. Q 1. Where can I find a critique of a paper in which statistical errors are explicitly identified in order that I can better understand what to look out for.

A. Have a look at the presentation

This presentation should be read in conjunction with the following paper:

. Q 2. Can you provide examples of bad statistical practice in the analysis of or in the interpretation of the results of clinical trials?

A. There are plenty within the resource
Improving Medical Statistics and the Interpretation of Medical Studies

Additonally, recently the Journal of Medical Internet Research published an article entitled

“Is Cybermedicine Killing You?” – The Story of a Cochrane Disaster

The three critical flaws listed in the abstract are “(1) data extraction and coding errors, (2) the pooling of heterogeneous studies, and (3) a problematic and ambiguous scope and, possibly, some overlooked studies.”

Click here to learn more on what went wrong with this Cochrane systematic review.

. Q 3. Can you provide some examples of the misuse of basic statistics?

A. Yes. Here is a recent in which the authors have wrongly assumed that we can extrapolate with linear regression:

The conclusions of the research are therefore unsound.

Have a look at the following reviews to see what else the readers had to say and to help you further in understanding the relevant issues:

. Q 4. What is regression towards the mean and where can I learn more about it?

A. Regression towards the mean is a statistical phenomenon according to which extreme data (or data, with atypically high or low values) must revert back to the overall average for the data. Without an adequate approach to study design and interpretation of data, this phenomenon can be mistaken for a deterioration or improvement over time (or “change from baseline”) which arises from the intervention used. To learn more, you are encouraged to refer to the following resources:

1. The February 2015 Year 3 MBChB Programme Theme session notes entitled  Getting into the groove with questionnaire design and statistical analysis: Preparing for interprofessional collaboration below.

Please download the files below into the same folder as the PowerPoint version of the presentation in order to ensure access to all audio and audiovisual content.

These particular lecture notes contain 4 key questions on regression to the mean (see slide 31 of 33) which are fully addressed through the resources recommended below.

2.
a) Statistics notes: Regression towards the mean
Authors: J Martin Bland and Doug G Altman
 

b) Regression to the Mean (Author: J Martin Bland)

c) Regression towards the mean or why was Terminator III such a disappointment? (Author: J Martin Bland)

(The above resources illustrate, among other things, why merely comparing repeated measures over time on administration of a particular intervention can be a misleading way of assessing the effectiveness of the intervention. Links to articles for further reading are also provided within resource c).)  

3. Regression to the mean: what it is and how to deal with it  [Look out for the recommended statistical methods.] Authors: Adrian G Barnett, et al.
(This resource provides recommendations on statistical methods for managing regression to the mean. You should pay particular attention to section 2, which provides a gentle introduction to ANCOVA (ANalysis of COVAriance).  In this section you will learn about the effectiveness of adjusting for the baseline measure as a covariate. In simple terms, you should think of a covariate as a continuous variable which may influence the outcome. Within the context of ANCOVA, we need to adjust for such a variable.)

 

 

N.B. If you discover a favourite example of regression towards the mean which you would like to share, please be sure to let me know.

Misuse of Statistics: Some Statistical Blunders Discovered Within Refereed Publications and Some Statistical Phenomena to Look out for by Margaret MacDougall is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.