Klesges L M, Klesges R C
Department of Psychology, Memphis State University, TN 38152.
Med Sci Sports Exerc. 1987 Oct;19(5):511-7.
The authors assessed the convergent validity and the sources of error in an electronic single-plane accelerometer (i.e., Caltrac Personal Activity Computer (Hemokinetics, Inc., Vienna, VA). The device was validated against observed all-day physical activity levels of children in their natural environment. Thirty pre-school children were observed in non-structured activity for periods of approximately 9 h while wearing the accelerometer. The results revealed moderately high but variable Spearman rank-order correlations between hourly readings of the accelerometer and the observational system (range of correlations = 0.62 to 0.95). An all-day accelerometer reading significantly correlated with the observational instrument (rho = 0.54). A step-wise regression analysis revealed that the best behavioral predictor of the all-day accelerometer reading was the observed behavior of walking, explaining 32% of the total variance. Older vs younger children (i.e., greater than 32.5 months), females vs males, and overweight (i.e., 75 percentile or greater) vs normal weight children tended to show higher correlations between direct observation and accelerometer readings. Implications of these findings and the utilization of the accelerometer in epidemiologic research are discussed.
作者评估了一款电子单平面加速度计（即Caltrac个人活动计算机，Hemokinetics公司，弗吉尼亚州维也纳）的收敛效度和误差来源。该设备根据在自然环境中观察到的儿童全天身体活动水平进行了验证。30名学龄前儿童佩戴加速度计，在非结构化活动中被观察约9小时。结果显示，加速度计每小时读数与观测系统之间的斯皮尔曼等级相关系数中等偏高但存在差异（相关系数范围为0.62至0.95）。全天加速度计读数与观测仪器显著相关（rho = 0.54）。逐步回归分析表明，全天加速度计读数的最佳行为预测指标是观察到的行走行为，解释了总方差的32%。年龄较大与较小的儿童（即大于32.5个月）、女性与男性、超重（即第75百分位数或更高）与正常体重儿童在直接观察与加速度计读数之间往往显示出更高的相关性。讨论了这些发现的意义以及加速度计在流行病学研究中的应用。
