Retractions
Each Welch two sample t-test was performed for each pair of years as
described in the Methods section. Table \ref{735551} shows the mean percent
increase in the number of submissions a journal received if it had one
or more retractions in the previous year, as well as the mean percent
increase in the number of submissions a journal received if it had no
retractions. The p-values are reported; 5 out of the 11 pairs are statistically significant. (For clarification: 0.01 would represent
a 1% increase and -0.01 would represent a 1% decrease; for example.)
For every pair of years that is statistically significant, the percent
increase in the number of submissions a journal receives is lower for
journals that issue retractions versus journals that issue no
retractions. The journals that issued retractions in 2007 show on
average a percent decrease in the number of submissions received in
2008.
One reason for the null results may be the removal of data from the analysis. There may also be an effect due to new journal starts, which increased in later years and are not representative of established journals.