statistiek assumpties


Assumptions (General)

  • Variable Matching:

    • Ensure that the variables align with the analyses you intend to conduct.

  • Outliers and Influential Cases:

    • Check for outliers or influential cases that may distort results; these should not be included in the sample.

  • Confounding Variables: No confounding variables are assumed to affect the outcome.

  • Essential Assumptions:

    • Independent observations/errors

    • The sample is a representation of the population intended for generalization

    Errors must be normally distributed.

    • Homoscedasticity (homogeneity of variances): assessed through PRED/RESID plots, Levene’s test, or Sphericity tests.

Correlations

  • Type of Variables:

    • Continuous variables are necessary for calculating correlations.

  • Extra Assumption:

    • Linearity: There should be a linear relationship between the variables.

Multiple Regression Analysis

  • Types of Variables:

    • Outcome/Dependent Variable: Must be continuous.

    • Predictors/Independent Variables: Can be either continuous or dichotomous.

  • Extra Assumptions:

    • There must be a linear relationship between the outcome variable and the predictors.

    • No Multicollinearity: There should be no multicollinearity among the predictors, indicating they should not be highly correlated with each other.

Hierarchical Regression Analysis

  • Types of Variables:

    • Outcome/Dependent Variable: Continuous.

    • Predictors/Independent Variables: Can be continuous or dichotomous.

    • Control Variables: Can also be continuous or dichotomous.

  • Extra Assumptions:

    • Linear relationship between the outcome variable and predictors.

    • No Multicollinearity: Same as in multiple regression; no multicollinearity among predictors or controls.

Mediation/Moderation Analysis

  • Types of Variables:

    • Outcome/Dependent Variable: Continuous.

    • Mediator: Continuous.

    • Moderator: Can be continuous or dichotomous (dummy variables are dichotomous).

    • Predictors/Independent Variables: Can be continuous or dichotomous.

  • Extra Assumptions:

    • Identical to those in regression analyses; must verify linearity and multicollinearity.

AN(C)OVA (Analysis of Covariance)

  • Types of Variables:

    • Dependent Variable: Must be continuous.

    • Independent Variable: Nominal (categorical).

    • Optional:

    • Covariate: Can be continuous or dichotomous.

  • Extra Assumptions: (only when including a covariate)

    • There must be a linear relationship between the dependent variable and the covariate.

    • Homogeneity of Regression Slopes: The slopes of the regression lines must be homogenous for the covariate across the levels of the independent variable.

    • Independence: Covariate must be independent of the independent variables.

    • Multiple Covariates: If using multiple covariates, check for multicollinearity among them.

Contrast Tests (Field 4th Ed. Page 456)

  • Types of Contrast Tests:

    • Deviation: Compares every level (except “first” or “last”) to the total effect.

    • Simple: Compares each level to the “first” or “last” level.

    • Repeated: Every level is compared to the next level sequentially.

    • Helmert: Compares each level to the total of the consecutive levels.

    • Difference: The reverse of the Helmert contrast.

    • Polynomial: Determines the type of effect for within factors (linear, quadratic, cubic, etc.).