Schwabe Inga, van den Berg Stéphanie M
Department of Research Methodology, Measurement, and Data Analysis, University of Twente, Drienerlolaan 5, 7522 NB, Enschede, The Netherlands,
Behav Genet. 2014 Jul;44(4):394-406. doi: 10.1007/s10519-014-9649-7. Epub 2014 Mar 20.
Considerable effort has been devoted to establish genotype by environment interaction (G x E) in case of unmeasured genetic and environmental influences. Although it has been outlined by various authors that the appearance of G x E can be dependent on properties of the given measurement scale, a non-biased method to assess G x E is still lacking. We show that the incorporation of an explicit measurement model can remedy potential bias due to ceiling and floor effects. By means of a simulation study it is shown that the use of sum scores can lead to biased estimates whereas the proposed method is unbiased. The power of the suggested method is illustrated by means of a second simulation study with different sample sizes and G x E effect sizes.
在未测量的遗传和环境影响的情况下，人们投入了大量精力来建立基因型与环境的相互作用（G×E）。尽管许多作者已经指出，G×E的出现可能取决于给定测量尺度的属性，但仍然缺乏一种无偏的方法来评估G×E。我们表明，纳入一个明确的测量模型可以纠正由于天花板效应和地板效应导致的潜在偏差。通过模拟研究表明，使用总分可能会导致有偏差的估计，而所提出的方法是无偏的。通过第二个模拟研究，使用不同的样本量和G×E效应大小，说明了所建议方法的功效。
