A Test by Any Other Name: P Values,Bayes Factors,and Statistical Inference

Procedures used for statistical inference are receiving increased scrutiny as the scientific community studies the factors associated with insuring reproducible research. This note addresses recent negative attention directed at p values, the relationship of confidence intervals and tests, and the role of Bayesian inference and Bayes factors, with an eye toward better understanding these different strategies for statistical inference. We argue that researchers and data analysts too often resort to binary decisions (e.g., whether to reject or accept the null hypothesis) in settings where this may not be required.     

Verbalization of Decision Strategies in Multiple‐Cue Probabilistic Inference

In multiple‐cue probabilistic inference, people choose between alternatives based on several cues, each of which is differentially associated with an alternative's overall value. Various strategies have been proposed for probabilistic inference (e.g., weighted additive, tally, and take‐the‐best). These strategies differ in how many cue values they require to enact and in how they weight each cue. Do decision makers actually use any of these strategies? Ways to investigate this question include analyzing people's choices and the cues that they reveal. However, different strategies often predict the same decisions, and search behavior says nothing about whether or how people use the information that they acquire. In this research, we attempt to elucidate which strategies participants use in a multiple‐cue probabilistic inference task by examining verbal protocols, a high‐density source of process data. The promise of verbal data is in their utility for testing detailed information processing models. To that end, we apply protocol analysis in conjunction with computational simulations. We find converging evidence across outcome measures, search measures, and verbal reports that most participants use simplifying heuristics, namely take‐the‐best. Copyright © 2015 John Wiley & Sons, Ltd.     

A Regression Framework for Causal Mediation Analysis with Applications to Behavioral Science

We introduce and extend the classical regression framework for conducting mediation analysis from the fit of only one model. Using the essential mediation components (EMCs) allows us to estimate causal mediation effects and their analytical variance. This single-equation approach reduces computation time and permits the use of a rich suite of regression tools that are not easily implemented on a system of three equations. Additionally, we extend this framework to non-nested mediation systems, provide a joint measure of mediation for complex mediation hypotheses, propose new visualizations for mediation effects, and explain why estimates of the total effect may differ depending on the approach used. Using data from social science studies, we also provide extensive illustrations of the usefulness of this framework and its advantages over traditional approaches to mediation analysis. The example data are freely available for download online and we include the R code necessary to reproduce our results.     

相对熟悉度和同音线索在谐音型歇后语理解中的作用

谐音型歇后语是汉语特殊的语汇表达形式, 对其加工常常需要通达后一语节的同音线索来完成语音、语义隐喻映射。本研究通过2个实验探讨相对熟悉度和同音线索类型对谐音型歇后语加工的作用。实验1结果表明, 在高熟悉度条件下, 被试对语汇的加工策略取决于歇后语的同音线索类型。当后一语节为同音字时, 加工较为快速; 当后一语节为谐音字时, 通达歇后语隐喻意义的路径受阻。对熟悉度较高的谐音型歇后语进行加工, 语义通达表现出预存性; 在缺乏同音线索的条件下, 语料加工时间较长, 反应准确率较低, 支持概念隐喻模型和概念合成假说。而对熟悉度较低的歇后语进行加工, 被试更倾向于采用即时(on-line)策略, 支持概念合成理论和基于心理模型的语用推理假说。实验2结果表明, 呈现“错误同音”线索对歇后语的加工形成干扰, 不支持概念合成理论和基于心理模型的语用推理理论。总的结果表明, 相对熟悉度和同音线索类型影响谐音型歇后语的通达, 谐音歇后语加工需要同时激活语音和语义两条通道。     

Forward inference in risky choice: Mapping gaze and decision processes

The study of cognitive processes is built on a close mapping between three components: overt gaze behavior, overt choice, and covert processes. To validate this overt–covert mapping in the domain of decision‐making, we collected eye‐movement data during decisions between risky gamble problems. Applying a forward inference paradigm, participants were instructed to use specific decision strategies to solve those gamble problems (maximizing expected values or applying different choice heuristics) during which gaze behavior was recorded. We revealed differences between overt behavior, as indicated by eye movements, and covert decision processes, instructed by the experimenter. However, our results show that the overt–covert mapping is for some eye‐movement measures not as close as expected by current decision theory, and hence question reverse inference as being prone to fallacies due to a violation of its prerequisite, that is, a close overt–covert mapping. We propose a framework to rehabilitate reverse inference.     

Creating Brand Meaning: A Review and Research Agenda

In this paper I review, from the perspective of experimental research, studies that have examined how brands acquire cultural meaning, and suggest future research directions. McCracken's (Journal of Consumer Research, 13 , 1986 and 71) model of the meaning transfer process gained influence about thirty years ago, but experimental studies of the processes it posited have been limited in their scope. The review is organized around three questions. First, what should be the dependent variables: the types of meanings that can adhere to brands? Second, what have we learned from studies on the types of visual, sensory, and human cues that are the sources of particular types of brand meaning—our independent variables? Third, what do we know, and need to know, about the inferential and other processes through which consumers “take possession” of these brand meanings from these cues? The review concludes with a research agenda.     

Metamorphic malicious code behavior detection using probabilistic inference methods

Existing antivirus programs detect malicious code based on fixed signatures; therefore, they have limitations in detecting metamorphic malicious code that lacks signature information or possesses circumventing code inserted into it. Research on the methods for detecting this type of metamorphic malicious code primarily focuses on techniques that can detect code based on behavioral similarity to known malicious code. However, these techniques measure the degree of similarity with existing malicious code using API function call patterns. Therefore, they have certain disadvantages, such as low accuracy and large detection times. In this paper, we propose a method which can overcome the limitations of existing methods by using the FP-Growth algorithm, a data mining technique, and the Markov Logic Networks algorithm, a probabilistic inference method. To perform a comparative evaluation of the proposed method's malicious code behavior detection, we performed inference experiments using malicious code with an inserted code for random malicious behavior. We performed experiments to select optimal weights for each inference rule to improve our malicious code behavior inferences’ accuracy. The results of experiments, in which we performed a comparative evaluation with the General Bayesian Network, showed that the proposed method had an 8% higher classification performance.     

On the blessing of abstraction

The “blessing of abstraction” refers to the observation that acquiring abstract knowledge sometimes proceeds more quickly than acquiring more speci?c knowledge. This observation can be formalized and reproduced by hierarchical Bayesian models. The key notion is that more abstract layers of the hierarchy have a larger “effective” sample size, because they combine information across multiple speci?c instances lower in the hierarchy. This notion relies on speci?c variables being relatively concentrated around the abstract “overhypothesis”. If the variables are highly dispersed, then the effective sample size for the abstract layers will not be appreciably larger than for the speci?c layers. Moreover, the blessing of abstraction is counterbalanced by the fact that data are more informative about lower levels of the hierarchy, because there is necessarily less stochasticity intervening between speci?c variables and the data. Thus, in certain cases abstract knowledge will be acquired more slowly than speci?c knowledge. This paper reports an experiment that shows how manipulating dispersion can produce both fast and slow acquisition of abstract knowledge in the same paradigm.