Abstract
A survey of farmers rating the severity of crop pest infestation in their fields was conducted in the Midwest in 1992. The purpose of the present study was to detennine summary variables of the pest infestation ratings and the effect of region, soil type, and tillage on these summary variables. The pests were in the following six categories: perennial and annual weeds, insects and diseases of com (Zea mays L.) and insects and diseases of soybean (Glycine max (L.) MerritT). Categorical models were used to analyze individual pest ratings. A non-parametric method, the Sheirer-Ray-Hare extension of the Kruskal-Wallis test to factorials, was used as a substitute when categorical model analysis failed. When both methods could be performed the results were very similar and were also close to results of general linear model analysis of the raw data. Region and tillage were the most significant factors. Variables for which tillage was significant had higher mean ratings in no-till than in conventional till. When region was significant the eastern region had higher mean ratings than the western region in most cases. Principal component analyses produced several informative summary sets of three, seven, and thirteen eigenvalues, respectively. The three rotated components using three eigenvalues consisted of soybean pests, perennial and annual weeds, and corn pests, respectively. These rotated components showed a strong partitioning for region and tillage, and to a lesser degree for soil type. Using seven eigenvalues resulted in further division of components. The corn pest component of the three eigenvalues was divided into a component of corn insects and a component of corn diseases. The rotated components based on thirteen eigenvalues further divided the soybean pests into a component of insects and a component of diseases, and divided each of the perennial weeds, annual weeds, and corn insects into components of higher and lower mean ratings. Analyses of each of the sums of variables loading on the thirteen rotated components resulted in a very highly significant (p < 0.0001) tillage effect in all but one sum. The region effect was very highly significant in half of these variable sums, while the soil type effect was very highly significant in only three sums.
Keywords
Non-parametric factorial analysis, principal components, component analysis
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Aref, Susanne
(1997).
"ANALYSIS OF A MIDWEST FARMER SURVEY OF PEST INFESTATION,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1298
ANALYSIS OF A MIDWEST FARMER SURVEY OF PEST INFESTATION
A survey of farmers rating the severity of crop pest infestation in their fields was conducted in the Midwest in 1992. The purpose of the present study was to detennine summary variables of the pest infestation ratings and the effect of region, soil type, and tillage on these summary variables. The pests were in the following six categories: perennial and annual weeds, insects and diseases of com (Zea mays L.) and insects and diseases of soybean (Glycine max (L.) MerritT). Categorical models were used to analyze individual pest ratings. A non-parametric method, the Sheirer-Ray-Hare extension of the Kruskal-Wallis test to factorials, was used as a substitute when categorical model analysis failed. When both methods could be performed the results were very similar and were also close to results of general linear model analysis of the raw data. Region and tillage were the most significant factors. Variables for which tillage was significant had higher mean ratings in no-till than in conventional till. When region was significant the eastern region had higher mean ratings than the western region in most cases. Principal component analyses produced several informative summary sets of three, seven, and thirteen eigenvalues, respectively. The three rotated components using three eigenvalues consisted of soybean pests, perennial and annual weeds, and corn pests, respectively. These rotated components showed a strong partitioning for region and tillage, and to a lesser degree for soil type. Using seven eigenvalues resulted in further division of components. The corn pest component of the three eigenvalues was divided into a component of corn insects and a component of corn diseases. The rotated components based on thirteen eigenvalues further divided the soybean pests into a component of insects and a component of diseases, and divided each of the perennial weeds, annual weeds, and corn insects into components of higher and lower mean ratings. Analyses of each of the sums of variables loading on the thirteen rotated components resulted in a very highly significant (p < 0.0001) tillage effect in all but one sum. The region effect was very highly significant in half of these variable sums, while the soil type effect was very highly significant in only three sums.
