When the Penn State football scandal exploded in 2012, observers tended to frame events in terms of individuals behaving badly or irresponsibly. The perpetrator of child abuse was convicted and sent to prison; the head football coach was fired; the president of the University and several senior administrators were terminated; and the former head of the Board of Trustees was forced to resign. Certainly, these actions were understandable under the circumstances. Terrible crimes had been committed and covered up for over a decade. Nevertheless, an exclusively individual focus overlooks the roles of 9 groups whose collective behavior first allowed the criminal acts to occur and then put an end to them. The groups included the children's families and high school coaches, the Penn State football coaching staff, the Penn State senior administration, the Penn State Board of Trustees, the Second Mile charitable organization, the Centre County Pennsylvania criminal justice system, Penn State students, the Big 10 athletic conference, and the National Collegiate Athletic Association. This article employs a group and intergroup perspective to analyze key events and to explain both the dysfunctional systemic behavior and the corrective actions.

For many, Penn State is synonymous with football and Coach Joe Paterno, an association that has been a source of pride for many alumni, staff, and students for years. As Alderfer's (2013) article summarizes, that overlap has now made the horrific behaviors of the former football coach Sandusky indelibly connected to Penn State. Alderfer argues that the scandal can be explained by the actions or inactions of nine Penn State groups “whose collective behavior first allowed the criminal acts to occur and then put an end to them” (p. 117).

The tragic failure of Penn State University to effectively respond to years of child sex abuse perpetrated by Jerry Sandusky was both a breakdown of leaders and of leadership systems. Numerous individual leaders at Penn State had the knowledge, power, and interpersonal influence to effectively intervene in support of Jerry Sandusky's victims. However, fully understanding how this tragedy occurred also requires an examination of the organizational system of leadership that enabled each leader to rationalize the cover up of the sexual abuse of children. Alderfer's (2011) laws of embedded intergroup relations are useful for understanding the organizational dysfunction at Penn State; especially when those laws are integrated with other theories of organizational psychology, including the social identity maintenance theory of groupthink, the romance of leadership, and authentic leadership. The integration of these theories suggests specific strategies and tactics for preventing similar lapses of ethical behavior in the future through the development of leaders and leadership systems.

Alderfer's (2013) piece on the Sandusky/Penn State tragedy reminds us that leader-centric analyses, the norm in leadership studies, often miss the mark. Alderfer joins a growing list of writers who increasingly recognize that leadership consists of three key elements in a triangle: leaders, followers, and environments. The Penn State scandal highlights how a conducive environment, typified by centralized power and an absence of checks and balances, coupled with flawed leaders and the actual assistance or quiet submission of certain followers, can lead to disastrous outcomes. As Alderfer observes, leadership is a social, or group, process. Leadership success or failure depends on group results, and group results involve more than just leaders and their characteristics and actions. Yet, over three-quarters of articles in scholarly journals consistently overlook the role of organizational environments and followers (Porter & McLaughlin, 2006), focusing instead on leader traits and behaviors (Kaiser, Hogan, & Craig, 2008; Thoroughgood, Padilla, Hunter, & Tate, 2012).

The series of events in the 2011 Penn State sexual abuse scandal were tumultuous and complex. Alderfer's (2013) focal article on the group-level phenomena surrounding the scandal provides a unique lens to view these events. However, questions remain about how relationships both within and between the groups involved in the scandal resulted in these outcomes. In particular, why did members of groups within Penn State fail to act, whereas people who belonged to other social groups took action? In short, we agree that group and intergroup boundaries were important in this situation, but would like to further elaborate on the underlying mechanisms behind their significance. Beyond the fact that individuals belonged to different groups, what aspects of group dynamics explain the differences in their perceptions, attitudes, and behaviors in this scandal?

