I once listened to Professor Thomas Allen Lang's lecture on thesis writing. He is very nice, and his books are also good, and he is very operational. How to write and publish in his manuscript &; On page 154 of Presentation of Health Science, he concluded that the statistical analysis part should include the following contents: statistical description part, all basic statistical methods and analysis schemes (such as ITT or PP), description of sample size, grouping method, setting of test level and statistical analysis software used.
Page 155 of this book also says: statistical analysts can help authors analyze data reasonably and interpret the analysis results correctly, and can also be responsible for writing statistical analysis parts. He suggested that statistical analysts should be one of the authors, so that they might not be careless and irresponsible.
Regarding the writing of medical statistical analysis, in fact, he also has a book "How to report statistics in medicine", which is more professional in the reporting of statistical analysis.
Anyway, since this article is a summary of "statistical analysis", let's summarize it. Personally, I think the writing of Statistical Analysis should include the following contents:
(1) Sample size estimation and follow-up/data collection;
(2) Software and methods for data entry and management;
(3) Statistical analysis software and analysis scheme used in this study;
(4) The methods of statistical description are divided into measurement and counting data;
(5) Statistical inference methods can be divided into single factor and multi-factor;
(6) Selection of inspection level.
For some "you know" reasons, many common papers do not estimate the sample size, nor do they distinguish different analysis schemes (ITT/PP). Then a simple example is as follows:
In this study, ... database is used for data entry and management, and data entry is verified by double entry. Statistical analysis of research data with …… software. The comparison between the measurement data groups that conform to the normal distribution is carried out by independent sample t-test or one-way ANOVA, that is, the comparison between the measurement data groups that conform to the normal distribution. In the multivariate analysis, multiple linear /logistic regression analysis methods are used to analyze the influencing factors. All the tests were conducted with bilateral P.
Some people say that I want to write Statistical Analysis in English. What should I do? Similarly, you need to read more excellent articles of others, and then use their sentence patterns to construct your own statistical analysis content. The sentence patterns for reference are:
(1) data collection: collect the research data on the standard form, check the integrity of the data, and type it into the ... database.
(2) Statistical software: All statistical analysis was carried out using SAS 9.2 (SAS Institute Inc, Cary, North Carolina).
(3) Statistical description: ... using the mean, median, standard deviation and the 75th percentile of continuous variables to describe; Frequency and proportion are used to classify variables.
(4) Single factor analysis: use two-sample independent test/one-way analysis of variance (ANOVA)/ nonparametric test (Kruskal-Wallis test)/Pearson's x2 test or Fisher exact test to compare the differences between ...
(5) Multivariate analysis: multivariate linear regression/multivariate binary logistic regression /Cox proportional risk was used to estimate.
(6) Test level: P value less than 0.05 (bilateral significance test) is considered statistically significant among all analysts.