Skip to main content

Archived Comments for: How to spot a statistical problem: advice for a non-statistical reviewer

Back to article

  1. Independent statistical review for any study under review

    Siddharudha Shivalli, Yenepoya Medical College, Yenepoya University, Mangaluru, Karnataka, India

    5 February 2016

    I must thank Greenwood and Freeman for providing a crisp overview on statistical issues while reviewing an article. I completely agree for an independent statistical review for any study under review.

    Following are the few common issues which I have come across

    • Not reporting the outcome variable with their precision (eg, 95% confidence interval)
    • Not testing the normal distribution of the data before applying parametric tests
    • Reporting of ‘p’ values as ‘0.0000’ instead of <0.001
    • Not reporting the R2 of the applied regression model
    • Interpreting ‘association’ as ‘causality’ in cross sectional studies
    • Failure to justify the sample size and representativeness
    • Generalizing the findings of hospital based studies on community etc.

    One of the best ways to tackle these issues is to follow the standard reporting guidelines. Journals should always ask the authors to upload the relevant research reporting guideline/checklist/statement. In addition, submitting all the relevant data along with the manuscript would be of great use in statistical review.

    Competing interests

    Author declares that there is not competing interest about this publication