PURPOSE: Measurement of the testis is a more readily available method of estimating spermatogenesis. Doubt remains about the best instrument for measuring testicular volume. Lack of bias or accuracy of instruments has received too much emphasis in some studies, while to our knowledge no one has yet appropriately compared reliability statistically. We propose a simple new method for measuring testicular size based on visual comparison with graphic models, and describe the reliability and bias of this and 4 traditional methods. MATERIALS AND METHODS: Measurements of 42 adolescent testes were made in a certain sequence: graphic method, dimensional measurement, Prader orchidometer, ring orchidometer and ultrasound with ultrasound assumed to be the standard. Statistical analysis was based on the linear structural model. RESULTS: Statistical tests indicated that all 5 methods are equally reliable (R > 0.9). Although they are not equally accurate, actual testicular size can be calculated using each of these 5 methods and the equations of the linear structural model. CONCLUSIONS: The new graphic method proposed in this study is as reliable as other well-known methods for measuring testicular size. Actual testicular volume can be estimated without bias and with equal reliability from any of the 5 methods using the equations of the linear structural model. This statistical approach is more relevant than the sole comparison of lack of bias or accuracy, which has been the main concern of previous studies.
PURPOSE: Measurement of the testis is a more readily available method of estimating spermatogenesis. Doubt remains about the best instrument for measuring testicular volume. Lack of bias or accuracy of instruments has received too much emphasis in some studies, while to our knowledge no one has yet appropriately compared reliability statistically. We propose a simple new method for measuring testicular size based on visual comparison with graphic models, and describe the reliability and bias of this and 4 traditional methods. MATERIALS AND METHODS: Measurements of 42 adolescent testes were made in a certain sequence: graphic method, dimensional measurement, Prader orchidometer, ring orchidometer and ultrasound with ultrasound assumed to be the standard. Statistical analysis was based on the linear structural model. RESULTS: Statistical tests indicated that all 5 methods are equally reliable (R > 0.9). Although they are not equally accurate, actual testicular size can be calculated using each of these 5 methods and the equations of the linear structural model. CONCLUSIONS: The new graphic method proposed in this study is as reliable as other well-known methods for measuring testicular size. Actual testicular volume can be estimated without bias and with equal reliability from any of the 5 methods using the equations of the linear structural model. This statistical approach is more relevant than the sole comparison of lack of bias or accuracy, which has been the main concern of previous studies.