The relationship between scientific publishing retractions and democracy

Objective: To determine the magnitude and shape of the relationship between the proportion of retracted papers and the level of democracy at the country level.Methods: This was a cross-sectional study. The number of retracted articles for each country was collected from the Retraction Watch Database. The mean Democracy Index score 2006–2021 by the Economist Intelligence Unit was used as the independent variable. The number of citable documents 1996–2021 from SCImago was used to account for the population size of retractions. Poisson family regression and linear regression were used to explore the relationship between the proportion of retractions and the Democracy Index score. The adjusted model included: non-violent mass campaigns, gross domestic product per capita, Human Development Index, industry’s share of the economy in percent, length of executive tenure, location, the Muslim share of the population, the number of top universities, and plurality/majority system. R was used for data handling, analysis and reporting.Results: Overall, data from 241 countries were analyzed. As of January 24, 2023, the Retraction Watch Database had retractions from 166 countries and territories. The highest number of retractions belonged to China (n=18,352), the United States (n=4,668), and India (n=2,554). Twenty-four countries had one retraction. The mean (and standard deviation) for the number of retractions and Democracy Index score were 180.4 (1246.42) and 5.5 (2.19), respectively. In the Poisson-inverse Gaussian regression models, the coefficient for the mean Democracy Index score was −0.120 (exp=0.877, P<0.001) in the unadjusted model and −0.040 (exp=0.961, P=0.630) after adjustment. In the linear regression models, the coefficients were −0.102 (P=0.007, R2=0.037) in the unadjusted model and −0.033 (P=0.623, R2=0.185) in the adjusted model. Sensitivity analyses using zero-truncated and outlier-eliminated datasets yielded similar results.Conclusion: There was an inverse association between the proportion of retractions and the Democracy Index scores. Democratic nations can provide an accountable research environment promoting ethical behaviour resulting in fewer number of publication retractions.