One in eight psychology research papers contain massive errors
Half of psychology research papers contain at least one statistical error. One in eight contain such large errors that it affects the paper's statistical conclusions.
This is according to a research study by Michele B. Nuijten, a PhD student at Tilburg University. Her study, titled "The prevalence of statistical reporting errors in psychology (1985 - 2013)", focused on finding and preventing statistical errors.
Nuijten found that 49.6 percent of the 16,695 papers she checked contained at least one inconsistency. And 12.9 percent contained at least one gross inconsistency, an inconsistency large enough to affect the paper's conclusion.
The prevalence of inconsistencies remained relatively stable over the years. And there were also no significant differences between social, clinical and experimental psychology papers.
Nuijten used the tool "statcheck" to help her analyse the thousands studies she used in her research. Statcheck is a statistical tool that detects flaws in studies, similar to spell checkers checking for spelling errors. The tool was developed by Nuijten herself and Sacha Epskamp of the University of Amsterdam.
Statcheck is online and available for everyone to use.
According to the Volkskrant, Tilburg psychologists Jelte Wicherts did a similar study in 2011 and came to the same findings. Daniel Lakens, a statistician at the Eindhoven University of Technology, analyzed Nuijten's research and supports her conclusions. "It's mostly about small mistakes. But there are also researchers who adjust there results t their conclusions rather than vice versa. Shameful", he said to the newspaper.