When attaining the best sample is out of reach: Nonprobability alternatives when engaging in public administration research

ABSTRACT

Across all social science disciplines, but in particular public administration, there is a shared concern about the costs of using traditional random samples to generate data, and its impact on researchers’ ability to engage in “quality” research. As a result of these costs, more academics, practitioners, and students are turning to nonprobability sampling methods. However, beyond the notion that these sampling strategies reduce the external validity of findings, individuals engaging in these strategies are doing so in ill-conceived ways due to the lack of attention and examples within mainstream public administration literature that provide researchers with the knowledge on how to best utilize these strategies. As a result, this article seeks to provide public administration practitioners, Master of Public Administration students, and scholars an understanding of and guidance in deciding to utilize three nonprobabilistic methods, convenience sampling, purposive sampling, and sample matching. This article is intended to be used as a supplement to materials and texts already currently being used within methods courses.

KEYWORDS:

• Convenience sampling
• nonprobability sampling
• purposive sampling
• propensity score matching
• sample matching
• sampling methods

Acknowledgment

The author thanks Amitra Wall, Patrick McGovern, and Peter Yacobucci for their aid in the development of this article.

Notes

^1. It should be noted that by adding a third step to this process that controls for the times in the day in which a researcher attempts to recruit participants is known as time–space sampling, which has been used to draw samples of certain minorities, hard-to-reach (Kish, 1991; Sudman, Sirken, & Cowan, 1988), and hidden populations (Ferreira, De Oliveira, Raymond, Chen, & McFarland, 2008; Mackellar et al., 2007; Remafedi, Jurek, & Oakes, 2008; Semaan, 2010). However, these samples also suffer from similar limitations indicative to convenience and purposive sampling.

^2. Other alternative methods of matching include one-to-many (Ming & Rosenbaum, 2001; Rubin & Thomas, 2000; Smith, 1997; Thoemmes & Kim, 2011) and full matching (Hansen, 2004; Rosenbaum, 1991; Stuart & Green, 2008) for the creation of samples and analysis.

Additional information

Notes on contributors

Jason D. Rivera

Jason D. Rivera is an assistant professor in the Public Administration and Nonprofit Management program at SUNY Buffalo State. His research interests include disaster and emergency management, representative bureaucracy, and governance.