Dataset Information

Reducing the standard deviation in multiple-assay experiments where the variation matters but the absolute value does not.

ABSTRACT: When the value of a quantity x for a number of systems (cells, molecules, people, chunks of metal, DNA vectors, so on) is measured and the aim is to replicate the whole set again for different trials or assays, despite the efforts for a near-equal design, scientists might often obtain quite different measurements. As a consequence, some systems' averages present standard deviations that are too large to render statistically significant results. This work presents a novel correction method of a very low mathematical and numerical complexity that can reduce the standard deviation of such results and increase their statistical significance. Two conditions are to be met: the inter-system variations of x matter while its absolute value does not, and a similar tendency in the values of x must be present in the different assays (or in other words, the results corresponding to different assays must present a high linear correlation). We demonstrate the improvements this method offers with a cell biology experiment, but it can definitely be applied to any problem that conforms to the described structure and requirements and in any quantitative scientific field that deals with data subject to uncertainty.

SUBMITTER: Echenique-Robba P

PROVIDER: S-EPMC3813515 | biostudies-literature | 2013

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Reducing the standard deviation in multiple-assay experiments where the variation matters but the absolute value does not.

Echenique-Robba Pablo P Nelo-Bazán María Alejandra MA Carrodeguas José A JA

PloS one 20131030 10

When the value of a quantity x for a number of systems (cells, molecules, people, chunks of metal, DNA vectors, so on) is measured and the aim is to replicate the whole set again for different trials or assays, despite the efforts for a near-equal design, scientists might often obtain quite different measurements. As a consequence, some systems' averages present standard deviations that are too large to render statistically significant results. This work presents a novel correction method of a v ...[more]

PMID: 24205158

Similar Datasets

Project description:BackgroundA recent review found that 45% of meta-analyses included statistical errors, of which, the most common was the calculation of effect sizes based on standard error (SE) rather than standard deviation (SD) [the SE/SD mix-up].ObjectivesThe first aim of this study was to assess the impact of the SE/SD mix-up on the results of one highly cited meta-analysis. Our second aim was to identify one potential source of the SE/SD mix-up, by assessing how often SE is reported as a measure of sample variability in randomised controlled trials in sports medicine.MethodsWe checked for potential SE/SD mix-ups in a 2015 meta-analysis of randomised controlled trials reporting the effects of recreational football interventions on aerobic fitness in adults. We corrected effect sizes affected by SE/SD mix-ups and re-analysed the data according to the original methodology. We compared pooled estimates of effect sizes from our re-analysis of corrected values with those of the original study. To assess how often SE was reported instead of SD as a measure of sample variance, we text mined results of randomised controlled trials from seven sports medicine journals and reported the proportion reporting of SE versus SD.ResultsWe identified potential SE/SD mix-ups in 9/16 effect sizes included in the meta-analysis describing the effects of football-based interventions versus non-exercise control. The published effect size was standardised mean difference (SMD) = 1.46 (95% confidence interval [CI] 0.91, 2.01). After correcting for SE/SD mix-ups, our re-analysis produced a smaller pooled estimate (SMD = 0.54 [95% CI 0.37, 0.71]). The original pooled estimate for trials comparing football versus running interventions was SMD = 0.68 (95% CI 0.06, 1.4). After correcting for SE/SD mix-ups and re-analysis, the effect was no longer statistically significant (SMD = 0.20 [95% CI - 0.10, 0.49)]). We found that 19.3% of randomised controlled trials reported SE rather than SD to describe sample variability. The relative frequency of the practice ranged from 0 to 25% across the seven journals sampled.ConclusionsWe found the SE/SD mix-up had inflated estimates for the effects of football on aerobic fitness. Meta-analysts should be vigilant to avoid miscalculating effect sizes. Authors, reviewers and editors should avoid and discourage (respectively) the practice of reporting SE as a measure of sample variability in sports medicine research.

Dataset Information

Reducing the standard deviation in multiple-assay experiments where the variation matters but the absolute value does not.

Publications

Reducing the standard deviation in multiple-assay experiments where the variation matters but the absolute value does not.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets