Publications

We present new evidence on the relationship between employee productivity and job tenure using data from the teacher labor market. Econometric challenges require identifying assumptions to model the within-teacher returns to experience with teacher fixed effects. We describe the bias introduced by violations of two common assumptions, and we propose a third approach with a different and empirically-testable assumption. Consistent with past research, we find that teachers experience rapid productivity growth early in their careers. However, we find suggestive evidence of returns to experience later in the career, indicating that teachers continue to build human capital beyond these first years.

Courts have articulated a number of legal tests to distinguish corporate transactions that have a legitimate business or economic purpose from those carried out largely, if not solely, for favorable tax treatment. We outline an approach to analyzing the economic substance of corporate transactions based on the property rights theory of the firm, and describe its application in two recent tax cases.

How accurate are people when predicting whether they will vote? These self-predictions are used by political scientists, public opinion researchers, and psychologists. Responses from phone surveys over three elections are compared to administrative voting records to determine pre-election vote intention and actual voting behavior (N=29,403). Unsurprisingly, many who predict that they will vote actually do not vote. However, many who predict that they will notvote actually do (29% to 56%). Records of past voting behavior predict turnout substantially better than self-prediction. Self-prediction inaccuracy is not caused by lack of cognitive salience of past voting or by inability to recall past voting. Moreover, self-reported recall of turnout inone past election predicts future turnout slightly better than self-prediction of future turnout. Wediscuss implications for political science research, behavioral prediction, election administrationpolicy, and public opinion.

One hundred years after J. M. Barrie published the novel Peter and Wendy, Maria Tatar revisits a story that, like Alice in Wonderland, bridges the generations, animating both adults and children with its kinetic energy. The adventures of the Darling children with Peter Pan and Tinkerbell in Neverland are the seminal tale of escape and fantasy. Inspired by Barrie's real-life adventures with the five Llewelyn Davies boys he adopted, the story of Peter Pan has a deep and controversial history of its own that comes alive in Tatar's new edition. This brilliantly designed volume—with period photographs, full-color images by iconic illustrators, commentary on stage and screen versions, and an array of supplementary material, including Barrie's screenplay for a silent film—will draw readers into worlds of incandescent beauty, flooding them with the radiance of childhood wonder and the poignancy of what we lose when we grow up.

This paper examines the distribution of thematic infinitive endings in early Greek epic in the context of the long-standing debate about the transmission and development of Homeric epic diction. Ιn Homer we find both Ionic -εῖν and Aeolic -έμεν, the latter mostly occupying the biceps of the fourth or the fifth foot, conforming to the well-known preference for a dactylic word-end before the bucolic dieresis and before the sixth foot. Forms in -έμεν have been viewed either as (1) remnants from the “Αeolic stage” of epic diction, not “ionicized” by Ionian bards, because contracted Ionic -ẹ̄n (< *-ehen) would otherwise fill the biceps, resulting in an undesirable spondaic foot, or as (2) products of secondary “aeolicization”, whereby Aeolic -έμεν from the neighboring tradition was substituted for the metrically inept Ionic -ẹ̄n. This paper provides a novel argument in favor of the second solution (the “diffusionist” approach), starting from the fact that there are no aorist infinitives in -έμεν in Homer which would scan as υ υ - before a consonant or caesura (e.g. *βαλέμεν): instead we find unexplained forms in -έειν (e.g. βαλέειν). It is argued that this artificial “distraction” should be viewed as an actual analogical innovation, resulting from a proportional analogy to the “liquid futures” in *-ehe/o-: inf. fut. βαλεῖν : βαλέειν = inf. aor. βαλεῖν : Χ, where X is resolved as βαλέειν As the paper proceeds to argue, the total absence of aoristic -έειν from Hesiod is unlikely to be coincidental: this artificial form must have been a product of specifically East Ionic Kunstsprache (aor. inf. in -έειν are absent from epichoric Ionic) and was unknown in the different Ionian school of epic poetry where Hesiod may have been trained. Returning to the debate about the stages of Homeric epics, this paper argues that the striking avoidance of anapaestic aorist infinitives in -έμεν cannot be adequately explained under the “Aeolic phase” theory: it remains unclear why in the process of Ionicisation an Ionian singer would replace an archaic/foreign form in -έμεν by a form in -έειν which likewise did not belong to his vernacular (e.g. *βαλέμεν δέ → βαλέειν δέ). Artificial forms generated by the preference for dactylic word-end in certain metrical contexts abound, but there are no cases in Homer where an Aeolic form can be suspected to have been replaced by a metrically equivalent artificial form. Under the “diffusionist” approach, however, it can be plausibly supposed that in the cases where contraction of *-ehen to -ẹ̄n distorted the meter, the Ionian singers were nevertheless able to continue employing traditional formulae by replacing the former *-ehen before a consonant (e.g. *βαλέhεν δέ) by a form in -έειν (e.g. βαλέειν δέ), created and embedded in their own poetic tradition. But -έειν could not be used if the resulting form had a cretic shape (*ἐλθέειν - υ - ), and therefore the Ionian singers had to resort to borrowing Aeolic aor. inf. ἐλθέμεν. The “aeolicization” theory is thus able to fully explain the distribution of Aeolic έμεν and Ionic -έειν in Homeric epics.

At every moment, we face choices: Is it time to work, or to play? Should I listen to this lecture, or check my e-mail? Should I pay attention to what my significant other is saying, or do a mental inventory of what work I need to accomplish today? Should I keep my hands on the wheel, or change the radio station? Without any change to the external environment, it is possible to select a subset of these possibilities for further action. The process of selection is called attention, and it operates in many domains, from selecting our higher-level goals, to selecting the sensory information on which we focus, to selecting what actions we perform. This chapter focuses on the relationship between visual attention (selecting visual inputs), and action (selecting and executing movements of the body). As a case study, we focus on visual-spatial attention, the act of choosing to attend to a particular location in the visual field, and its relationship to eye-movement control. Visual attention appears to select the targets for eye-movements, as attention to a location necessarily precedes an eye-movement to that location. Moreover, there is a great deal of overlap in the neural mechanisms that control spatial attention and eye-movements, and the neural mechanisms that are specialized for spatial attention or eye-movements are highly intertwined. This link between spatial attention and eye-movements strongly supports the idea that a computational goal of the visual attention system is to select targets for action, and suggests that many of the design properties of the spatial attention system might be optimized for the control of eye-movements. Whether this relationship will hold broadly between other forms of attention (e.g., goal selection, auditory selection, tactile selection), and other forms of action (e.g., hand movements and locomotion) is an important topic of contemporary and future research.