Assessing Sensitivity to Unmeasured Confounding Using a Simulated Potential Confounder

ABSTRACT

A major obstacle to developing evidenced-based policy is the difficulty of implementing randomized experiments to answer all causal questions of interest. When using a nonexperimental study, it is critical to assess how much the results could be affected by unmeasured confounding. We present a set of graphical and numeric tools to explore the sensitivity of causal estimates to the presence of an unmeasured confounder. We characterize the confounder through two parameters that describe the relationships between (a) the confounder and the treatment assignment and (b) the confounder and the outcome variable. Our approach has two primary advantages over similar approaches that are currently implemented in standard software. First, it can be applied to both continuous and binary treatment variables. Second, our method for binary treatment variables allows the researcher to specify three possible estimands (average treatment effect, effect of the treatment on the treated, effect of the treatment on the controls). These options are all implemented in an R package called treatSens. We demonstrate the efficacy of the method through simulations. We illustrate its potential usefulness in practice in the context of two policy applications.     

Grading leniency, grade discrepancy, and student ratings of instruction

The purpose of this study was to examine how grading leniency and grade discrepancy (the difference between expected grades and deserved grades) were associated with various dimensions of student ratings of instruction. A sample of 754 undergraduate college students completed a student ratings of instruction instrument and provided responses to a number of other questions on topics such as course difficulty and workload. A series of multilevel regression analyses were conducted and results showed that an instructor's grading leniency, as perceived by students, was positively associated with student ratings on 11 of 12 dimensions of instruction examined. This finding suggests that more lenient instructors tend to receive higher student ratings. The second finding shows that grade discrepancy was negatively associated with most dimensions of instruction. This supports the self-serving bias hypothesis under attribution theory (Gigliotti & Buchtel, 1990) in that students tended to punish instructors with lower ratings when expected grades were lower than students believed they deserved, yet little evidence of a pattern of rewards existed in student ratings when students expected grades higher than they deserved.     

Effects of Achieving Tendency,Gender, and Outcome on Causal Attributions following Motor Performance

A test of Weiner's propositions of the effects of achieving tendency on causal attributions was conducted using a stabilometer performance as the criterion task. The Achieving Tendency Scale and the Internal-External Control Scale were administered to 930 college students. From these, 120 subjects who had extreme scores on the ATS and moderate scores on the I-E Scale were selected for the experiment. Subjects performed one 30-second trial on either a spring-assisted or a non-spring-assisted stabilometer. The importance of the causal factors of ability, effort, task difficulty, and luck as influences on performance were then rated. A 2 × 2 × 2 MANOVA was performed across the ratings of the four factors, with achieving tendency, gender, and outcome considered as independent variables. The only significant effect was a main effect for outcome (p < .001). Subsequent univariate analyses indicated that only effort attributions differed as a function of outcome (p < .001). Those who succeeded perceived effort to be more important than those who failed. Also, a significant univariate effect (p < .02) was found for gender on the luck factor; females perceived luck to be more important than did males. These results are discussed in terms of methodological and theoretical considerations.     

Confirmation biases in selective exposure to political online information: Source bias vs. content bias

The present work examines the role of source vs. content cues for the confirmation bias, in which recipients spend more time with content aligning with preexisting attitudes. In addition to testing how both source and content cues facilitate this biased pattern of selective exposure, the study measures subsequent attitude polarization. An experiment (N?=?120) presented messages with opposing political stances, associated with unbiased or slanted sources. Software tracked selective exposure in seconds, and attitudes were measured before, immediately after, and two days after message exposure. Further, information processing styles were assessed. The confirmation bias emerged regardless of source quality. Information processing styles moderated the confirmation bias as well as selective exposure to messages from unbiased vs. slanted sources. Selective exposure reinforced attitudes days later.     

A ‘power law’ based method to reduce size-related bias in indicators of knowledge performance: An application to university research assessment

The knowledge production provided by universities is essential to sustaining a country’s long-term economic growth and international competitiveness. Many nations are thus driving to create sustainable and effective funding environments. The evaluation of university knowledge, productivity and research quality becomes critical, with ever increasing share of public funding allocated on the basis of research assessment exercises. Nevertheless, the existing methods to assess the universities’ knowledge production are often affected by limits and biases, extensively discussed in the scientific literature.In this paper we study how to reduce the effect of size-related bias due to university size on the indicators of knowledge performance used in evaluation exercises. We propose an innovative utilization of the scale-free property of the power laws as a scaling relationship, to normalize research productivity indicators, and provide results independent by the university size. Our method has evident policy implications and gives a contribution for the future design of assessment exercises.We apply our findings in a recent Italian research assessment exercise.     

The effects of narrative messages on optimistic bias in South Korea: a focus on controllability,collectivism, and risk perception in a massive fire crisis

This study examined the effect of narrative messages of a massive fire crisis on optimistic bias by experimentally comparing the effect of narrative describing a personal story on the crisis incident and that of non-narrative message (Study 1). Researchers further sought the interaction between controllability and the narrative message and the mediated moderation model of risk perception. In Study 2, the effect of narrative message describing a group story on the crisis incident on optimistic bias was further tested in terms of South Korea’s collectivistic culture. Collectivism, along with controllability, was used as a moderator, and mediated moderation models of risk perception were tested. The present research offers several major findings: (1) a narrative message describing a personal story decreased optimistic bias, (2) among people who read a narrative describing a personal story, those with high controllability had a lower level of optimistic bias than those with low controllability, (3) among people who read the narrative of a group story, those with high collectivism had a lower level of optimistic bias than those with low collectivism, and (4) the interaction between message types and collectivism affected risk perception and this risk perception increased optimistic bias. Theoretical implications of these findings are discussed.     

英语中的性别歧视现象及其规避策略

语言是伴随着人类社会的发展而发展.从社会语言学的角度,从历史与现实两方面探讨性别歧视,此现象在词法、性、语义、社会习俗及婚姻中各有不同的表现形式,可有六点规避策略.     

元分析方法研究综述

本文对目前医学及社会科学领域学者采用的定量化文献综述方法----元分析方法进行了文献调研,从元分析的理论入手对该方法的实质、优势进行了简要介绍,通过对该方法在国内外医学、社会科学及图书情报学领域的应用研究梳理,归纳整理了元分析方法在实践中的操作流程。本文以为作为一种有效集成已有研究的方法,元分析方法在情报学领域得到的关注还不够,其原因是多方面的,如缺乏实证数据支撑、出版偏好等。     

论基督教作家笔下的犹太人形象

在欧洲文学作品中，犹太人一般以贪婪、守财、狡黠、多虑的高利贷者、投机奸商等形象出现，且处于陪衬基督徒形象的从属地住。这一形象与犹太民族的客观形象是不吻合的。基督教作家对犹太人的偏见有着社会时代、文化心理、宗教情绪等多方面的原因。莎士比亚和司各特笔下的犹太人以不同的方式反抗歧视和侮辱，使这些形象具有了不同寻常的积极意义。     

Language effects in international testing: the case of PISA 2006 science items

We investigate the extent to which language versions (English, French and Arabic) of the same science test are comparable in terms of item difficulty and demands. We argue that language is an inextricable part of the scientific literacy construct, be it intended or not by the examiner. This argument has considerable implications on methodologies used to address the equivalence of multiple language versions of the same assessment, including in the context of international assessment where cross-cultural fairness is a concern. We also argue that none of the available statistical or qualitative techniques are capable of teasing out the language variable and neutralising its potential effects on item difficulty and demands. Exploring the use of automated text analysis tools at the quality control stage may be successful in addressing some of these challenges.