Exam 3: Pre-Analysis Data Screening

If the researcher determines that the data have deviated from normal, she or he can consider transforming the data by:

One of the characteristics of multivariate normality is that any linear combination of the variables must be nonnormally distributed.

Univariate outliers are cases with extreme values on one variable.

Robustness refers to the relative insensitivity of a statistical test to violations of the underlying inferential assumptions.

If a distribution differs only moderately from normal, a log transformation should be obtained.

Homoscedasticity is the assumption that the variability in scores for one continuous variable is roughly the same at all values of another continuous variable.

The problem with outliers is that they can distort the results of a statistical test.

Cases with unusual or extremely large values at one or both ends of a sample distribution are known as outliers.

Residuals are defined as the portion of scores not accounted for by the multivariate analysis.

A statistical procedure known as Mahalanobis distance can be used to identify outliers of any type.