Grissom R J
Department of Psychology, San Francisco State University, California 94132, USA.
J Consult Clin Psychol. 1996 Oct;64(5):973-82. doi: 10.1037//0022-006x.64.5.973.
The "probability of superiority estimate" (PS) estimates the probability that a randomly sampled client from a population given a treatment will have an outcome that is superior to that of a randomly sampled client from a population given another treatment. The meta-analytic clinical outcome literature was examined to calculate mean PS (PS) for comparisons involving therapy versus control, therapy versus placebo, therapy versus therapy, and placebo versus control. The range of PS was found to be approximately .7 +/- .2, with median PS greatest when therapy and control are compared (Mdn PSTC = .70, where T = therapy and C = control) and least when 2 therapies are compared (Mdn PSTT = .56). Results suggested that there is more to therapeutic success than placebo effects (Mdn PSTP = .66, where T = therapy and P = placebo) and that placebo is typically better than do-nothing control conditions (Mdn PSPC = .62). The present exceptionally large study, controlling for dependencies and confounding variables, may put to rest the question of the superiority of therapy to placebo. It also appears that the strength of effect of therapy is typically at least average among the effects of independent variables in psychology.
“优势概率估计”（PS）估算的是，在接受一种治疗的总体中随机抽取的一位患者，其治疗结果优于在接受另一种治疗的总体中随机抽取的一位患者的概率。我们查阅了荟萃分析临床结果文献，以计算在治疗与对照、治疗与安慰剂、治疗与治疗以及安慰剂与对照的比较中的平均PS（PS）。结果发现，PS的范围约为0.7±0.2，当比较治疗与对照时，中位数PS最大（Mdn PSTC = 0.70，其中T =治疗，C =对照），而当比较两种治疗时，中位数PS最小（Mdn PSTT = 0.56）。结果表明，治疗成功不仅仅是安慰剂效应（Mdn PSTP = 0.66，其中T =治疗，P =安慰剂），而且安慰剂通常比不做任何处理的对照情况要好（Mdn PSPC = 0.62）。这项控制了相关性和混杂变量的规模极大的研究，或许可以平息关于治疗相对于安慰剂的优越性的问题。此外，治疗效果的强度在心理学自变量的效果中似乎通常至少处于平均水平。
