Matskási István (szerk.): A Magyar Természettudományi Múzeum évkönyve 94. (Budapest 2002)

Radoinova, D., Tenekedjiev, K. ; Yordanov, Y.: Stature estimation according to bone length in Hungarian population

terval. The adjusted coefficient of multiple determination is R

2 = 0.959. The 95% confidence intervals of the model coefficients are respectively: from 87.92 cm to 94.96 cm and from 2.285 cm to 2.493 cm. The regression parameters are significant (t-tests with P-value's < 0.0005) and the model (6) is adequate (ANOVA with P-value < 0.0005). The three formulae (BREITINGER, TROTTER-GLESER & DUPERTUIS-HADDEN) described the observed sample inadequately (P-value = 0.000% for each of the three formulae). The nomogram on humerus is shown on Fig. 5. II.

2. Maximal stature regression on fibula - Seven outliers are rejected - 3 in the first loop and 4 in the second loop (n = 186-7 = 179). Heteroskedasticity is detected by 6 tests: WHITE test (linear) with P-value = 0.0449, GLEJSER test (reciprocal) with P-value = 0.0499, GOLDFELD-QUANDT test with P-value = 0.0051 and by three tests of BREUSCH-PAGAN - linear with P-value = 0.0302, reciprocal with P-value = 0.0262 and root with P-value = 0.0291. Just the reciprocal model for residuals' module is adequate (ANOVA P-value = 0.0499) but is not acceptable (R

2

= 0.016). This proved practically negligible heteroskedasticity. The derived regression formula is: (7) E[stature] = 88.21 + 2.283*Fi (1.063) (0.028) Fig. 6. Nomogram for predicting the maximal stature of Hungarian males using the length of fibula with age correction according to GILES and BORCAN (bold line). Three confidence margins are plotted (95%- solid line, 99% dashed line and 99.9% dashdotted). The outliers are plotted with circles, and the experimental data is shown with dots