Document Detail

Limitations to the robustness of binormal ROC curves: effects of model misspecification and location of decision thresholds on bias, precision, size and power.
MedLine Citation:
PMID:  9131755     Owner:  NLM     Status:  MEDLINE    
This paper concerns robustness of the binormal assumption for inferences that pertain to the area under an ROC curve. I applied the binormal model to rating method data sets sampled from bilogistic curves and observed small biases in area estimates. Bias increased as the range of decision thresholds decreased. The variance of area estimates also increased as the range of decision thresholds decreased. Together, minor bias and inflated variance substantially altered the size and power of statistical tests that compared areas under bilogistic ROC curves. I repeated the simulations by applying the binormal assumption to data sampled from binormal curves. Biases in area estimates were minimal for the binormal data, but the variance of area estimates was again higher when the range of decision thresholds was narrow. The size of tests that compared areas did not vary from the chosen significance level. Power fell, however, when the variance of area estimates was inflated. I conclude that inferences derived from the binormal assumption are sensitive to model misspecification and to the location of decision thresholds. A narrow span of decision thresholds increases the variability of area estimates and reduces the power of area comparisons. Model misspecification produces bias that alters test size and can exaggerate the loss of power that accompanies increased variability.
S J Walsh
Related Documents :
4058025 - Theorems on log-convex disposition curves in drug and tracer kinetics.
8345445 - Analyzing individual status and change with hierarchical linear models: illustration wi...
20805055 - On linear combinations of dichotomizers for maximizing the area under the roc curve.
23004965 - Search for supersymmetry in events with three leptons and missing transverse momentum i...
19276275 - Randomized phase ii designs.
10582035 - Error mode prediction.
Publication Detail:
Type:  Journal Article; Research Support, Non-U.S. Gov't    
Journal Detail:
Title:  Statistics in medicine     Volume:  16     ISSN:  0277-6715     ISO Abbreviation:  Stat Med     Publication Date:  1997 Mar 
Date Detail:
Created Date:  1997-06-26     Completed Date:  1997-06-26     Revised Date:  2006-11-15    
Medline Journal Info:
Nlm Unique ID:  8215016     Medline TA:  Stat Med     Country:  ENGLAND    
Other Details:
Languages:  eng     Pagination:  669-79     Citation Subset:  IM    
Department of Community Medicine, School of Medicine, University of Connecticut Health Center, Farmington 06030-1910, USA.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Bias (Epidemiology)*
Binomial Distribution*
Decision Support Techniques*
Logistic Models*
ROC Curve*
Reproducibility of Results
Sensitivity and Specificity

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine

Previous Document:  Group sequential monitoring of distribution-free analyses of repeated measures.
Next Document:  Blinded subjective rankings as a method of assessing treatment effect: a large sample example from t...