Dear Editor and readers of the Journal of Strength and Conditioning Research, this letter was written to clarify some issues of the researcher Rodrigo Bini, in relation to our study entitled “Influence of saddle height on lower limb kinematics in well trained cyclists” (3).
The first issue was related to the possible discrepancy between our study (3) and the recent study published by Peveler et al. (6). From our point of view, both studies are complementary instead of discrepant. We proposed a novel equation based on kinematic and anthropometric measurements to adjust the saddle height using clipless pedals. Previous equations as far back as the 1970s and 1980s were valid only to toe-clip pedals (4,5). In our study, we highlighted that “Besides anthropometrics, we recommend that coaches should consider a kinematic analysis of their bicycle configuration to optimize pedaling efficiency” (3). These arguments complement Peveler et al.'s conclusions: “Bike fit sessions are typically conducted with the cyclists in a stationary position. While stationary measurements are supported in the literature and are initially important in the bike-fitting process, this study supports the use of dynamic measures in such a process, too” (6). In the introduction of our article, we stated that dynamic fit was not recognized as a method to adjust the saddle height in a recent review (1).
The second question was related to Bini's affirmation “…it is not logical to set cyclists in a standard saddle height. Evidence has shown that changes in saddle height <5% of lower leg length do not affect cycling performance or injury risk.” First, this affirmation was contradictory in essence. If varying 5% of the saddle height does not affect cycling performance or injury risk, then no method would be valid, because this variation represents a wide range of saddle height (3–4 cm). Second, to the best of our knowledge, no longitudinal study demonstrated that changing the saddle height affected the injury rate in cycling. This consideration was performed by Reviewer 1 of our article (3), and we discarded any reference on this topic. In addition, various studies (4,5,7) demonstrated that varying the saddle height by 4% affects the pedaling efficiency. In conclusion, we consider that our study and the study of Peveler are complementary. Both studies highlighted the importance of initiating saddle height adjustment with stationary methods and recommending the use of kinematics analysis in the bike fit process.
The third issue was related to the statistical methods used in our study (1). The Kolgomorov-Smirnov test showed that the sample was normally distributed. All the variables (inseam length, knee flexion angle, hip flexion angle, ankle flexion angle, hamstring flexibility and saddle back) obtained p-values >0.05. However, it is known that the Shapiro-Wilk test is much more sensitive when analyzing samples less then 30 subjects (8). For this reason, after the comment of Bini “how authors could ensure a gaussian distribution using Kolgomorov-Smirnov test,” we decided to use the Shapiro-Wilk test obtaining similar results (p values >0.05). Concerning the multivariate lineal regression analysis, we agree with Bini that, in our study, we should have clarified the statistical process used to reach the final equation. A backward stepwise method was used to predict saddle height based on kinematics and anthropometric data. To avoid multicollinearity, hip angle was not used in the regression because it presented a high correlation with knee angle. Therefore, 4 independent variables (inseam length, knee angle, ankle angle, and flexibility) were assessed obtaining the subject to variable ratio of 5.75. Cattell (2) suggested that this ratio should be in the range of 3–6 (2) and Tabachnick and Fidell (9) argued for a minimum ratio of 5 (9). Considering these recommendations, in our study, the subject to variable ratio seems to be within the recommendation limits, although the sample used was not large. The backward regression method tested each variable, discarding in every step the least significant one. Variables were allowed to enter the model at p ≤ 0.10, but only those variables with p ≤ 0.05 remained in the model. Finally, inseam length and knee angle were the most appropriate variables to predict saddle height (R2 = 0.937; F = 148.008; p < 0.001). Standard error of prediction was <1%.
where SH is the saddle height in centimeters, E is the inseam length in centimeters, and KA is the knee flexion angle in degrees.
To assess the validity of the regression model, a scatter plot between the residual and predicted values was used (Figure 1). We found that residuals were randomly distributed with mean zero and constant variance. The value of the Durbin-Watson statistic was 1.96 showing non–autocorrelation among the residuals. In addition, the Shapiro-Wilk test showed that residuals were normally distributed (p > 0.05) (Table 1).
1. Bini RR, Hume PA, Croft JL. Effects of bicycle saddle height on knee injury risk and cycling performance. Sports Med 41: 463–476, 2011.
2. Catell RB. The Scientific Use of Factor Analysis in Behavioral and Life Sciences. New York: Plenum Press, 1978.
3. Ferrer-Roca V, Roig A, Galilea P, García-López J. Influence of saddle height on lower limb kinematics in well-trained cyclists: Static versus dynamic evaluation in bike fitting. J Strength Cond Res. In press. doi: 10.1519/JSC.0b013e318245c09d.
4. Hamley EJ, Thomas V. Physiological and postural factors in the calibration of the bicycle ergometer. J Physiol 191: 55–56, 1967.
5. Nordeen-Snyder KS. The effect of bicycle seat height variation upon oxygen consumption and lower limb kinematics. Med Sci Sports 9: 113–117, 1977.
6. Peveler WW, Shew B, Johnson S, Palmer TG. A kinematic comparison of alterations to knee and ankle angles from resting measures to active pedaling during a graded exercise protocol. J Strength Cond Res. In press. doi:10.1519/JSC.0b013e318243fdcb.
7. Price D, Donne B. Effect of variation in seat tube angle at different seat heights on submaximal cycling performance in man. J Sports Sci 15: 395–402, 1997.
8. Shapiro SS, Wilk MB. An analysis of variance test for normality (complete samples). Biometrika 52: 591–611, 1965.
© 2013 National Strength and Conditioning Association
9. Tabachnick BG, Fidell LS. Using Multivariate Statistics. (2nd ed.). Cambridge, MA: Harper and Row, 1989.