Multiplicative interaction models are widely used in social science to examine whether the relationship between an outcome and an independent variable changes with a moderating variable. Current empirical practice tends to overlook two important problems. First, these models assume a linear interaction effect that changes at a constant rate with the moderator. Second, estimates of the conditional effects of the independent variable can be misleading if there is a lack of common support of the moderator. Replicating 46 interaction effects from 22 recent publications in five top political science journals, we find that these core assumptions often fail in practice, suggesting that a large portion of findings across all political science subfields based on interaction models are modeling artifacts or are at best highly model dependent. We propose a checklist of simple diagnostics to assess the validity of these assumptions and offer flexible estimation strategies that allow for nonlinear interaction effects and safeguard against excessive extrapolation. These statistical routines are available in both R and STATA.

We study the short and long-term spillover effects of a pay reform that substantially increased the returns to schooling in Israeli kibbutzim. This pay reform, which induced kibbutz students to improve their academic achievements during high school, spilled over to non-kibbutz members who attended schools with these kibbutz students. In the short run, peers of kibbutz students improved their high school outcomes and shifted to courses with higher financial returns. In the medium and long run, peers completed more years of postsecondary schooling and increased their earnings. We discuss three main spillover channels: diversion of teachers’ instruction time towards peers, peer effects from improved schooling performance of kibbutz students, and the transmission of information about the returns to schooling. While each of these channels likely contributed to improving the outcomes of peers, we provide suggestive evidence that the estimates are more consistent with the effects operating mainly through transmission of information.

Striving better to uncover causal effects, political science is amid a revolution in micro-empirical research designs and experimental methods. This methodological development—although quite promising in delivering new findings and discovering the mechanisms that underlie previously known associations—raises new and unnerving ethical issues that have yet to be confronted by our profession. We believe that addressing these issues proactively by generating strong, internal norms of disciplinary regulation is preferable to reactive measures, which often come in the wake of public exposés and can lead to externally imposed regulations or centrally imposed internal policing.

This article probes the extent to which post secondary Spanish learners can substantively increase their knowledge of Spanish over a two-week period within a context of language and content instruction for four hours per day. The article considers the relationship of an immersion experience to upper-level literature and culture classes. Insights into integrating service learni ng as a key part of the Spanish learning experience are also provided. Oral and writing data were collected in 2012, 2013, 2014, and 2015 to demonst rate that participants showed significant im provement in their Spanish language abilities within the two weeks of the intensive experience. The enhancement of participants’ Spanish language profi ciency did not, however, bridge to upper-level Spanish literature and culture courses, or necessarily to an interest in study abroad. Using service learning as a motivating factor in student participation had a positive affective impact.

Multiplicative interaction models are widely used in social science to test whether the relationship between an outcome and an independent variable changes with a moderating variable. Current empirical practice overlooks two important problems. First, these models assume a linear interaction effect that changes at a constant rate with the moderator. Second, reliably estimating the conditional effects of the independent variable at all values of the moderator requires sufficient common support. Replicating nearly 50 interaction effects from 22 recent publications in five top political science journals, we find that these core assumptions fail in a majority of cases, suggesting that a large portion of findings across all subfields based on interaction models are modeling artifacts or are at best highly model dependent. We propose simple diagnostics to assess the validity of these assumptions and offer flexible estimation strategies that allow for nonlinear interaction effects and safeguard against excessive extrapolation.

“What was the Enlightenment” is a question scholars have been asking since the eighteenth century, but far less often has it been asked who the Enlightenment was. Who were the members of the social, professional, and academic classes that made up the Enlightenment? Who was in and, by extension, who was out? Much is known about the primary producers of Enlightenment works, from the famed philosophes to some of the lesser-known authors (e.g., in France, the Encyclopédistes, the Grub Street writers, or the libellistes).¹ Studies have also revealed some of the readers of Enlightenment texts and explored some of its social institutions. And a certain degree of attention has been paid to the major patrons and political supporters of the philosophes: Mme de Pompadour, Frederick the Great, and other influential governmental officials such as Malesherbes. But rarely, if ever, have the different participants of the Enlightenment been considered together as a single entity, or, as we would say today, a network.

That is the goal of this article: to study the social composition of the French Enlightenment network. What sectors of eighteenth-century society were most present? How involved was the government? What role did the aristocracy play? Which intellectual disciplines and fields were most represented? Which ones were not? What was the role of women? And, perhaps most critically, how did it function as a network? These are the kinds of questions we raise, and we seek to answer them by means of a hybrid quantitative and qualitative methodology. We use basic statistical calculations to provide rough estimates of the size and importance of different social groups; given the nature of our data, however, we do not analyze them with standard social network analysis (SNA) methods. Rather, we corroborate, refine, and defend our findings through comparisons with arguments from the secondary literature. While our reasons for this approach are largely driven by the shape of our data (which make most SNA metrics unfeasible), we also advocate a method of network analysis that does not rely on mapping relationships between nodes and calculating such metrics as betweenness centrality or clustering coefficients; rather, it focuses on the relative size of, and overlap between, different subgroups in order to understand the overall structure and social composition of a historical network.

While some of our findings, detailed in full below, will not come as a surprise to specialists (the network is, for instance, overwhelmingly male), the overall picture of the French Enlightenment network that emerges from this study is nonetheless striking. Two features in particular stand out. First, men and women of science are significantly underrepresented. The scholars and writers we find in this network were largely gens de lettres, much more engaged with history, philosophy, political economy, and literature than with mathematics, medicine, or astronomy. Second, the elite segments of society, be they aristocratic, social, or governmental, are remarkably overrepresented. The French Enlightenment network incorporates the crème de la crème of the French state and high society. The presence of so many notables in this network suggests that French Enlightenment authors were more likely to be engaged in collaborative, reformist efforts than in subversive plots; conversely, it also hints at the existence of a fairly extensive and well-represented (at different levels of social hierarchy) parti philosophique within the French state itself.

Reading, writing, and discussion are the most common—and, most would agree, the most valuable—components of a university-level humanities seminar. In humanities courses, all three activities can be conducted with a variety of digital and analog tools. Digital texts can create novel opportunities for teaching and learning, particularly when students’ reading activity is made visible to other members of the course. In this paper, we introduce Lacuna, a web-based software platform which hosts digital course materials to be read and annotated socially. At Stanford, Lacuna has been collaboratively and iteratively designed to support the practices of critical reading and dialogue in humanities courses. After introducing the features of the platform in terms of these practices, we present a case study of an undergraduate comparative literature seminar, which, to date, represents the most intentional and highly integrated use of Lacuna. Drawing on ethnographic methods, we describe how the course instructors relied on the platform’s affordances to integrate students’ online activity into course planning and seminar discussions and activities. We also explore students’ experience of social annotation and social reading.

In our case study, we find that student annotations and writing on Lacuna give instructors more insight into students’ perspectives on texts and course materials. The visibility of shared annotations encourages students to take on a more active role as peer instructors and peer learners. Our paper closes with a discussion of the new responsibilities, workflows, and demands on self-reflection introduced by these altered relationships between course participants. We consider the benefits and challenges encountered in using Lacuna, which are likely to be shared by individuals using other learning technologies with similar goals and features. We also consider future directions for the enhancement of teaching and learning through the use of social reading and digital annotation.