Detecting treatment-subgroup interactions in clustered data with generalized linear mixed-effects model trees.

Identification of subgroups of patients for whom treatment A is more effective than treatment B, and vice versa, is of key importance to the development of personalized medicine. Tree-based algorithms are helpful tools for the detection of such interactions, but none of the available algorithms allow for taking into account clustered or nested dataset structures, which are particularly common in psychological research. Therefore, we propose the generalized linear mixed-effects model tree (GLMM tree) algorithm, which allows for the detection of treatment-subgroup interactions, while accounting for the clustered structure of a dataset. The algorithm uses model-based recursive partitioning to detect treatment-subgroup interactions, and a GLMM to estimate the random-effects parameters. In a simulation study, GLMM trees show higher accuracy in recovering treatment-subgroup interactions, higher predictive accuracy, and lower type II error rates than linear-model-based recursive partitioning and mixed-effects regression trees. Also, GLMM trees show somewhat higher predictive accuracy than linear mixed-effects models with pre-specified interaction effects, on average. We illustrate the application of GLMM trees on an individual patient-level data meta-analysis on treatments for depression. We conclude that GLMM trees are a promising exploratory tool for the detection of treatment-subgroup interactions in clustered datasets.

Comparison of optotypes of Amsterdam Picture Chart with those of Tumbling-E, LEA symbols, ETDRS, and Landolt-C in non-amblyopic and amblyopic patients.

PURPOSE: To compare optotypes of the Amsterdam Picture Chart (APK) with those of Landolt-C (LC), Tumbling-E (TE), ETDRS and LEA symbols (LEA), to assess their reliability in measuring visual acuity (VA). METHODS: We recruited healthy controls with equal VA and amblyopes with ≥2 LogMAR lines interocular difference. New logarithmic charts were developed with LC, TE, ETDRS, LEA, and APK with identical size and spacing (four optotypes) between optotypes. Charts were randomly presented at 5 m under DIN EN ISO 8596 and 8597 conditions. VA was measured with LC (LC-VA), TE, ETDRS, LEA, and APK, using six out of ten optotypes answered correctly as threshold. RESULTS: In 100 controls aged 17-31, LC-VA was -0.207 ± SD 0.089 LogMAR. Visual acuity measured with TE differed from LC-VA by 0.021 (positive value meaning less recognizable), with ETDRS 0.012, with Lea 0.054, and with APK 0.117. In 46 amblyopic eyes with LC-VA <0.5 LogMAR, the difference was for TE 0.017, for ETDRS 0.017, for LEA 0.089, and for APK 0.213. In 13 amblyopic eyes with LC-VA ≥0.5 LogMAR, the difference was for TE 0.122, ETDRS 0.047, LEA 0.057, and APK 0.019. APK optotypes had a lower percentage of passed subjects at each LogMAR line compared to Landolt-C. The 11 APK optotypes had different thresholds. CONCLUSIONS: Small APK optotypes were recognized worse than all other optotypes, probably because of their thinner lines. Large APK optotypes were recognized relatively well, possibly reflecting recognition acuity. Differences between the thresholds of the 11 APK optotypes reduced its sensitivity further.

Investigating Group Differences on Cognitive Tests Using Spearman's Hypothesis: An Evaluation of Jensen's Method.

Jensen has posited a research method to investigate group differences in cognitive tests. This method consists of first extracting a general intelligence factor by means of exploratory factor analysis. Secondly, similarity of factor loadings across groups is evaluated in an attempt to ensure that the same constructs are measured. Finally, the correlation is computed between the loadings of the tests on the general intelligence factor and the mean differences between groups on the tests. This part is referred to as a test of "Spearman's Hypothesis", which essentially states that differences in g account for the main part of differences in observed scores. Based on the correlation, inferences are made with respect to group differences in general intelligence. The validity of these inferences is investigated and compared to the validity of inferences based on multi-group confirmatory factor analysis. For this comparison, population covariance matrices are constructed which incorporate violations of the central assumption underlying Jensen's method concerning the existence of g and/or violations of Spearman's Hypothesis. It is demonstrated that Jensen's method is quite insensitive to the violations. This lack of specificity is observed consistently for all types of violations introduced in the present study. Multi-group confirmatory factor analysis emerges as clearly superior to Jensen's method.

Pseudolikelihood estimation of the Rasch model.

An estimation method is proposed for the Rasch model on the basis of the pseudolikelihood theory of Arnold and Strauss (1988). A simulation study was conducted to compare the proposed maximum pseudolikelihood estimates with the well known conditional maximum likelihood and unconditional maximum likelihood estimates for the item parameters of the Rasch model. The results show great similarity between the methods.

Guilford's Facet Theory Of Intelligence: An Empirical Comparison Of Models.

A general linear model was described for tests constructed in a facet design. Guilford's structure of intellect model and an alternative model were specified as special cases. From the Guilford research seven data sets were selected such that both models could be compared. The comparison was done with oblique and orthogonal factors. Using covariance structure analysis the fit of the models was assessed, and the parameters estimated. Models with orthogonal factors did not fit the data. The fit of the oblique Guilford model was better than the alternative oblique model. For three of the data sets the oblique Guilford model yielded an acceptable fit to the data with parameter estimates that could be interpreted by the structure of intellect model.