Readings

Antonakis, J., Bendahan, S., Jacquart, P., & Lalive, R. (2010). On making causal claims: A review and recommendations. The Leadership Quarterly21(6), 1086-1120. doi:10.1016/j.leaqua.2010.10.010

Video:   (32:19)

Wooldridge, J. M. (2013). Introductory econometrics: a modern approach (5th ed.). Mason, OH: South Western, Cengage Learning. (Chapters 6-8, 9.5, 15)

Hekman, D. R., Aquino, K., Owens, B. P., Mitchell, T. R., Schilpzand, P., & Leavitt, K. (2010). An Examination of Whether and How Racial and Gender Biases Influence Customer Satisfaction. Academy of Management Journal, 53(2), 238-264. doi:10.5465/AMJ.2010.49388763 (AMJ best paper winner for 2010)

Deephouse, D. L. (1999). To be different, or to be the same? It's a question (and theory) of strategic balance. Strategic Management Journal, 20(2), 147-166. doi:10.1002/(SICI)1097-0266(199902)20:2<147::AID-SMJ11>3.0.CO;2-Q

Mochon, D., Johnson, K., Schwartz, J., & Ariely, D. (2017). What Are Likes Worth? A Facebook Page Field Experiment. Journal of Marketing Research, 54(2), 306–317.

Instructions

Read the listed material. You should understand the main principles discussed by Wooldridge, but you do not necessarily need to fully understand all the details or the equations. If you must choose, reading all four chapters quickly is preferable to reading just a part of the material in depth. The paper by Antonakis is likely to be challenging, but you do not need to understand all the techniques. Focus on the first part of the paper that discusses the conditions for causality and threats to causal claims.

You can skim through the non-methodological parts of the three empirical papers. The versions available in the course Zotero library contain highlighted sections that are discussed in class. Note that the empirical papers are used in multiple classes so not all highlighted sections are discussed in the first class where a paper is used.

Questions

1) Search definitions or explanations for following concepts in Wooldridge's book. When applicable, quote the definition from Wooldridge. If there is no clear quotable definition, cite the page numbers on which Wooldridge discusses these concepts. Explain the concepts and how they are used in the empirical papers (Hekman, Deephouse, Mochon). (Hint: Most, but not all concepts are used in the three empirical papers. Some concepts that are used are not obvious and it is OK to say that you did not find the concept used in the papers.)

  1. Beta coefficients or standardized coefficients
  2. Interaction term and quadratic term
  3. Dummy variable
  4. Heteroskedasticity robust inference
  5. Endogenous explanatory variable, endogenous sample selection, and endogeneity
  6. Instrumental variable
  7. Marginal effect

2) Mochon et al conducted a study involving three smaller studies with some experimental elements in each. First, they did a screening survey by choosing a sample of Vitality customers based on their joining date and chose those customers that had a Facebook account but had not liked the Vitality page. This formed the main study sample, which was assigned to treatment and control groups. The treatment group were invited to like the Facebook page using four different messages that were assigned to subjects randomly. Those subjects that agreed to like the page were directed to Vitality Facebook page to click on the like button. After this manipulation, the activity of the subjects were followed first for a four month "organic" period and then for two month "boosted" period, where Vitality targeted promotion to Facebook followers. (See Figure 1 in the article). They then present three sets of analysis results: 1) the effect of the four different invitation emails comparing on the decision to accept the invitation to like the Facebook page (Table 1, Table 2), 2) the effect on the decision to like on points acquisition (Table 3) 3) the effect of boosted and organic period on point acquisition (Table 3). Which of these three effects are based on quasi-experimental and which on true experimental designs? What justifies a causal interpretation of each effect and under which assumptions?

3) Hekman et al's second study ("Bookcorp Stydy") is an randomized experiment. In randomized experiments, the subjects are assigned to the treatment and control conditions randomly, and this makes the two groups statistically equivalent. Because the group are initially statistically indistinguishable, there is no need to statistically control for initial group differences in this kind of study. However, the article also reports that control variables were used in the analysis. Is the use of control variables completely unnecessary in this case or is it necessary. Justify your opinion. You do not need to evaluate whether these control variables are good or not, just answer the general question of whether controls are needed in this scenario and why. 

Submit your answers two days before the lecture and bring them also to the lecture. The answer should be 3-6 pages. If you have done this assignment as a part of a previous attempt at this course, you can use your old answer as a starting point and do a revision. If you need help or have questions about the assignment, please post those to the course forum.

Sorry, no guest users are allowed to access this plugin. Please login.

window