Parametric tests require data that meet certain conditions. Which of the following describes this?

Parametric tests require data that meet certain conditions. Which of the following describes this?

• A. They require only ordinal data with no distribution assumptions.
• B. They rely on assumptions about the population distribution and data scale.
• C. They do not assume equal variances.
• D. They are unaffected by outliers.

Answer: (B)

Parametric tests rely on assumptions about the data in two key ways: the population distribution and the measurement scale. Most parametric methods assume the population from which the sample is drawn is normally distributed (or that the sampling distribution of the statistic is normal enough for the sample size), and that the data are measured on an interval or ratio scale so that means and standard deviations are meaningful. These conditions let us use the mean as a measure of central tendency and the variance to quantify spread, which underpins hypothesis tests and confidence intervals. If data are ordinal or the distribution is seriously nonnormal, those mean-based inferences can be invalid, which is why nonparametric methods are often preferred in such cases. It’s also not accurate to say parametric tests are unaffected by outliers—outliers can distort means and variances and thus affect results—nor is the claim about equal variances a universal requirement for all parametric tests, since some tests handle unequal variances while others assume they are equal.

Parametric tests rely on assumptions about the data in two key ways: the population distribution and the measurement scale. Most parametric methods assume the population from which the sample is drawn is normally distributed (or that the sampling distribution of the statistic is normal enough for the sample size), and that the data are measured on an interval or ratio scale so that means and standard deviations are meaningful. These conditions let us use the mean as a measure of central tendency and the variance to quantify spread, which underpins hypothesis tests and confidence intervals. If data are ordinal or the distribution is seriously nonnormal, those mean-based inferences can be invalid, which is why nonparametric methods are often preferred in such cases. It’s also not accurate to say parametric tests are unaffected by outliers—outliers can distort means and variances and thus affect results—nor is the claim about equal variances a universal requirement for all parametric tests, since some tests handle unequal variances while others assume they are equal.