Publication Bias

In statistics, biases are systematic errors that can occur in studies or experiments. When bias occurs, some responses are chosen or others are preferred. Interestingly, this bias occurs in a large proportion of studies, at least to some extent. So what does publication bias consist of and what are the reasons behind it?

The key is to understand this concept so that it can be avoided, minimized, or corrected. Misconceptions can occur at any stage of the research process. Especially in the design, implementation, analysis, presentation and publication phases.

Therefore, publication bias refers to the tendency to publish studies that show only positive results.

Positive results are when researchers favor some findings over others. These findings are statistically significant, such as correlations or differences.

On the other hand, researchers do not publish negative results. As you can imagine, negative results are the opposite of positive results, so there are no statistically significant correlations or differences.

Another form of bias is that researchers publish more nice or popular results or articles.  Therefore, it would be more likely that a journalist would publish an author’s article.

In the middle of the last century, Theodore D. Sterling first introduced the term publication bias.  He pointed out that magazines are more likely to publish studies whose results are statistically significant. On the other hand, journals generally do not publish studies whose results are less relevant.

Demonstration of publication bias

So one of the first handlers of the bias was Sterling. He reviewed all articles in four magazines for one year (1955–1956). Afterwards, he noted that when researchers conducted relevant experiments, 97% of the articles rejected the null hypotheses.

According to the results of the study, there are more published studies than unpublished studies with a ratio of 128: 1 and 1: 1. In most studies, the ratio is 10: 1 to 1: 1.

Factors affecting bias

According to Maria Carmen Rosa Garrido, a member of the Spanish Research Ethics Committee, there are several factors in the publication bias:

• First, the author’s decision not to publish results that are not statistically significant in their study.
• Second, the journalist’s refusal to publish studies that show negative results. This is the case even if the methodological quality is sufficient to ensure the reliability of the results.

• Third, the exclusion of these studies from bibliographic searches by other researchers.

Eliminate or prevent publication biases

To prevent or eliminate bias, some suggest that:

• Hypothesis tests would be removed
• Peer reviews would be carried out
• The publication of studies with insufficient sample sizes would be discontinued
• A more positive attitude towards unmarked results would be developed
• Peer review and publication processes would be improved

Bias assessment

Several statistical procedures have been developed to assess whether the survey sample is biased and to assess the impact of this distortion.

Most are based on the assumption that in some specific areas, the small sample size of studies can produce a relatively large impact size. Surveys with a large sample size should also produce a similar effect as population size.

Journalists and publishing bias

What causes publication bias? It seems that the publication of heresies that attract attention in the medical literature is the work of journalists.

This suspicion has some basis. In 1980, the British Medical Journal stated that their ideal article would describe “results that affect clinical practice… and findings that improve prognosis or simplify the treatment of common diseases”.  Consequently, they inadvertently referred to the publication bias.

The most effective way to prevent delusions is to register all clinical trials. In order to move forward, the aim should be prevention.

Finally, it is important to stress the importance of this bias in the scientific community. When using biased evidence, decisions may not be optimal. In the medical community, therefore, this can lead to inappropriate prescriptions that may not be optimal for patients or the scientific community in general.