On Graphically Checking Goodness-of-fit of Binary Logistic Regression Models

OBJECTIVES: This paper is concerned with checking goodness-of-fit of binary logistic regression models. For the practitioners of data analysis, the broad classes of procedures for checking goodness-of-fit available in the literature are described. The challenges of model checking in the context of binary logistic regression are reviewed. As a viable solution, a simple graphical procedure for checking goodness-of-fit is proposed. METHODS: The graphical procedure proposed relies on pieces of information available from any logistic analysis; the focus is on combining and presenting these in an informative way. RESULTS: The information gained using this approach is presented with three examples. In the discussion, the proposed method is put into context and compared with other graphical procedures for checking goodness-of-fit of binary logistic models available in the literature. CONCLUSION: A simple graphical method can significantly improve the understanding of any logistic regression analysis and help to prevent faulty conclusions.

Survival of bonded lingual retainers with chemical or photo polymerization over a 2-year period: a single-center, randomized controlled clinical trial

INTRODUCTION The objective of this trial was to compare the survival rates of mandibular lingual retainers bonded with either chemically cured or light-cured adhesive after orthodontic treatment. METHODS Patients having undergone orthodontic treatment at a private orthodontic office were randomly allocated to fixed retainers placed with chemically cured composite or light-cured composite. Eligibility criteria included no active caries, restorations, or fractures on the mandibular anterior teeth, and adequate oral hygiene. The main outcome was any type of first-time lingual retainer breakage; pattern of failure (adapted adhesive remnant index scores) was a secondary outcome. Randomization was accomplished with random permuted blocks of 20 patients with allocation concealed in sequentially numbered, opaque, sealed envelopes. Blinding was applicable for outcome assessment only. Patients were reviewed at 1, 3, and 6 months and then every 6 months after placement of the retainer until completion of the study. Data were analyzed using survival analysis including Cox regression; sensitivity analysis was carried out after data imputation for subjects lost to follow-up. RESULTS Two hundred twenty patients (median age, 16 years; interquartile range, 2; range, 12-47 years) were randomized in a 1:1 ratio to either chemical or light curing. Baseline characteristics were similar between groups, the median follow-up period was 2.19 years (range, 0.003-3.64 years), and 16 patients were lost to follow-up. At a minimum follow-up of 2 years, 47 of 110 (42.7%) and 55 of 110 (50.0%) retainers had some type of failure with chemically cured and light-cured adhesive, respectively (log-rank test, P = 0.35). Data were analyzed on an intention-to-treat basis, and the hazard ratio (HR) was 1.15 (95% confidence interval [CI], 0.88-1.70; P = 0.47). There was weak evidence that age is a significant predictor for lingual retainer failures (HR, 0.96; 95% CI, 0.93-1.00; P = 0.08). Adhesive remnant index scoring was possible for only 66 of the 102 (64.7%) failures and did not differ between composites (Fisher exact test, P = 0.16). No serious harm was observed other than gingivitis associated with plaque accumulation. CONCLUSIONS The results of this study indicated no evidence that survival of mandibular lingual retainers differs between chemically and light-cured adhesives. The overall failure rate was 46.4%; however, this included any type of failure, which may have exaggerated the overall failure rate.

Stochastic search for semiparametric linear regression models

This paper introduces and analyzes a stochastic search method for parameter estimation in linear regression models in the spirit of Beran and Millar [Ann. Statist. 15(3) (1987) 1131–1154]. The idea is to generate a random finite subset of a parameter space which will automatically contain points which are very close to an unknown true parameter. The motivation for this procedure comes from recent work of Dümbgen et al. [Ann. Statist. 39(2) (2011) 702–730] on regression models with log-concave error distributions.