More than any other man, Mr. Paterno is Penn State – the man who brought the institution national recognition, the man who built a football program based on honor for 46 years, the winningest football coach in Division I history…Paterno is at the core of the university's sense of identity. (Guarino, 2011, November 10)

We believe that Alderfer's (2013) intergroup analysis of the Penn State scandal can be enriched by focusing on the special role that Joe Paterno played in the events leading up to the 2010 grand jury that brought the scandal to light outside the university community. On the basis of the social identity theory of leadership (Hogg, 2001), we argue that the community afforded Paterno the trust that he appears to have abused to suppress the scandal because he embodied the norms, values, and goals of Penn State in a way that made him prototypical of the university. By making the broader group category salient, Paterno was able to shape the behavior of representatives of many sub-groups described by Alderfer within the administration, athletics, and community. Coupled with a high degree of institutional ambiguity, Paterno's prototypicality likely enabled him to suppress the “constructive diversity” (p. 123) of checks and balances that should have operated at Penn State. This disruption to effective intergroup processes prohibited necessary oversight over the PSU athletic program and the Second Mile that otherwise might have uncovered Jerry Sandusky's crimes much earlier.

The current research literature on voluntary employee turnover exhibits at least four shortcomings: low predictive validity, excessive exuberance with predictors, low rigor, and little relevance. Existing theories of voluntary employee turnover research are under specified, as none contains the full range of variables originally hypothesized by March and Simon as contributing to individual employees' decisions to quit their jobs. March and Simon (1958) contributed to the current state of affairs by making at least one assumption that subsequent results do not support. Subsequent efforts to expand theory and to detect and integrate new explanatory constructs have led to a much deeper understanding of the same small portion of turnover variance explained in employee decisions to quit. Deficiencies in current approaches and what a nondeficient model of voluntary employee turnover might look like are described. Directions for future voluntary turnover research are identified, as are past trends that need to be discontinued.

Dr. Russell (2013) provocatively critiqued turnover research, expressing a sentiment that we share—namely, the lamentable modest predictability of turnover. All the same, we disagree with certain criticisms of turnover theory, methodology, and practicality. We organize our reactions into sections: predictive validity for the standard turnover criterion; other criteria for model evaluation; incremental validity controlling quit intentions; Russell's proposed methodology, the potential biases of the Russell and Van Sell (2012) test; and an alternate approach by Hom, Mitchell, Lee, and Griffeth (2012).

Professor Russell proposes a decision model of turnover in which the attractiveness of the current job is compared with that of an alternative. In turn, an employee chooses the option with the highest judged attractiveness. For example, “Employees make decisions to quit based on the relative attractiveness of their current job compared to alternative jobs or activities” (2013, p. 163). The attractiveness of one's current job and alternative are estimated by a regression equation assessing various attributes of the two targets (i.e., current job and alternative). Evoking March and Simon (1958) for a theoretical foundation, Professor Russell offers a subjectively “rational model” for the choice to stay or leave based on expectancy and expected value type decision models. In his empirical work, he uses a “policy capturing” simulation to identify how new hires personally weigh various job attributes when deciding whether they would quit hypothetical jobs varying in those attributes (Russell & Van Sell, 2012). When these weights are applied to employees' actual survey perceptions of the levels of job attributes, the resulting “simulated turnover intention” score predicts turnover better than a survey measure of quit intentions or job attributes alone. The inference is that turnover scholars can make substantial progress toward the prediction of actual turnover by using this model.

Craig Russell has done us all a service in his review of current turnover research (2013), but I think he is too severe in his judgment of the validities and relevance of our field's research, and he encourages us to go down some questionable research paths in the future. Despite his expertise in the field, there are several points where I believe other views are more attractive.

We can recognize the usefulness of assessing so many studies and meta-analyses done in recent decades, as summarized in his Table 1 (drawn from Griffeth, Hom, & Gaertner, 2000). But let us also note that Russell would have done readers a bigger favor if the results were presented in some logical order. Right now, not even an alphabetic order among the variables is evident. This is a flaw that obscures key understandings. Looking at the data another way can make one more optimistic about its predictive validity and relevance.