Managing the Gaps: How Performance Gaps Shape Managerial Decision Making

Abstract

Much literature provides insights on the effect of managerial decisions on organizational performance. This research has given less attention to the determinants, rather than the effects, of variance in managerial decisions. This study seeks to determine whether decisions vary when performance gaps are based on subjective clientele ratings or more objective performance output information. By combining data from an original survey of hospital CEOs, the American Hospital Association and the Centers for Medicare & Medicaid Services, we find that multiple managerial decisions are explained by both historical and social aspirational gaps, but that shifts in priorities vary depending on how performance is defined.

Keywords:

• managerial decisions
• past performance
• performance gaps
• social aspirations

Notes

Notes

1 We use this language to signal that no performance information is entirely objective. Meier and O’Toole (2013) offer a more detailed discussion of the terminology of perceptual and archival performance measures. This terminology has been adopted in other recent empirical studies on performance information and performance management (Meier et al., 2015).

2 Much literature on these two gaps uses the language of social comparisons, which might also be termed peer comparisons. We largely follow this literature in terminology in this study, although we consider peer comparison an appropriate synonym in this context.

3 Appendix A presents more details about the survey response rate. We analyze survey nonresponse bias by first coding in-sample cases as “1” and all other cases as “0” and then estimate a logistic regression model to predict survey participation. Although we observe varying response rates by ownership and service type, the nonresponse biases are minimal (with near 1 odds ratio coefficients).

4 Beyond this overall rating measure, the HCAHPS Survey asks patients specific questions regarding their direct experiences with the hospital where they receive health care (e.g., satisfaction about nurses, physicians, ER treatment, ER waiting time, satisfaction with specialized care services, etc.). We use the overall rating measure because this question is asked across all patients and all hospitals consistently. It also reflects patients’ overall image of a hospital.

5 Because these hospitals operate largely in a fee-for-service system, managing optimal service outputs is a common task for hospital administrators across the three sectors. Research in health care economics in the U.S. context has long demonstrated that hospital managers are often under considerable pressure to increase service outputs through mergers and/or facility expansions. The primary motivation is to maintain and optimize the revenue stream by increasing hospital size and scaling up service outputs (Gaynor & Haas-Wilson, 1999; Hollingsworth, 2008).

6 In our calculation of the two performance gap measures, we logged the number of hospital visits (outpatient and ER visits) given two considerations. Statistically, hospitals vary substantially by size, thus the scale of total service outputs varies substantially across hospitals. Substantively, it is unlikely that a performance gap by one hospital visit or several visits would be deemed as salient enough to trigger managerial responses. The logged measure better reflects performance gaps at scale.

7 An analysis of pairwise correlations shows that the performance gap measures based on patient satisfaction rating have a correlation of 0.34 while the performance gap measures based on hospital service outputs are correlated at 0.10. Performance gaps based on patient satisfaction rating and those based on service outputs do not correlate with each other (all are near zero). These correlation statistics suggest that performance gaps calculated according to different performance indicators need to be modeled separately.

8 Analyses were also run with the unmodified ten categories. Results of ordered logit models containing the original 10-point scale lead to substantively consistent conclusions with the reported models with three response categories. However, models with 10 categories are difficult to interpret in a straightforward way. For example, if a positive and significant coefficient is detected in a 10-category ordered logit, it only means that an increase in the performance gap measure increases the probability of giving the highest priority over all lower categories combined and decreases the probability of giving a score of “1” compared with that for all higher categories.

9 We also tested models that included state fixed effects (not shown) but found these models to have high levels of collinearity. Still, results were not substantively different in direction and significance.

10 If one were considering the effect size associated with every one percentage point change in the satisfaction-based performance gap measure, the effect is rather small. A one unit increase in patient satisfaction rating only decreases the probability of prioritizing community outreach by approximately 0.04 and decreases the probability of giving high priority to patient outcomes by about 0.05.

Additional information

Notes on contributors

Ling Zhu

Ling Zhu is an associate professor of Political Science at the University of Houston. Her research interests include public management, health disparities, social equity in healthcare access, as well as implementation of public health policies at the state and local level.

Amanda Rutherford

Amanda Rutherford is an assistant professor at the School of Public and Environmental Affairs at Indiana University. Her research interests include performance accountability policies, managerial decision making, representation and equity, and education policy.