Recent research studied how relative salary affects job satisfaction. It gave a random sample of University of California employees information about their coworkers’ salaries and estimated the effect of this information on job satisfaction. This article suggests ways the dataset created by this research can be used in econometrics and statistics classes. It provides examples using these data to calculate frequency distributions, contingency tables, Chi-square tests, and linear probability models. It also explains how these examples can be used productively in class.

Introduction

This article introduces a valuable dataset for teaching introductory-level econometrics and statistics classes. In addition, it provides examples of how the dataset can be used to teach effectively. The dataset is valuable because it allows an instructor to use real data, take advantage of students’ interest in income inequality, and connect his/her class to topics students study elsewhere. The dataset was developed as part of seminal research by Card et al. (2012). They asked how earnings inequality within units of an organization affects employees’ job satisfaction. I show how instructors can use the dataset to provide students opportunities to practice using frequency distributions, contingency tables, Chi-square tests, linear probability models, and logit models. I also show how an instructor might use the dataset to introduce students to experimental treatments to identify causal effects. The examples I give are taken from my teaching of a business statistics class for MBA students, one which emphasizes data literacy. However, most would be relevant in econometrics classes and other statistics classes. The learning outcomes emphasized in my exposition are students’ abilities to apply each statistical method and to interpret the output each method generates. For example, I show how one can use the dataset to illustrate the use of contingency tables to explore the association between a worker’s knowledge of peer salaries and their level of job satisfaction. I also suggest ways one can help students interpret particular relative frequencies from a contingency table. One reason this dataset is valuable is that it allows an instructor to connect their class to the real world and real research. Singer and Willett argue for using real data to teach statistics (1992).