Kineziologija, Vol. 5. No. 1.-2., 1975.
Izvorni znanstveni članak
The reliability of some composite tests of primary motoric abilities
Konstantin Momirović
; Institut za kineziologiju fakulteta za fizičku kulturu
Boris Wolf
; Centar za klasifikaciju i selekciju ljudstva za potrebe JNA
Sažetak
On the sample of males 19-27 years odl a battery of 110 motoric tests was measured. The experiment showed that it is possible to construct very reliable motoric tests of a composite type, even to estimate motoric abilities such as coordination ans accuracy of aiming and shooting, for which such tests didn't exist before. It is of course possible, if the construction of motoric tasks and necessary number of items are determined on the basis of precise pre-experimental work and if the measuring conditions are strictly standardised, including, of course, the kinesiometric education of the experimenters.
Different measures of reliability don't have the same value for estimating of this metrical characteristic. For a great deal of motoric tests reliability coefficients, inferred from classic error theory, are of little or no use. Also, reliability coefficients, inferred on the basis of the hypothesis that the best result would be obtained by summing the results in items, are of no use.
The most dependable reliability measures were obtained by methods which hypothesised unequal value of different items for determining the whole test result. It seems that it is the most reliable to estimate the total test result as the first principal component of an antiimage covariance matrix of items.
Besides classic generalizability coefficients of so inferred results, for determining the real reliability of motric tests, coefficients of representativeness, based on the non-zero error covariance model, can be used.
Ključne riječi
Hrčak ID:
240633
URI
Datum izdavanja:
1.1.1975.
Posjeta: 1.383 *