Overdispersion tests in count-data analysis

Count data are commonly assumed to have a Poisson distribution, especially when there is no diagnostic procedure for checking this assumption. However, count data rarely fit the restrictive assumptions of the Poisson distribution. The violation of much of such assumptions commonly results in overdispersion, which invalidates the Poisson distribution. Undetected overdispersion may entail important misleading inferences, so its detection is essential. In this study, different overdispersion diagnostic tests are evaluated through two simulation studies. In Exp. 1, the nominal error rate is compared under different sample sizes and lamda conditions. Analysis shows a remarkable performance of the chi2 df test. In Exp. 2 and 3, statistical power is compared under different sample sizes, lamda, and overdispersion conditions. Chi2 and LR tests provide the highest statistical power.     

Q-Coh: A tool to screen the methodological quality of cohort studies in systematic reviews and meta-analyses

The evaluation of the methodological quality of primary studies in a systematic review is a key process to enhance the likelihood of achieving valid results. When considering non-randomized designs as cohort studies, this process becomes even more critical, since these designs are more susceptible to bias than randomized controlled trials are. Taking this into account, a tool, named Q-Coh, was designed with the aim to screen the methodological quality of the primary studies with a cohort design priming specificity over sensitivity in a reasonable application time. After applying it to 21 prospective cohort studies by three raters, all domains had a moderate to good agreement, with all except one of them having statistically significant kappa values. Despite there is no gold standard for the methodological quality, arguments supporting its validity are given. Future research should assess the psychometric properties of Q-Coh in the context of real meta-analyses, evaluate the influence of the raters’ substantive and methodological expertise on these properties, and explore different ways of including the domains-based ratings of the quality provided by Q-Coh into meta-analyses.     

Using the linear mixed model to analyze nonnormal data distributions in longitudinal designs

Using a Monte Carlo simulation and the Kenward–Roger (KR) correction for degrees of freedom, in this article we analyzed the application of the linear mixed model (LMM) to a mixed repeated measures design. The LMM was first used to select the covariance structure with three types of data distribution: normal, exponential, and log-normal. This showed that, with homogeneous between-groups covariance and when the distribution was normal, the covariance structure with the best fit was the unstructured population matrix. However, with heterogeneous between-groups covariance and when the pairing between covariance matrices and group sizes was null, the best fit was shown by the between-subjects heterogeneous unstructured population matrix, which was the case for all of the distributions analyzed. By contrast, with positive or negative pairings, the within-subjects and between-subjects heterogeneous first-order autoregressive structure produced the best fit. In the second stage of the study, the robustness of the LMM was tested. This showed that the KR method provided adequate control of Type I error rates for the time effect with normally distributed data. However, as skewness increased—as occurs, for example, in the log-normal distribution—the robustness of KR was null, especially when the assumption of sphericity was violated. As regards the influence of kurtosis, the analysis showed that the degree of robustness increased in line with the amount of kurtosis.     

Generalized Communicative Suspicion (GCS) Among Police Officers: Accounting for the Investigator Bias Effect1

Research shows that police officers are no more accurate than non‐officers in discerning between truthful and deceptive statements, but more biased than non‐officers to judge statements as deceptive. This was labeled by Meissner and Kassin (2002) as the investigator bias effect. A likely explanation could be that, since officers are normally involved in situations where questioning the truthfulness of the interviewee is essential, they could develop a generalized communicative suspicion (GCS). Levine and McCornack's (1991) GCS scale was adapted to the Spanish context and used to compare the GCS ratings of 152 undergraduates, 88 experienced officers, and 89 police recruits. Experienced officers’ GCS scores were significantly higher than those of the other groups, which did not differ from each other. This suggests that socialization within the police force increases the officers’ suspicion, leading them to make deceptiveness judgments.     

Median decision techniques as an alternative to the classical statistical inference test

In the research area, the criterion to test a hypothesis is frequently the distance of each subject from the mean of the distribution. This concept of distance is constantly used and re-used as a source of evidence in the search for significant contrasts with respect to models of theoretical probability. This paper proposes a number of ideas regarding the use of new distance indicators related to statistics that rely less heavily on parametric assumptions, in particular, the use of median confidence intervals. We have simulated via EXCEL different data samples for a design of two independent groups under 15 different conditions in two different situations: mean equality and mean differences. The results showed the high sensitivity of the comparison of the confidence intervals, especially if a strict decision criterion is used. However, we did not obtain good results for sensitivity. In fact, in some cases, the comparison of the confidence intervals of the mean worked better than those of median.     

Exercise motivation in university community members: a behavioural intervention

The purpose of this study was to examine how exercise level, exercise motives, and barriers changed from the baseline phase to the follow-up phase after a behavioural and cognitive intervention aimed at increasing exercise. Seventy-five members of our university community (43 subjects in the control group and 32 in the experimental group), all of whom received cognitive feedback, agreed to complete the baseline phase. Only the experimental group received behavioural feedback and a free-access gym ticket with personal training in order to facilitate their adherence to exercise. The results suggest that a combination of behavioural and psychological techniques is an efficient strategy for increasing exercise level. In addition, the results showed that extrinsic motivation predominates the early stages of change-of-exercise behaviour, and that intrinsic motivation is important for progression towards maintenance. Subjects who decreased their exercise level increased their extrinsic exercise motivation and subjects who increased their exercise level decreased the barriers related to intrinsic motivation. These findings indicate that, in order to facilitate exercise adherence, feedback about motives for undertaking exercise is needed, combined with advice about how to improve physical condition. This combination could help eliminate certain barriers that hinder engaging in an active and healthy life-style.     

Evaluating effects in short time series: alternative models of analysis

In the applied context, short time-series designs are suitable to evaluate a treatment effect. These designs present serious problems given autocorrelation among data and the small number of observations involved. This paper describes analytic procedures that have been applied to data from short time series, and an alternative which is a new version of the generalized least squares method to simplify estimation of the error covariance matrix. Using the results of a simulation study and assuming a stationary first-order autoregressive model, it is proposed that the original observations and the design matrix be transformed by means of the square root or Cholesky factor of the inverse of the covariance matrix. This provides a solution to the problem of estimating the parameters of the error covariance matrix. Finally, the results of the simulation study obtained using the proposed generalized least squares method are compared with those obtained by the ordinary least squares approach. The probability of Type I error associated with the proposed method is close to the nominal value for all values of rho1 and n investigated, especially for positive values of rho1. The proposed generalized least squares method corrects the effect of autocorrelation on the test's power.     

Autocorrelation problems in short time series

The work of Huitema (1985) on autocorrelation in behavioral data suggests that the use of conventional statistical methods is justified. The present study restates the problem of autocorrelation by analyzing 100 baselines of small samples designs published in the Journal of Applied Behavior Analysis during 1992. The results show a negative bias in the autocorrelations, especially with very small samples. The autocorrelation values are normally distributed, and the method of Davies, Trigg, and Newbold (1977) is the most accurate in calculating the standard deviation.