Bradley-Terry Models in<i>R</i>: The<b>BradleyTerry2</b>Package

Turner, Heather; Firth, David

doi:10.18637/jss.v048.i09

Cited by 195 publications

(156 citation statements)

References 22 publications

Supporting

Mentioning

149

Contrasting

Order By: Relevance

“…For this reason, the Bradley and Terry (1952) approach was used to analyse the ranking data of tree species and ES. Hence, the data was analysed in R (R Core Team 2015) using the BradleyTerry2 package (Turner and Firth 2012) and three other R functions as explained Van der Wolf et al (2016).…”

Section: Surveys and Data Collectionmentioning

confidence: 99%

Local tree knowledge can fast-track agroforestry recommendations for coffee smallholders along a climate gradient in Mount Elgon, Uganda

et al. 2017

View full text Add to dashboard Cite

Arabica coffee (Coffea arabica) is economically important for many smallholder farmers in the Mount Elgon region of East Uganda, but its production is increasingly threatened by climate change. However, ecosystem services (ES) provided by companion trees in coffee agroforestry systems (AFS) can help farmers adapt to climate change. The objectives of this research were to develop agroforestry species recommendations and tailor these to the farmers' needs and local context, taking into consideration gender. Local knowledge of agroforestry species and ES preferences was collected through farmer interviews and rankings. Using the Bradley-Terry approach, analysis was done along an altitudinal gradient in order to study different climate change scenarios for coffee suitability. Farmers had different needs in terms of ES and tree species at different altitudes, e.g. at low altitude they need a relatively larger set of ES to sustain their coffee production and livelihood. Local knowledge is found to be gender blind as no differences were observed in the rankings of species and ES by men and women. Ranking species by ES and ranking ES by preference is a useful method to help scientists and extension agents to use local knowledge for the development of recommendations on companion trees in AFS for smallholder farmers.

show abstract

Section: Surveys and Data Collectionmentioning

confidence: 99%

Local tree knowledge can fast-track agroforestry recommendations for coffee smallholders along a climate gradient in Mount Elgon, Uganda

et al. 2017

View full text Add to dashboard Cite

show abstract

“…We analyzed farmers' ranking of (i) ES and (ii) trees for providing ES at plot level; using Bradley and Terry (BT) model (Bradley and Terry 1952) in 'BradleyTerry2' package of R (Turner and Firth 2012;R Core Team 2015). The model based on how various farmers ranked trees for ES provision to generate a score upon which it ranks the trees.…”

Section: Farmers' Local Knowledge Of Coffee-shade Treesmentioning

confidence: 99%

Integrating local knowledge with tree diversity analyses to optimize on-farm tree species composition for ecosystem service delivery in coffee agroforestry systems of Uganda

et al. 2017

View full text Add to dashboard Cite

Coffee agroforestry systems deliver ecosystem services (ES) critical for rural livelihoods like food but also disservices that constrain livelihoods like fostering coffee-pests. Since such ES are treebased, maximizing ES and limiting constraints requires knowledge on optimizing on-farm tree composition especially trees adapted to local conditions. The study was in three sites along a rainfall gradient in Central Uganda where we: assessed tree diversity in coffee agroforestry; ranked tree suitability for providing ES according to farmers' knowledge; and then proposed an approach for optimizing on-farm tree composition for delivery of ES. We collected data on tree diversity and, farmers' knowledge of tree species and the ES they provide. Farmers ranked ES in order of importance to their livelihoods ('Needs rank') and ranked trees according to suitability for providing ES. Using Bradley Terry modeling, we grouped trees into 'ES groups' according to suitability for providing different ES and ranked 'ES groups' according to tree diversity ('Diversity rank'). Tree-suitability for providing ES and importance of ES to farmers varied with rainfall regime but tree diversity did not match farmers' needs for ES. We propose the FaD-FaN (matching farm tree diversity to farmers' needs) approach for optimizing tree species composition with respect to tree-suitability for farmers' priority ES. Farmers locally prioritize ES needed and identify trees that best serve such ES. The approach then focuses on modifying on-farm tree diversity to match/suit 123Agroforest Syst https://doi.org/10.1007/s10457-017-0172-8 farmers' priority ES. The FaD-FaN approach caters for varying socio-ecological conditions; it's adaptable for other coffee and cocoa-growing areas worldwide.

show abstract

“…Most of the available add-on packages for paired comparison models in R are restricted to the case of binary response and cannot deal with ordered response. The most popular packages are prefmod (Hatzinger and Dittrich, 2012) and BradleyTerry2 (Turner and Firth, 2012). The former uses the log linear representation of BT-models and can handle draws in the response variable.…”

Section: Discussionmentioning

confidence: 99%

“…An alternative way to constrain estimates is to use a random effects model instead of a fixed effect model. By assuming that the strengths are random effects, for example, by assuming γ r ∼ N (0, σ 2 ), parameters can be estimated within a random effects model, see Firth (2005) and Turner and Firth (2012) who used random effects models to account for correlations between responses. A disadvantage of random effects models is that they assume that random effects and covariates are uncorrelated, certainly not realistic in football if the covariates contain the budget of teams, because it might be the main source of the strength of a team.…”

Section: A Model With Team-specific Explanatory Variablesmentioning

confidence: 99%

Extended ordered paired comparison models with application to football data from German Bundesliga

Tutz

Schauberger

2014

AStA Adv Stat Anal

View full text Add to dashboard Cite

A general paired comparison model for the evaluation of sports competitions is proposed. It efficiently uses the available information by allowing for ordered response categories and team-specific home advantage effects. Penalized estimation techniques are used to identify clusters of teams that share the same ability. The model is extended to include team-specific explanatory variables. It is shown that regularization techniques allow to identify the contribution of explanatory variables to the success of teams. The usefulness of the methods is demonstrated by investigating the performance and its dependence on the budget for football teams of the German Bundesliga.

show abstract

Bradley-Terry Models inR: TheBradleyTerry2Package

Cited by 195 publications

References 22 publications

Local tree knowledge can fast-track agroforestry recommendations for coffee smallholders along a climate gradient in Mount Elgon, Uganda

Local tree knowledge can fast-track agroforestry recommendations for coffee smallholders along a climate gradient in Mount Elgon, Uganda

Integrating local knowledge with tree diversity analyses to optimize on-farm tree species composition for ecosystem service delivery in coffee agroforestry systems of Uganda

Extended ordered paired comparison models with application to football data from German Bundesliga

Contact Info

Product

Resources

About