Statistics for Making Decisions

Decision-making is a common mental activity that involves choosing one course of action from a list of options. Statistics for Making Decisions proposes a new paradigm for statistical practice that emphasizes the use of prior information and the consequences of errors. The volume appeals to analysts who deal with simple statistical problems and promotes a principled solution based on the idea of a currency for error, which is implemented by selecting the option for which the expected loss is smallest. Nicholas T. Longford is a senior statistician at Imperial College, London, specializing in statistical methods for neonatal medicine.

Format: Hardback
Length: 291 pages
Publication date: 03 March 2021
Publisher: Taylor & Francis Ltd

Decision-making is a fundamental mental process that permeates our personal, professional, and public lives. It involves selecting one option from a limited set of possibilities. In the realm of statistics, decision-making takes center stage, with its theory emerging as a novel paradigm for statistical practice. This paradigm emphasizes the importance of prior information and the potential consequences of different types of errors. The ultimate goal is to tailor a course of action that aligns with the specific needs and perspectives of the client or sponsor of the analysis.

The authors of this volume aim to replace hypothesis testing as the primary method in statistical inference. They argue that hypothesis testing fails to account for the significance of errors, which are inherently relevant to the client's interests. The book appeals to analysts who encounter basic statistical tasks such as comparing two samples or determining the positivity or negativity of a parameter. It combines a critical assessment of hypothesis testing with the promotion of a principled solution based on the concept of a currency for error. This currency is intended to be spent as minimally as possible by selecting the option with the smallest expected loss, known as the Bayes rule.

However, the cost of implementing this approach involves providing a more detailed description of the options and eliciting and quantifying the consequences (ramifications) of errors. This is a task that clients often perform informally and sometimes inexpertly after receiving analysis outputs in established formats, such as the verdict of a hypothesis test or an estimate and its standard error. As a scientific discipline and profession, statistics has the potential to excel in this area and provide clients with more accurate and valuable insights.

In conclusion, decision-making is a crucial mental activity that plays a significant role in our daily lives. Statistics, as a field, recognizes the importance of decision-making and strives to develop new paradigms and methodologies that account for the complexities and consequences of errors. By adopting a rigorous approach that incorporates prior information, the Bayes rule.

Weight: 604g
Dimension: 162 x 241 x 25 (mm)
ISBN-13: 9780367342678