Skip to main content
ARS Home » Pacific West Area » Maricopa, Arizona » U.S. Arid Land Agricultural Research Center » Pest Management and Biocontrol Research » Research » Publications at this Location » Publication #362690

Research Project: Ecologically Based Pest Management in Western Crops Such as Cotton

Location: Pest Management and Biocontrol Research

Title: Common statistical mistakes in entomology: blocking and inference space

Author
item Spurgeon, Dale

Submitted to: American Entomologist
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/29/2019
Publication Date: 12/12/2019
Citation: Spurgeon, D.W. 2019. Common statistical mistakes in entomology: blocking and inference space. American Entomologist. 65(4):268-271. https://doi.org/10.1093/ae/tmz064.
DOI: https://doi.org/10.1093/ae/tmz064

Interpretive Summary: Agricultural research often involves experiments that simultaneously examine the effects of multiple factors (i.e., multifactor experiments). Such experiments are important because they allow examination of how the multiple factors affect each other (whether and how much they interact). However, reports of multifactor experiments often ignore the interactions between or among treatments because they are deemed too complicated to explain, or it is not clear how they should be interpreted. When an interaction occurs between or among individual treatments or factors, interpretation of the effects of the individual factors is not straightforward. Their effects always depend on the treatment(s) with which they are combined. In those cases, treating the tests of the individual factors as if they were meaningful will generally result in incorrect or incomplete conclusions. Where an interaction is found to occur, it contains all of the interpretable information from the experiment regardless of the tests of the individual factors. The interaction should be explored through what are called simple-effect tests, where the effects of one factor or treatment are compared within the levels of another factor. The use of simple-effect tests, which are easy to conduct with widely-available statistical software, greatly simplifies the interpretation of factor interactions. Correct interpretation of these interactions not only provide a more accurate assessment of the experimental treatments, but also can expose previously unrecognized biological responses that contribute to an improved understanding of the system under study.

Technical Abstract: Agricultural research often involves experiments that simultaneously examine the effects of multiple factors (i.e., multifactor experiments). Such experiments are important because they allow examination of how the multiple factors affect each other (whether and how much they interact). However, reports of multifactor experiments often ignore the interactions between or among treatments because they are deemed too complicated to explain, or it is not clear how they should be interpreted. When an interaction occurs between or among individual treatments or factors, interpretation of the effects of the individual factors is not straightforward. Their effects always depend on the treatment(s) with which they are combined. In those cases, treating the tests of the individual factors as if they were meaningful will generally result in incorrect or incomplete conclusions. Where an interaction is found to occur, it contains all of the interpretable information from the experiment regardless of the tests of the individual factors. The interaction should be explored through what are called simple-effect tests, where the effects of one factor or treatment are compared within the levels of another factor. The use of simple-effect tests, which are easy to conduct with widely-available statistical software, greatly simplifies the interpretation of factor interactions. Correct interpretation of these interactions not only provide a more accurate assessment of the experimental treatments, but also can expose previously unrecognized biological responses that contribute to an improved understanding of the system under study.