Model Specification in Media Effects Research.
Model Specification in Media Effects Research.
复制标题
媒体效果研究中的模型规范。
DOI：
10.1007/s10508-023-02714-3
复制
发表时间：
2023
影响因子：
3.8
通讯作者：
Herbenick,Debby
中科院分区：
文献类型：
--
作者：
Wright,PaulJ;Tokunaga,RobertS;Herbenick,Debby
在最近的一封给编辑的信（赖特，2021）中，我提出了性媒体效应研究人员过度强调第三变量的概念化作为观察（即，相关性）研究中的“潜在混淆”的可能性。从混淆的角度来看，接触性媒体和性结果之间的联系是虚假的，是由一个或多个第三变量（即混淆）引起的（Baer等人，2015; Wright，2020 a）。我注意到其他媒体效应学者最近提出的类似建议（Bushman &安德森，2023;斯莱特，2015），并指出其他学科的社会和行为科学家也提出了类似的建议（贝克尔，2005;伯纳斯和阿吉尼斯，2016;卡尔森和吴，2012;米尔，1971;斯特曼等人，2009; Spector & Brannick，2011）。最近，作为例证，我们在一项关于美国成年人错误性观念的研究中发现，数据与人口统计学变量（例如，婚姻状况、年龄、种族、教育）被设想为预测感知色情现实主义（这反过来又预测了错误的性观念）而不是传统的方法，在传统的方法中，观众的人口统计数据被视为感知的色情现实主义和错误的性信仰之间的关联的混淆（Wright等人，2022 a，2022 b;另见Wright & Tokunaga，2023）。在对这些发现的讨论中，我们鼓励后续的研究努力探索其他表面上的第三变量混淆是否可能在替代能力中更好地拟合数据。
In a recent Letter-to-the-Editor (Wright, 2021), I broached the possibility that sexual media effects researchers have overemphasized the conceptualization of third-variables as “potential confounds” in observational (ie, correlational) studies. From the confounding perspective, links between exposure to sexual media and sexual outcomes are spurious and caused by one or more third-variables (ie, confounds)(Baer et al., 2015; Wright, 2020a). I noted similar recent suggestions from other media effects scholars (Bushman & Anderson, 2023; Slater, 2015) and pointed to parallel refrains from social and behavioral scientists in other disciplines (Becker, 2005; Bernerth & Aguinis, 2016; Carlson & Wu, 2012; Meehl, 1971; Schisterman et al., 2009; Spector & Brannick, 2011). More recently, as illustration, we showed in a study of US adults’ erroneous sexual beliefs that the data were more consistent with a model wherein demographic variables (eg, marital status, age, race, education) were envisioned as predictors of perceived pornography realism (which in turn predicted erroneous sexual beliefs) than the conventional approach wherein audience demographics were treated as confounds of the association between perceived pornography realism and erroneous sexual beliefs (Wright et al., 2022a, 2022b; see also Wright & Tokunaga, 2023). In the discussion of these findings, we encouraged subsequent research efforts probing whether other ostensible third-variable confounds might fit the data better in alternative capacities.