在本文中，我们提出了用于检测倾向得分设定是否错误的有效检验统计量。这些统计量可用于评估依赖于倾向得分的正确设定的治疗效果估计的有效性。当协变量向量维数较高时，我们的检验统计量较少受到“维数灾难”的影响，并且它是完全数据驱动的，不需要调整如带宽之类的参数，而且能够检测以参数速率n^(-1/2)收敛到原假设的局部备择假设，n为样本大小。相较Song的检验统计量，我们提出的检验统计量不是方向性的，并且避免在了某些情况下失去功效。通过数值模拟和实际数据分析来测试有限样本下检验统计量的性能，发现我们的统计量在所考虑情形下均表现良好。

In this paper, we develop e?cient test statistic for detecting propensity score mis-speci?cation. These tests may be applied to assess the validity of di?erent treatment e?ects estimators that rely on the correct speci?cation of the propensity score. Our tests do not su?er from the “curse of dimensionality” when the vector of covariates is of high-dimensionality, are fully data-driven, do not require tuning parameters such as bandwidths, and are able to detect local alternatives converging to the null at the parametric rate n^(-1/2), with n the sample size. Comparing the test statistics of Song, our proposed test statistics are not directional and avoid losing powers for certain cases. The ?nite sample performance of the tests are examined by means of a Monte Carlo experiment and an empirical application, and our test perform well in all these situation.