Statistics mean the systematic error that can occur in the results of studies or research, when a bias occurs, what actually happens is the production or selection of some responses to the detriment of others (1). Bias occurs in much of the research, to a greater or lesser extent. However, how does publication bias occur?
The fundamental question is to know it to avoid, minimize or correct it (2), bias can occur at any stage of the search process, i. e. in planning, performing, analyzing, presenting results or in subsequent publication ( 3).
- So is publication bias defined as the tendency to publish research with positive results?Where are there associations or differences between groups? And the tendency to stop doing it with these surveys that yield negative results?Where no association is found.
- No difference is found.
This problem arises in relation to the publication of articles and books by researchers for certain reasons, such as that positive research is much more peer-reviewed and also because it is easier to publish, so the author favors this type of result when sending to an editor or magazine for publication (4).
In the middle of the last century, Sterling defined publication bias, said studies showing statistically significant results are more likely to be published; on the contrary, studies with non-significant results are less published (5).
As mentioned, one of the first authors to draw attention to publication bias was Sterling, who reviewed all articles published in 4 journals for one year (1955 or 1956), so he showed that 97% of articles using the statistical parameter of meaning rejected the null hypothesis (9).
Based on the results of the research on the subject, the published studies refer to those that are not significant in a ratio between 128: 1 and 1: 1, most studies have a quote ratio between 10: 1 and 1: 1 (7).
According to the author María Carmen Rosa Garrido, there are currently several factors that affect the publication bias (5):
To prevent and eliminate publication biases, some authors (6) propose:
Several statistical procedures have already been developed to assess whether a sample of a particular study is biased, or also to assess the impact of that bias.
Most are based on the assumption that, for a given area, studies with small samples should produce a relatively imperfect range of effect size, while studies with large samples should produce data on effect size closer to the population. of the actual effect (6).
Why does a publication bias occur? It appears that, at least with regard to medical literature, it is thought that the bias of posting positive results stems from revisions (7).
This suspicion has some basis. In 1980, the British Medical Journal reported that its ideal article described “results that affect clinical practice. “And discoveries that improve prognosis or simplify the treatment of common diseases?. Thus, they alluded to the publication bias, even if it is not intentional.
The most effective measure to prevent publication bias is the registration of all clinical trials (7), so prevention may be the goal to target (7).
It is important to understand the importance of this bias in the scientific community. When using brain tests, the decisions made when analyzing the results of certain investigations may not be the best (4). This could trigger inadequate prescribing decisions that may not be best for the patient, or for the scientific community at large.