Data considerations for Test for Equal Variances

To ensure that your results are valid, consider the following guidelines when you collect data, perform the analysis, and interpret your results.

The data should include at least one categorical factor
For more information on factors, go to Factors and factor levels.
The response variable should be continuous
If a response variable is categorical, you cannot use the equal variances test because a valid standard deviation does not exist. The response column must contain numeric values, such as weights.
The sample sizes should be greater than 20
If some of the groups have fewer than 20 samples, the multiple comparisons p-value might not be valid.
Each observation should be independent from all other observations
If your observations are dependent, your results might not be valid. Consider the following points to determine whether your observations are independent:
  • If an observation provides no information about the value of another observation, the observations are independent.
  • If an observation provides information about another observation, the observations are dependent.
The sample data should be selected randomly

Random samples are used to make generalizations, or inferences, about a population. If your data were not collected randomly, your results might not represent the population.

Collect data using best practices
To ensure that your results are valid, consider the following guidelines:
  • Make certain that the data represent the population of interest.
  • Collect enough data to provide the necessary precision.
  • Measure variables as accurately and precisely as possible.
  • Record the data in the order it is collected.