We emphasize that science and practice issues are equally salient when pursuing thinking and research on employee engagement. We agree with much of what the commentaries have to say, especially that organizational competitive advantage is the relevant focus of engagement research and practice and that engagement is not a new construct but one that required clarification vis-a-vis existing constructs. We also agree that state engagement can be highly variable, that disengagement needs study, that negative situations can induce engagement behaviors, that engagement surveys should yield actionable data, and that people can be hired who are more likely to be engaged. We disagree with the idea that all employee attitudes are essentially equal and that existing conceptualizations of performance make engagement behavior a nonuseful construct.

My focal article generated a good deal of both heat and light embodied in 13 commentaries. In this response, I address both the heat and the light by clarifying some of my original comments, responding to criticisms in the commentaries, and using the commentaries to suggest a research agenda for studying the possible influence of stereotypes on real-world personnel decisions.

In this paper, I suggest that our methods for how to train are based historically on what we understand learning to be. I briefly trace the history of instructional design models, which embody both first-generation objectivist and second-generation cognitive constructivist views of learning. I then suggest we are on the cusp of a third-generation instructional model, one that places greater emphasis on the learner forming understanding largely through a process of social negotiation, either in training or on the job. I argue that Web-based instruction offers a technology ideally suited to the social constructivist approach at the core of this third-generation model. I discuss the importance of interaction in learning and suggest why Web-based networked learning may result in more, not less, interaction during training. Finally, I discuss other applications of third-generation learning such as peer mentoring and professional forums.

Assessment centers (ACs) are often designed with the intent of measuring a number of dimensions as they are assessed in various exercises, but after 25 years of research, it is now clear that AC ratings that are completed at the end of each exercise (commonly known as postexercise dimension ratings) substantially reflect the effects of the exercises in which they were completed and not the dimensions they were designed to reflect. This is the crux of the long-standing “construct validity problem” for AC ratings. I review the existing research on AC construct validity and conclude that (a) contrary to previous notions, AC candidate behavior is inherently cross-situationally (i.e., cross-exercise) specific, not cross-situationally consistent as was once thought, (b) assessors rather accurately assess candidate behavior, and (c) these facts should be recognized in the redesign of ACs toward task- or role-based ACs and away from traditional dimension-based ACs.