This page contains links to data and/or code from research projects. Please acknowledge and/or cite appropriately and please send me an e-mail if you use proprietary data (data you could not or no longer assemble from public sources).
Work in progress™
Exploration and Exploitation
In 2018 I held the Ph.D. seminar Advanced Strategy at Harvard Business School. The regular paper discussion sessions were prefaced by a number of sessions on the philosophy of science and research methods. For one of these sessions I reprogrammed parts of the simulation in March’s (1991) seminal paper: Exploration and Exploitation in Organizational Learning, Organization Science 2(1): 71-87, based on the description he gave in the article. The commented code reproduces the simulations underlying figures 1, 2, and 4 and the figures themselves using the R statistical software, which is available free of charge. In the article, March is not sufficiently specific about all assumptions. More precisely specified and auxiliary assumption(s) are found underneath March’s original description in the code file. In no variation of the simulation have I been able to reproduce the crossover pattern or contingency effect in March’s figure 1. Specifically, in no variation have I found that with a high rate of members learning from the code (high p1), a slow rate of the code learning from the members (low p2) is superior. Instead, I found in all variations that a slow rate of members learning from the code and a fast rate of the code learning from its members is superior. Irrespective of whether the mistake is mine or March’s, it means that this contingency effect is likely not robust to minor changes in assumptions. All other results hold.
Grands crûs classés of the Médoc
Classification, wine ratings, and price data for the Grands crûs classés of the Médoc, Bordeaux, France, as analyzed in Malter (2014) On the Causality and Cause of Returns to Organizational Status: Evidence from the Grands Crus Classés of the Médoc. Administrative Science Quarterly 59(2):271-300.
Wine auctions
A hand-collected dataset of the right-censored online auction life cycle of a newly released vintage of an Australian prestige wine on German eBay. At the time the data was collected, bids and bidder handles were identifiable. The data contains information about roughly 250 distinct bidders placing roughly 2500 bids in roughly 250 auctions, most of which were single-unit auctions. To the best of my memory, the next auction of the same product did not occur until about six months after right-censoring.
2008 state-level U.S. presidential election polls and results
Data on state-level polls and results of the 2008 U.S. presidential elections (Obama vs. McCain), complented by state-level census data, data on religious homogeneity computed from the American religious data archive, and data on the 2004 Kerry/Bush margin, Bush presidential approval, and the Obama/Clinton margin in the 2008 primaries/caucuses collected for a class project and described in Malter (2010) “Analyzing Obama’s Out- and Under Performance in the 2008 Presidential Elections: Social Desirability Bias, Sample Selection, and Momentum Neglect in the Polls.”