Half of the psychology articles in scientific journals contain
one or more statistical mistakes. In
most cases, this is the result of negligence, such as typos and mistaken
rounding up or down of statistics. In
one out of eight cases, however, major negligence was found in the
calculations, which impacted directly on the statistical results. This was discovered in a study carried out by
methods expert Michele Nuijten, who tested a random sample of 30,000
articles. In most of the cases, the
mistakes were in line with the opinions expressed by the researchers themselves
on the subject of their papers. Together
with another researcher, Nuijten has developed a program which can check
statistical inaccuracies. It can assess
whether the mistakes are accidental or not.
In papers where the mistakes were not accidental, frequently more errors
were found.
No comments:
Post a Comment