Parametrial statistial riteria (parametri statistial tests)
with. To., unlike nonparametri or free riteria from distribution, promotion of the assumptions, onerning harateristis, properties and demand the form of distributions of sets, from to- samples analyzed by means of them are taken. Assumptions onern to number of most often demanded assumptions about normalnosti distributions of sets (t, F, r/R) uniformity of a dispersion (t, F), linearities of regress (bxy, r/R), ontinuity of measures/variables and equality of intervals of a measuring sale. In that, as to hek of zero hypotheses, parametrial riteria possess greater apaity in omparison with nonparametri, at equality of all other fators. However, more likely by virtue of the universality, than apaity, parametrial riteria beame most rasprostr. Tools of the researher in the field of psyhology.
In multivariate statistis parametrial riteria are based, gl. obr., on two important assumptions, namely, that) inluded in the analysis of data variables have multivariate normal distribution and that dispersive matrixes (basially, matrixes of dispersions and kovariatsy) are idential to all groups. The entral plae among parametrial riteria in multivariate statistis borrow -= Uilksa and offered Bartletom approximation of distribution through distribution (V - riterion). - riterion Uilksa - typial statistis for hek of the importane of distintions between group tsentroidami.
V-statistis Bartleta is a natural logarithm , to- it is possible to use as parametrial statistial riterion for hek of a zero hypothesis about equality nesk. Group tsentroidov. Statistis V has distribution with number of degrees of freedom n (G - ). Versions And V are applied as parametrial methods of hek of a zero hypothesis in various appendies of multivariate statistis, napr. In MANOVA, the multivariate disriminantal analysis and the analysis of initial orrelations.
see also Nonparametri statistial riteria, Statistis in psyhology