Edgar Erdfelder

G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences

G*Power (Erdfelder, Faul, & Buchner, 1996) was designed as a general stand-alone power analysis program for statistical tests commonly used in social and behavioral research. G*Power 3 is a major extension of, and improvement over, the previous versions. It runs on widely used computer platforms (i.e., Windows XP, Windows Vista, and Mac OS X 10.4) and covers many different statistical tests of thet, F, and χ2 test families. In addition, it includes power analyses forz tests and some exact tests. G*Power 3 provides improved effect size calculators and graphic options, supports both distribution-based and design-based input modes, and offers all types of power analyses in which users might be interested. Like its predecessors, G*Power 3 is free.

Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses

G*Power is a free power analysis program for a variety of statistical tests. We present extensions and improvements of the version introduced by Faul, Erdfelder, Lang, and Buchner (2007) in the domain of correlation and regression analyses. In the new version, we have added procedures to analyze the power of tests based on (1) single-sample tetrachoric correlations, (2) comparisons of dependent correlations, (3) bivariate linear regression, (4) multiple linear regression based on the random predictor model, (5) logistic regression, and (6) Poisson regression. We describe these new features and provide a brief introduction to their scope and handling.

GPOWER : A GENERAL POWER ANALYSIS PROGRAM

GPOWER is a completely interactive, menu-driven program for IBM-compatible and Apple Macintosh personal computers. It performs high-precision statistical power analyses for the most common statistical tests in behavioral research, that is,t tests,F tests, andχ2 tests. GPOWER computes (1) power values for given sample sizes, effect sizes andα levels (post hoc power analyses); (2) sample sizes for given effect sizes,α levels, and power values (a priori power analyses); and (3)α andβ values for given sample sizes, effect sizes, andβ/α ratios (compromise power analyses). The program may be used to display graphically the relation between any two of the relevant variables, and it offers the opportunity to compute the effect size measures from basic parameters defining the alternative hypothesis. This article delineates reasons for the development of GPOWER and describes the program’s capabilities and handling.

Toward unbiased measurement of conscious and unconscious memory processes within the process dissociation framework.

L.L. Jacobys (1991) process dissociation framework has been welcomed as a tool for differentiating controlled and automatic cognitive processes. Several variants of the original process dissociation measurement model are integrated in this article, and it is shown that the model ignores guessing and, hence, response bias. An extension of the original model is suggested that includes guessing parameters. The original model and the extended model are evaluated empirically. In 3 experiments using a yes-no recognition task, response bias was manipulated in various ways. The original model falsely attributes effects of response biases to either controlled or uncontrolled processes or to both. The extended model, in contrast, results in estimates of the contributions of controlled and uncontrolled memory processes that are relatively unaffected by response biases. The extended model is recommended as a measurement tool.

Multinomial processing tree models: A review of the literature.

Multinomial processing tree (MPT) models have become popular in cognitive psychology in the past two decades. In contrast to general-purpose data analysis techniques, such as log-linear models or other generalized linear models, MPT models are substantively motivated stochastic models for categorical data. They are best described as tools (a) for measuring the cognitive processes that underlie human behavior in various tasks and (b) for testing the psychological assumptions on which these models are based. The present article provides a review of MPT models and their applications in psychology, focusing on recent trends and developments in the past 10 years. Our review is nontechnical in nature and primarily aims at informing readers about the scope and utility of MPT models in different branches of cognitive psychology. In a now classical article, Riefer and Batchelder (1988) proposed a class of substantively motivated stochastic mod- els for categorical behavioral data which was relatively well known in statistical genetics at the time (e.g., Elandt- Johnson, 1971), but had received little attention in psycho- logical research up to the 1980s. These models are now known as multinomial processing tree (MPT) models. About 10 years later, Batchelder and Riefer (1999) already identified no less than 30 published MPT models in the psychological literature, most of which were applied to different agendas in cognitive research. The present article provides an update of Batchelder and Riefers review and focuses on models and their applications published in the past 10 years. Our review includes 70 MPT models and model variants from more than 20 research areas. In the first section, we will present a brief conceptual outline of MPT models using a simple example to illustrate the basics and main advantages of this approach. Technical details will be omitted almost entirely because they have been described elsewhere (e.g., Batchelder & Riefer, 1999; Hu & Batchelder, 1994). The second section sum- marizes MPT models and their applications in different branches of cognitive psychology, with a special focus on models for various memory paradigms. In the third sec- tion, psychological applications of MPT models outside the realm of cognitive psychology will be briefly summarized. The fourth section describes recent developments, general- izations, and innovations in the statistical methodology of MPT models that might be useful for those interested in applying such models. The fifth and final section of our review provides a sketch of computer programs that are currently available for statistical analyses in the MPT framework, along with a summary of the main advantages of each program.

