Session 237

Replication and verification

Track R

Date: Tuesday, October 6, 2015

 

Time: 14:15 – 15:30

Paper

Room: Plaza Court 6


Session Chair:

  • William Wan, City University of Hong Kong

Title: Introducing Text Analytics to Strategic Management Research

Authors

  • William Wan, City University of Hong Kong
  • Daphne Yiu, Chinese University of Hong Kong

Abstract: Extracting unique data from large amounts of text has always attracted strategy researchers. Computer-aided content analysis thus has seen increasing adoption. However, the data obtained from such analysis are primarily ‘facts’; ‘opinions’ are left out. This is an immense waste of information. The purpose of our paper is to introduce text analytics to strategic management research. Text analytics is a computational research method for mining and extracting themes and concepts in texts, as well as capturing opinions, sentiments, subjectivities, appraisals, and emotions toward entities, events and their attributes. We discuss its salient features, and highlight textual sources and research topics that the methodology can be gainfully employed in strategic management research.

Title: Performance Persistence in the Long Run: An Analysis using Markov Chains

Authors

  • Ingo Reinhardt, Simon-Kucher & Partners

Abstract: Strategy researchers frequently rely on the analysis of year-to-year changes to study performance persistence. Examples include the use of first-order autoregressive resource accumulation models or Markov chains to model transitions between performance percentiles. However, it is unclear if long-term persistence (e.g. over 20 years) can be explained as the result of consecutive year-to-year changes. We apply a Markov chain analysis to test if different models of year-to-year changes in performance can explain long-term persistence. To test our hypotheses, we study transitions between ROA deciles in the Fortune 500 lists from 1955 to 1994. We find that commonly used models of year-to-year persistence are unable to explain persistence over a time span of 10 to 20 year. Particularly, such models substantially overestimate regression to the mean.

Title: Replication Logic 2.0: What Makes a Case Study Rigorous?

Authors

  • Michael Gibbert, University of Lugano
  • Lakshmi Balachandran Nair, University of Lugano
  • Winfried Ruigrok, University of St Gallen

Abstract: We analyze the scope for methodological sophistication in using “replication logic” as theory-building device in published case study research (N=184) in the period 1996-2006 in Strategic Management Journal, Academy of Management Journal, Administrative Science Quarterly, Organization Science and Management Science. We empirically differentiate three types of replication logic: literal replication, theoretical replication and no replication, and illustrate their theory-building potential by demonstrating the implications for internal and external validity, describe the boundary conditions of each type in terms of theoretical sampling, the discussion of units of analysis and control as well as focal variables.

Title: The Looming Replication Crisis in Strategic Management

Authors

  • Donald Bergh, University of Denver
  • Barton Sharp, Northern Illinois University
  • Herman Aguinis, George Washington University
  • Ming Li, Hull University Business School

Abstract: We use data reported in SMJ studies in an attempt to reproduce empirical findings and gain insights into the status and scope of the replicability of knowledge in strategic management research. Results from two assessments – one of 79 articles reporting regression analysis and one from the top ten cited articles reporting structural equation modeling – reveal that most studies’ findings could not be reproduced, let alone replicated. Of those that were reproducible, coefficient signs were mostly confirmed but some significance levels were not, suggesting the possible presence of confirmation bias. Overall, the limited data disclosure raises concerns about whether attempts to replicate findings can yield valid confirmatory evidence. Our confidence in the field’s body of knowledge depends more on contributor integrity than independent substantiation tests.

All Sessions in Track R...

Sun: 08:00 – 09:15
Session 259: Multilevel Modelling
Sun: 11:15 – 12:30
Session 33: Methodological challenges in publishing international strategy research
Session 250: Foundational Issues in Research Methods
Sun: 17:45 – 00:00
Session 326: Research Methods Community Business Meeting
Mon: 08:00 – 09:15
Session 236: Improving methods
Mon: 11:15 – 12:30
Session 254: Ethics in Research Methods
Tue: 14:15 – 15:30
Session 237: Replication and verification


Strategic Management Society

Denver