PubMed · 8114842
The source did not provide an abstract. Follow the original record for more information.
Explore connections, maps & timelines
D B Stryer, W Browner, T Newman. 1994-03-24. More on torturing data.. https://doi.org/10.1056/nejm199403243301212
Cite the original work for its findings. Save a collection to share your selection of sources.
KEEP EXPLORING
Confusion in the teaching of statistical inference dates back to the conflict of Fisher's P-values and significance tests with the Neyman-Pearson hypothesis testing approach. To avoid the well-known pitfalls arising from over-reliance on significance tests and the division of results into 'significant' or 'not significant', many medical journals now insist that presentation of statistical analyses includes confidence intervals as well as or instead of P-values. The confusion over how to report statistical analyses which is evident in the recent medical literature is matched by divergent teaching of hypothesis tests between the 16 U.K. medical schools represented at the April 2000 Burwalls meeting. Suggested guidelines for the teaching of statistical inference to medical students are presented, and possible future developments are discussed.
In this paper we propose a quasi-exact alternative to the exact unconditional method by Chan and Zhang (1999) estimating confidence intervals for the difference of two independent binomial proportions in small sample cases. The quasi-exact method is an approximation to a modified version of Chan and Zhang's method, where the two-sided p-value of an observation is defined by adding to the one-sided p-value the sum of all probabilities of more "extreme" events in the unobserved tail. We show that distinctively less conservative interval estimates can be derived following the modified definition of the two-sided p-value. The approximations applied in the quasi-exact method help to simplify the computations greatly, while the resulting infringements to the nominal level are low. Compared with other approximate methods, including the mid-p quasi-exact methods and the Miettinen and Nurminen (M&N) asymptotic method, our quasi-exact method demonstrates much better reliability in small sample cases.
Standard deviation and standard error have a clear mutual relationship, but at the same time they differ strongly in the type of information they supply. This can lead to confusion and misunderstandings. Standard deviation describes the variability in a sample of measures of a variable, for instance the variability in ages of the members of a group. It represents the degree to which the values are scattered around their mean: the higher the standard deviation the wider the spread. The value of the standard deviation is not influenced by the number of observations in the sample. The standard error is always used for extrapolation: to estimate the intervals between which the true value of a statistic will occur, based on a sample of observations and with a certain degree of certainty. When interpreting a standard error, it is important to know which statistic (mean, percentage, relative risk, odds ratio) is to be estimated. The value of the standard error is strongly influenced by the number of observations in the sample: the bigger the sample, the smaller the standard error and the more accurate the estimation. To avoid confusion it is recommended to report no longer the standard error of the mean but instead the confidence intervals of the mean to estimate the true value of the mean.