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.

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.

The Blinder-Oaxaca decomposition for linear regression models

The counterfactual decomposition technique popularized by Blinder (1973, Journal of Human Resources, 436–455) and Oaxaca (1973, International Economic Review, 693–709) is widely used to study mean outcome differences between groups. For example, the technique is often used to analyze wage gaps by sex or race. This article summarizes the technique and addresses several complications, such as the identification of effects of categorical predictors in the detailed decomposition or the estimation of standard errors. A new command called oaxaca is introduced, and examples illustrating its usage are given.

Detecting outlying studies in meta-regression models using a forward search algorithm.

When considering data from many trials, it is likely that some of them present a markedly different intervention effect or exert an undue influence on the summary results. We develop a forward search algorithm for identifying outlying and influential studies in meta-analysis models. The forward search algorithm starts by fitting the hypothesized model to a small subset of likely outlier-free studies and proceeds by adding studies into the set one-by-one that are determined to be closest to the fitted model of the existing set. As each study is added to the set, plots of estimated parameters and measures of fit are monitored to identify outliers by sharp changes in the forward plots. We apply the proposed outlier detection method to two real data sets; a meta-analysis of 26 studies that examines the effect of writing-to-learn interventions on academic achievement adjusting for three possible effect modifiers, and a meta-analysis of 70 studies that compares a fluoride toothpaste treatment to placebo for preventing dental caries in children. A simple simulated example is used to illustrate the steps of the proposed methodology, and a small-scale simulation study is conducted to evaluate the performance of the proposed method. Copyright © 2016 John Wiley & Sons, Ltd.

Development and validation of energy expenditure prediction models based on GT3X accelerometer data in 5- to 9-year-old children

Background: Accelerometry has been established as an objective method that can be used to assess physical activity behavior in large groups. The purpose of the current study was to provide a validated equation to translate accelerometer counts of the triaxial GT3X into energy expenditure in young children. Methods: Thirty-two children aged 5–9 years performed locomotor and play activities that are typical for their age group. Children wore a GT3X accelerometer and their energy expenditure was measured with indirect calorimetry. Twenty-one children were randomly selected to serve as development group. A cubic 2-regression model involving separate equations for locomotor and play activities was developed on the basis of model fit. It was then validated using data of the remaining children and compared with a linear 2-regression model and a linear 1-regression model. Results: All 3 regression models produced strong correlations between predicted and measured MET values. Agreement was acceptable for the cubic model and good for both linear regression approaches. Conclusions: The current linear 1-regression model provides valid estimates of energy expenditure for ActiGraph GT3X data for 5- to 9-year-old children and shows equal or better predictive validity than a cubic or a linear 2-regression model.

ROBREG: Stata module providing robust regression estimators

robreg provides a number of robust estimators for linear regression models. Among them are the high breakdown-point and high efficiency MM-estimator, the Huber and bisquare M-estimator, and the S-estimator, each supporting classic or robust standard errors. Furthermore, basic versions of the LMS/LQS (least median of squares) and LTS (least trimmed squares) estimators are provided. Note that the moremata package, also available from SSC, is required.

Modeling absolute differences in life expectancy with a censored skew-normal regression approach.

Parameter estimates from commonly used multivariable parametric survival regression models do not directly quantify differences in years of life expectancy. Gaussian linear regression models give results in terms of absolute mean differences, but are not appropriate in modeling life expectancy, because in many situations time to death has a negative skewed distribution. A regression approach using a skew-normal distribution would be an alternative to parametric survival models in the modeling of life expectancy, because parameter estimates can be interpreted in terms of survival time differences while allowing for skewness of the distribution. In this paper we show how to use the skew-normal regression so that censored and left-truncated observations are accounted for. With this we model differences in life expectancy using data from the Swiss National Cohort Study and from official life expectancy estimates and compare the results with those derived from commonly used survival regression models. We conclude that a censored skew-normal survival regression approach for left-truncated observations can be used to model differences in life expectancy across covariates of interest.

Blood glucose and prognosis in children with presumed severe malaria: is there a threshold for 'hypoglycaemia'?

OBJECTIVES: Hypoglycaemia (glucose <2.2 mmol/l) is a defining feature of severe malaria, but the significance of other levels of blood glucose has not previously been studied in children with severe malaria. METHODS: A prospective study of 437 consecutive children with presumed severe malaria was conducted in Mali. We defined hypoglycaemia as <2.2 mmol/l, low glycaemia as 2.2-4.4 mmol/l and hyperglycaemia as >8.3 mmol/l. Associations between glycaemia and case fatality were analysed for 418 children using logistic regression models and a receiver operator curve (ROC). RESULTS: There was a significant difference between blood glucose levels in children who died (median 4.6 mmol/l) and survivors (median 7.6 mmol/l, P < 0.001). Case fatality declined from 61.5% of the hypoglycaemic children to 46.2% of those with low glycaemia, 13.4% of those with normal glycaemia and 7.6% of those with hyperglycaemia (P < 0.001). Logistic regression showed an adjusted odds ratio (AOR) of 0.75 (0.64-0.88) for case fatality per 1 mmol/l increase in baseline blood glucose. Compared to a normal blood glucose, hypoglycaemia and low glycaemia both significantly increased the odds of death (AOR 11.87, 2.10-67.00; and 5.21, 1.86-14.63, respectively), whereas hyperglycaemia reduced the odds of death (AOR 0.34, 0.13-0.91). The ROC [area under the curve at 0.753 (95% CI 0.684-0.820)] indicated that glycaemia had a moderate predictive value for death and identified an optimal threshold at glycaemia <6.1 mmol/l, (sensitivity 64.5% and specificity 75.1%). CONCLUSIONS: If there is a threshold of blood glucose which defines a worse prognosis, it is at a higher level than the current definition of 2.2 mmol/l.