A quantitative literature review of ecological statistical methods accounting for imperfect detection. We read 500+ ecology papers from 40+ years (collected using a stratified sampling method) and assessed which papers accounted for imperfect detection, if necessary. Probability that a paper accounted for imperfect detection was modeled as a function of several covariates including year, journal type, and the taxa studied. We used a hierarchical logistic regression model fit in a Bayesian framework.
The results are published in the following article:
The analysis_id_alltaxa.R file contain the framework of the Bayesian analysis in JAGS for the project. Additional information is provided in the comments of the file.
The analysis_id_septaxa.R file contain the framework of the Bayesian analysis in JAGS when each taxa was analyzed separately. Additional information is provided in the comments of the file.
The analysis_detection_covs.R file contains some simple calculations and characterizations of how reported detection probabilities varied in the sampled papers.
The models folder contains the BUGS models used; models/model_id_alltaxa.R for the full analysis and models/model_id_septaxa.R for the analysis separated by taxonomic group.
The figures folder contains code for Figure 1 in the paper as well as figures used in an oral presentation at the Midwest Fish & Wildlife Conference in 2015.
The data folder contains the raw data files (CSV format) used in the analyses as well as a script used to generate a basic version of the paper database.