Statistics in a horticultural journal: problems and solutions

Three researchers examined all articles in volume 139 and the first issue of volume 140 of the Journal of the American Society for Horticultural Science (JASHS) for statistical problems.

Slightly fewer than half appeared to have problems. This is consistent with what has been found for other biological journals. Problems ranged from inappropriate analyses and statistical procedures to insufficient (or complete lack of) information on how the analyses were performed.

A common problem arose from taking many measurements from the same plant, which leads to correlated test results, ignored when declaring significance at P = 0.05 for each test. In this case, experiment-wise error control is lacking.

The researchers believe that many of these problems could and should have been caught in the writing or review process; i.e., identifying them did not require an extensive statistics background. This suggests that authors and reviewers have not absorbed nor kept current with many of the statistical basics needed for understanding their own data, for conducting proper statistical analyses, and for communicating their results.

For a variety of reasons, graduate training in statistics for horticulture majors appears inadequate; the researchers suggest that researchers in this field actively seek out opportunities to improve and update their statistical knowledge throughout their careers and engage a statistician as a collaborator early when unfamiliar methods are needed to design or analyze a research study. In addition, the ASHS, which publishes three journals, should assist authors, reviewers, and editors by recognizing and supporting the need for continuing education in quantitative literacy.

Read the full paper at HortScience.

Publication date: Fri 14 Oct 2016