Book Reviews

According to the Preface, The Cambridge dictionary of statistics aims to provide students of statistics, working statisticians and researchers in many disciplines who are users of statistics with relatively concise deŽnitions of statistical terms. In my opinion, the dictionary fulŽls this promise and much more. First and foremost, it is an indispensable reference book, providing clear and succinct explanations of statistical terms. In general, I Žnd the topics included judiciously chosen and up to date. In fact, it was not easy for me to come up with an omitted term that would merit inclusion, with the possible exception of the currently vogue topic ‘false detection rate’. Importantly, the dictionary is not conŽned to (mathematical) statistics, but also includes entries from allied disciplines such as biostatistics, econometrics, psychometrics and genetics. I believe this modern and more general view of statistics is a deŽnitive merit of the dictionary, making it indispensable for applied statisticians. Another merit of The Cambridge dictionary of statistics is that many entries include references to articles and books. The references are aptly chosen and provide an invaluable guide to further reading. The dictionary also includes brief but entertaining biographies of deceased famous statisticians. Having similar entries on living statisticians would have been useful, but I appreciate that the choice of entries could invite controversy. Although my evaluation of The Cambridge dictionary of statistics is generally enthusiastic, there are inevitably entries that I am not entirely happy with. I will give a nonrandom sample here: for instance, it is stated under Hazard function that ‘It is a measure of how likely an individual is to experience an event as a function of the age of the individual’. However, the time variable need not be age (as is recognized elsewhere in the entry). IdentiŽcation is deŽned as ‘The degree to which there is sufŽcient information in the sample information to estimate the parameters in a proposed model’, although identiŽcation is a property of a statistical model (and not a sample) related to the uniqueness of model parameters. Random coefŽcient models are described only in the context of repeated measures data although applicable to clustered data in general. Duration time is deŽned as ‘The time that elapses before an epidemic ceases’. I Žnd this deŽnition too restrictive; duration or duration time is, for instance, used as a synonym for survival time in econometrics. The dictionary is appealing and is a pleasure to read. I personally enjoy picking an arbitrary page and start reading consecutive entries to gain insight into topics novel to me. The problem is that I Žnd it hard to stop. In summary, The Cambridge dictionary of statistics is a must for any serious statistician’s bookshelf.