The bradley-terry model for handling categorical response variables from farmer participatory trials
![Thumbnail](/bitstream/handle/11295/32376/The%20bradley-terry%20model%20for%20handling%20categorical.pdf.jpg?sequence=4&isAllowed=y)
View/ Open
Date
2006Author
Coe, Richard
Muchiri, Mary w.
Weke, Patrick G. O.
Type
ArticleLanguage
enMetadata
Show full item recordAbstract
This paper looks at responses from participatory on-farm farmer
participatory trials that are often measured as ratings (farmers score
each given treatment on a scale that is ordered but arbitrary) or
rankings (where farmers arrange the treatments in order from most
preferred to least preferred). Simple methods such as the preference
statistic that uses the proportion of responses, Kruskal-Wallis test
which is a one-way analysis of variance by ranks and the Friedman test
that is a two-way analysis of variance by ranks are outlined. The
Bradley-Terry model for ranks which is a logit model for paired
comparisons is described and used to fit models for plot level covariates.
Citation
Far East J. Thea. Stat. 20(2)(2006), 163-178Publisher
Pushpa publishing house Department of Mathematics University of Nairobi, Kenya