De keuze van het signifikantieniveau bij beleidsvoorbereidend sociaal-wetenschappelijk onderzoek

Samenvatting

Attention is drawn to the fact that in many instances in applied social policy research, levels of significance lower than the usual levels of alpha = .05 or alpha = .01 used in statistical tests are not only acceptable but even advisable. Various specific conditions are described where lower levels of significance are preferable, eg, when second order errors may have negative consequences in policy decisions. This condition is illustrated in such policy areas as industrial & labor relations, regional & Ur planning, & educational administration. 2 Figures. Modified HA.