Inferences from statistical model

The simple collision theory for bimolecular gas phase reactions is usually introduced to students in the early stages of their courses in chemical kinetics. They learn that the discrepancy between the rate constants calculated by use of this model and the experimentally determined values may be interpreted in terms of a steric factor, which is defined to be the ratio of the experimental to the calculated rate constants Despite its inherent limitations, the collision theory introduces the idea that molecular orientation (molecular shape) may play a role in chemical reactivity. We now have experimental evidence that molecular orientation plays a crucial role in many collision processes ranging from photoionization to thermal energy chemical reactions. Usually, processes involve a statistical distribution of orientations, and information about orientation requirements must be inferred from indirect experiments. Over the last 25 years, two methods have been developed for orienting molecules prior to collision (1) orientation by state selection in inhomogeneous electric fields, which will be discussed in this chapter, and (2) bmte force orientation of polar molecules in extremely strong electric fields. Several chemical reactions have been studied with one of the reagents oriented prior to collision. ... 

The scope of computational chemistry can be inferred from the methodologies it encompasses. Some of the more common tools include computer graphics, molecular modeling, quantum chemistry, molecular mechanics (MM), statistical analysis of structure-property relationships, and data management (informatics). As with any dynamic field of research, computational... 

Bergman and Hynen (1997) developed a method similar to that of Box and Meyer (1986), but with a simple and exact distribution theory for inference from the test statistic. The important observation of Bergman and Hynen was that the residuals from the fitted location model could complicate inference for the Box-Meyer statistic in two ways. First, the residuals in the two sums of squares could be correlated. Second, the residuals at the high (low) level of factor j typically depend on the actual variances at both levels of the factor, not just the level at which the run was made. 

Fig. 2. Method overview Ligand sets derived from existing databases (a) are used in set-wise comparisons (b) against a query set, the result of which is quantified by the statistical model inferred from that reference database (c).The generated probabilistic data can be used to construct chemical mappings of the ligand sets and correspondingly the biological targets (d).

Bayesian statistics Bayesian inference is a variant of statistics where prior information is allowed to influence the posterior probability of an event via application of Bayes rule. Complex problems of cheminformatics and bioinformatics often benefit from Bayesian models. A schism divides statisticians from Bayesians. 

This is not to say that I think that only randomization-based inference is valid for clinical trials far from it. Model-based inference is extremely important. What is also important, however, is studying effects and estimating contrasts, whatever the statistical machinery used to do it. Thus my general advice for clinical trials is (1) remember that they are experiments and (2) think comparatively. 

Random-effect model. A term which is used in at least two rather different senses by statisticians in the context of drug development. (1) A model for which more than one term is assumed random but the treatment effect is assumed fixed. (All statistical models, including so-called fixed ones have at least one error term which is random.) (2) A model in which the treatment effect itself is assumed to vary randomly from unit to unit. For balanced designs, random-effect models of the first sort can lead to identical inferences to fixed-effect models. Even for balanced designs, random-effect models of the second sort will not. 

Finally we remark that the majority of the parameters in the STCA filters have direct physical or mechanical interpretation, and that the transparency of the classification process is an important component in assuring the safety case for STCA. However, whether tuned by hand or optimised by a machine algorithm, the operational parameters are inferred from data. An alternative to direct physical modelling is to employ purely statistical classifiers, for example Ic-nearest neighbour classifiers or neural networks, for which there is no ready interpretation of the parameters. Nonetheless, these methods are highly effective in other areas and the machine optimisation of STCA parameters blurs the distinction between physical models on one hand and statistical black boxes on the other. We look forward to the construction of safety cases for purely statistical classifiers whose operational parameters are inferred from data and which have no ready physical interpretation. 

---