Experimental Research Methods (Fach) / L9 (Lektion)

Diese Lektion wurde von tklmn erstellt.

Lektion lernen

Diese Lektion ist leider nicht zum lernen freigegeben.

zurück  |  weiter 1 / 1

• only reduction of error variance use an experimental design for answering the research question
• Assumptions ANCOVA The first three assumptions of ANCOVA are identical to those of one-way ANOVA:1. Independence of observations2. Homogeneity of variance → Levene’s test3. Normally distributed residuals Four additional assumptions:4. Covariate is measured before the experimental manipulation5. Covariate is measured without measurement error6. Relation between X and Y is linear7. Relation between X and Y is the same for all levels of Group (“homogeneity of regression slopes”) →no interaction between X and GroupANCOVA is not robust to violations of assumptions 4-7
• Repeated measures ANOVA • Repeated measures ANOVA distinguishes itself from previous techniques because for each person wehave more than on data point or observation on Y → multiple dependent variables per person• Subjects/persons are measured at two or more measurement occasions (longitudinal designs)• Repeated measures ANOVA is also called a “within-subjects” design/analysis• Examples:• Test-retest designs• Panel studies• Diary studies
• Why do we need different analysis methods in the case of multiple observations per person (repeated measures ANOVA)? • One of the assumptions of the techniques that were already discussed (ANOVA, ANCOVA) is that allobservations are independent (errors uncorrelated)• Observations of one person are dependent!• AN(C)OVA is not robust against violation of the assumption of independence of observations: true p- value strongly deviates from the computed p-value• The more ‘dependent’ (higher correlation between) observations of a person, the larger the violationof assumption, the larger the difference between true p-value and computed p-value

zurück  |  weiter 1 / 1