Optimal Correction for Guessing in Multiple-Choice Tests

Building on Item Response Theory we introduce students’ optimal behavior in multiple-choice tests. Our simulations indicate that the optimal penalty is relatively high, because although correction for guessing discriminates against risk-averse subjects, this effect is small compared with the measurement error that the penalty prevents. This result obtains when knowledge is binary or partial, under different normalizations of the score, when risk aversion is related to knowledge and when there is a pass-fail break point. We also find that the mean degree of difficulty should be close to the mean level of knowledge and that the variance of difficulty should be high.

On the strategic equivalence of multiple-choice test scoring rules

A disadvantage of multiple-choice tests is that students have incentives to guess. To discourage guessing, it is common to use scoring rules that either penalize wrong answers or reward omissions. These scoring rules are considered equivalent in psychometrics, although experimental evidence has not always been consistent with this claim. We model students' decisions and show, first, that equivalence holds only under risk neutrality and, second, that the two rules can be modified so that they become equivalent even under risk aversion. This paper presents the results of a field experiment in which we analyze the decisions of subjects taking multiple-choice exams. The evidence suggests that differences between scoring rules are due to risk aversion as theory predicts. We also find that the number of omitted items depends on the scoring rule, knowledge, gender and other covariates.

An investigation of imputation methods for discrete databases and multi-variate time series

When it comes to information sets in real life, often pieces of the whole set may not be available. This problem can find its origin in various reasons, describing therefore different patterns. In the literature, this problem is known as Missing Data. This issue can be fixed in various ways, from not taking into consideration incomplete observations, to guessing what those values originally were, or just ignoring the fact that some values are missing. The methods used to estimate missing data are called Imputation Methods. The work presented in this thesis has two main goals. The first one is to determine whether any kind of interactions exists between Missing Data, Imputation Methods and Supervised Classification algorithms, when they are applied together. For this first problem we consider a scenario in which the databases used are discrete, understanding discrete as that it is assumed that there is no relation between observations. These datasets underwent processes involving different combina- tions of the three components mentioned. The outcome showed that the missing data pattern strongly influences the outcome produced by a classifier. Also, in some of the cases, the complex imputation techniques investigated in the thesis were able to obtain better results than simple ones. The second goal of this work is to propose a new imputation strategy, but this time we constrain the specifications of the previous problem to a special kind of datasets, the multivariate Time Series. We designed new imputation techniques for this particular domain, and combined them with some of the contrasted strategies tested in the pre- vious chapter of this thesis. The time series also were subjected to processes involving missing data and imputation to finally propose an overall better imputation method. In the final chapter of this work, a real-world example is presented, describing a wa- ter quality prediction problem. The databases that characterized this problem had their own original latent values, which provides a real-world benchmark to test the algorithms developed in this thesis.