instantiates, and evaluates true
ﬁeld experiments, quasi-experiments, quasi-ﬁelds, combo designs, and tri-
angulation. It also provides practical tips for overcoming deterrents to ﬁeld
experimentation. The review ends describing the merging of new technolo-
gies with classical experimental design and prophesying the bright future of
organizational ﬁeld experimentation.
I created a short .do file comparing several ways of estimating a two-period difference-in-differences model in Stata.
Code can be found on my Github page.
The University of North Carolina Population Center has a nice overview of ways to export Stata results to Word and other text processors.
Example of linear regression using machine learning in R: https://datarvalue.blogspot.com/2017/05/machine-learning-linear-regression-full.html
Steph Locke has a great post on interpreting regression coefficients at the Locke Data blog.
Find Steph Locke on Twitter.