In sharp contrast to previous studies suggesting that errors account for the majority of retracted scientific papers, a new analysis—the most comprehensive of its kind—has found that misconduct is responsible for two-thirds of all retractions. In the paper, misconduct included fraud or suspected fraud, duplicate publication and plagiarism. The paper's findings show as a percentage of all scientific articles published, retractions for fraud or suspected fraud have increased 10-fold since 1975.
This study was published in 2012, but I don't believe it got any better since then.
