Data Dispersion Issues
Biological processes are full of variations, and so is clinical research. Statistics can give no certainties, only chances and, consequently, their results are often reported with a measure of dispersion, otherwise called uncertainty. Mostly, standard errors are calculated as a measure for dispersion in the data. For example, in a hypertension study a mean systolic blood pressure after active treatment of 125 mmHg compared to 135 mmHg after placebo treatment may indicate that either the treatment was clinically efficacious or that the difference observed is due to random variation. To answer this the standard errors of the mean results, 5 mmHg each, and a pooled standard error are calculated, √(52 + 52) = 7.07 mmHg. According to the Student’s t-test this result is statistically insignificant: the t-value = (135 – 125)/7.07 = 1.4, and should have been larger than approximately 2. With such a result it is, usually, concluded that the treatment effect is not different from a placebo effect, and that the calculated mean difference is due to random variation, rather than a true treatment effect.
KeywordsTrue Treatment Effect Dispersion Issue Arrhythmic Patient Multiple Logistic Model Clinical Estimator
- Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig JG, Moher D, Rennie D, De Vet HC, for the STARD steering group (2003) Education and debate. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 326:41–44Google Scholar
- Gardner MJ (1989) Confidence interval analysis. BMJ Productions, LondonGoogle Scholar
- Hojsgaard S, Halekoh U (2005) Overdispersion. Danish Institute of Agricultural Sciences, Copenhagen. http://gbi.agrsci.dk/statistics/courses
- Lesterhuis W, Cleophas TJ (2009) Cardiovascular research: decision analysis using binary partitioning. Perfusion 22:88–91Google Scholar
- Levin MD, Van de Bos E, Van Ouwerkerk BM, Cleophas TJ (2008) Uncertainty of diagnostic tests. Perfusion 21:42–48Google Scholar