Statistical Power, P ‐Values, and the Positive Predictive Value
JC Barnes, SJ Linning�- The Encyclopedia of Research�…, 2021 - Wiley Online Library
The Encyclopedia of Research Methods in Criminology and Criminal�…, 2021•Wiley Online Library
This chapter focuses on a factor thought to have one of the greatest influences on the crisis
of confidence: low statistical power. Statistical power is clearly a foundational principle of
inferential statistics, yet it is commonly overlooked. If an entire discipline is built on
underpowered research, its whole body of evidence becomes suspect. This is all to say that
statistical power is the virus; the crisis of confidence is the symptom. The chapter describes
the fundamental principles behind this. It explains hypothesis testing and its limitations. The�…
of confidence: low statistical power. Statistical power is clearly a foundational principle of
inferential statistics, yet it is commonly overlooked. If an entire discipline is built on
underpowered research, its whole body of evidence becomes suspect. This is all to say that
statistical power is the virus; the crisis of confidence is the symptom. The chapter describes
the fundamental principles behind this. It explains hypothesis testing and its limitations. The�…
Summary
This chapter focuses on a factor thought to have one of the greatest influences on the crisis of confidence: low statistical power. Statistical power is clearly a foundational principle of inferential statistics, yet it is commonly overlooked. If an entire discipline is built on underpowered research, its whole body of evidence becomes suspect. This is all to say that statistical power is the virus; the crisis of confidence is the symptom. The chapter describes the fundamental principles behind this. It explains hypothesis testing and its limitations. The chapter then discusses the characteristics of statistical power and the positive predictive value (PPV) needed to generate more trustworthy inferences from these tests. The PPV can be estimated, but more information besides the α‐level and a P ‐value is needed. The chapter concludes with some suggestions that criminologists can incorporate into their research to help mitigate a crisis of confidence.
Wiley Online Library