Decomposing the hindsight bias : A multinomial processing tree model for separating recollection and reconstruction in hindsight

After having received feedback about the correct answer to a question, a memory judgment about ones own past answer, the original judgment (OJ), is often biased toward the feedback. The authors present a multinomial model that explains this hindsight bias effect in terms of both memory impairments and reconstruction biases for nonrecollected OJs. The model was tested in 4 experiments. As predicted, the parameters measuring OJ recollection could be influenced selectively by contrasting items whose OJs were or were not retrieved successfully earlier (Experiment 1). Increasing the feedback-recall delay reduced reconstruction biases exclusively (Experiment 2), whereas discrediting the feedback enhanced recollection of the OJs to feedback items (Experiment 3). hi Experiment 4, the models guessing parameters, but no other parameters, varied as a function of the number of response alternatives. The authors discuss implications for hindsight bias theories.

On the Irrelevance of Semantic Information for the llIrrelevant Speech Effect

Two experiments that tested whether semantic similarity between visually presented targets and auditorily presented distractors has an effect on serial recall of the visual targets are reported. In Experiment 1, we found no difference in the recall of two-digit numbers when distractors were either numbers or words and non-words that were designed to be phonologically similar to the targets. In Experiment 2 the “semantic distance” between targets and distractors had no effect on serial recall. Taken together, these experiments conceptually replicate and extend earlier results, and they establish constraints for models of the effect of unattended acoustic information on serial recall.

One-reason decision-making unveiled: A measurement model of the recognition heuristic

The fast-and-frugal recognition heuristic (RH) theory provides a precise process description of comparative judgments. It claims that, in suitable domains, judgments between pairs of objects are based on recognition alone, whereas further knowledge is ignored. However, due to the confound between recognition and further knowledge, previous research lacked an unbiased measure of RH use. Also, model comparisons have not been based on goodness-of-fit and model complexity as criteria. To overcome both limitations we introduce and test a multinomial processing tree model showing that it fits empirical data and provides an unbiased measure of RH use. Analyses of 8 data sets reveal that the RH alone cannot account for the data, not even when it is implemented in a probabilistic way. That is, information integration beyond recognition plays a vital role and cannot merely account for empirical data better due to model flexibility. Also, we present several validations of the central model parameter and provide demonstrations of how the model can be applied to study the less-is-more effect as well as determinants of (and individual differences in) RH use. (PsycINFO Database Record (c) 2009 APA, all rights reserved).

A Multinomial Model to Assess Fluency and Recollection in a Sequence Learning Task

We suggest that well-formedness judgements in conjunction with L.L. Jacobys (1991) process dissociation procedure and an appropriate measurement model can be used to obtain measures of implicit and explicit sequence knowledge. We introduce a new measurement model designed specifically for the sequence learning task. The model assumes that sequence identification is based on recollection, perceptual or motor fluency, systematicity detection, and guessing. The model and the application of the process dissociation procedure were empirically evaluated using auditory event sequences. In Experiment 1, the parameter reflecting recollection was higher in an intentional than in an incidental learning condition. Experiment 2 showed that random sequences interspersed among the systematic sequences during the acquisition phase may change this pattern of results. A manipulation of processing fluency in Experiment 3 was reflected in the appropriate model parameter. In sum, the new measurement model and the application of the process dissociation procedure appear to be useful tools in sequence learning research.

Determ inants of Positive and Negative Generation Effects in Free Recall

Better retention of self-produced as opposed to experimenter-presented material is called generation effect; the reverse phenomenon is the negative generation effect. Both are found in intentional-learning experiments in which generating versus reading is manipulated between subjects. The present article presents an overview of those findings and aims at clarifying the conditions under which these effects emerge. Experiments 1 and 2 demonstrate that if cuetarget relations are manipulated within one list, a negative generation effect in free recall can be obtained for all items, no matter which cue-target relation they bear. In Experiment 3, cue-target relations were manipulated between lists. Here, a negative generation effect in free recall was found only in lists in which items were cued with words that mismatched the intertarget relations, whereas a positive generation effect was observed in those lists in which the generation cues matched the inter-target relations. A subsequent cued-recall test demonstrated that in cases of mismatch of relations, participants in the generate condition process cue-target relations at the expense of inter-target relations. The three-factor theory can be integrated with the task-demand account in a transfer-appropriate processing framework to accommodate these findings.