THE PROMISE AND PITFALLS OF COMBINING GENETIC AND ECONOMIC RESEARCH
THE PROMISE AND PITFALLS OF COMBINING GENETIC AND ECONOMIC RESEARCH
复制标题
DOI：
10.1002/hec.1745
复制
发表时间：
2011-08-01
期刊：
影响因子：
2.1
通讯作者：
Fletcher, Jason M.
中科院分区：
文献类型：
--
作者：
Fletcher, Jason M.
随着人类基因组的测序以及由此产生的医学科学中基因型和表型之间关联研究的指数增长，分子遗传信息已经开始出现在几个最重要的社会科学数据集中，例如国家青少年健康纵向研究，脆弱家庭和威斯康星州纵向研究等。随着健康与退休研究、收入动态小组研究和全国青年纵向调查考虑增加此类数据，这种情况可能会进一步升级（见康利，2009）。这些新数据在社会科学研究中的两个早期用途是检查基因-环境相互作用和使用遗传数据作为工具;参见Conley和Rauscher（2010）和弗莱彻（2010），其中包括基因-环境相互作用研究的新方向。每一个方向都反映了在结合生物科学和社会科学的发现和方法方面取得的令人感兴趣的进展，但每一个方向也有相当大的局限性，需要认真加以解决。这篇社论将集中讨论基因作为工具的使用。迄今为止，在经济学文献中有两种类型的工具变量分析。第一个是由丁等人开创的。(2006，2009年），使用整个人口的遗传变异（包括家庭内和家庭之间）-在文献中通常称为孟德尔随机化。第二种是由弗莱彻和莱勒（2009 a，B）开创的，使用了家族内的遗传变异。Ding等人的工作主要使用被认为在大脑的“奖励”中心，多巴胺和血清素系统中运作的基因，在弗吉尼亚州高中生的平均成绩分析中，用于不良健康状况（例如ADHD，抑郁症，肥胖症）。这种方法的一个潜在的局限性（以及许多其他使用基因作为工具的经济学论文）是王朝效应的问题。如果学校表现和健康状况不佳存在很强的代际传递，那么在后代中起作用的遗传变异也会在父母一代中起作用。因此，这些工具与父母健康和学校表现之间可能存在机械相关性（包含在主方程的误差项中），这似乎使这些工具无效。这个问题，虽然看起来很普遍，但并不一定会使所有使用基因作为工具的分析无效，例如，对于没有这些王朝效应的行为。王朝影响的概念问题尚未被证明是无效的这种方法，但很少在文献中得到承认。除了潜在的王朝效应之外，孟德尔随机化框架可以说未能模仿实验设计，其中不同的个体具有相同的遗传某一等位基因的风险，并且偶然地是“幸运的”或“不幸的”。事实上，与王朝效应相关的问题是，只有父母携带“坏”等位基因的孩子才有风险。
With the sequencing of the human genome and the resulting exponential growth of research examining associations between genotype and phenotype in the medical sciences, molecular genetic information has begun to appear in several of the most important social science data sets, such as the National Longitudinal Study of Adolescent Health, Fragile Families, and the Wisconsin Longitudinal Study, among others. This appearance is likely to escalate further, as the Health and Retirement Study, the Panel Study of Income Dynamics, and the National Longitudinal Survey of Youth consider adding such data (see Conley, 2009). Two early uses of these new data in social science research have been the examination of gene–environment interactions and the use of genetic data as instruments; see Conley and Rauscher (2010) and Fletcher (2010), among others, for examples of new directions in gene–environment interaction research. Each direction reflects interesting steps forward in combining the findings and approaches of the biological and social sciences, yet each direction also has considerable limitations that will need to be carefully addressed. This editorial will focus on the use of genes as instruments.There have been two types of instrumental variables analyses so far in the economics literature. The first, pioneered by Ding et al.(2006, 2009), uses genetic variation across the population (both within and between families)–often called Mendelian Randomization in the literature. The second, pioneered by Fletcher and Lehrer (2009a, b), uses genetic variation within families. The Ding et al. work mainly used genes thought to operate in the ‘reward’centers of the brain, the dopamine and serotonin systems, to instrument for poor health conditions (eg ADHD, depression, obesity) in an analysis of grade point average outcomes for high school students in Virginia. One potential limitation with this approach (and many other papers in economics that use genes as instruments) is the issue of dynastic effects. If there is a strong intergenerational transmission of school performance and poor health, then the genetic variants operating in the offspring generation would have also operated in the parent generation. Thus, there is likely a mechanical correlation between the instruments and parent health and school performance (contained in the error term of the main equation), which would seem to invalidate the instruments. This issue, while seemingly pervasive, would not necessarily invalidate all analyses using genes as instruments–for example, for behaviors without these dynastic effects. The conceptual issue of dynastic effects has not been demonstrated to invalidate this approach, but has rarely been acknowledged in the literature. In addition to potential dynastic effects, the Mendelian Randomization framework arguably fails to mimic an experimental design, where different individuals are at the same risk of inheriting a certain allele, and by chance are either ‘lucky’or ‘unlucky’. Indeed, and related to the issue of dynastic effects, only children with parents with ‘bad’alleles are at risk of