What Two Assumptions Must Be Met When Using The Z Test

What Two Assumptions Must Be Met When You Are Using The Z Test To Test

When conducting a Z test to compare two means, there are two primary assumptions that must be satisfied to ensure the validity of the test results. First, the data should be drawn from populations that are normally distributed or that the sample sizes are sufficiently large, typically over 30, to invoke the Central Limit Theorem. This normality assumption ensures that the sampling distribution of the sample mean differences approximates a normal distribution, which is essential for applying the Z test. Second, the observations must be independent within and between groups; that is, the sampled individuals should not influence each other, and the groups should be independently selected. This independence assumption prevents bias and ensures the applicability of the statistical inference.

It is also important to note that when performing a Z test for the difference between two population means, the sample standard deviations (s₁ and s₂) cannot be used in place of the population standard deviations (σ₁ and σ₂) unless the population standard deviations are known. Typically, when the population standard deviations are unknown, a t-test is more appropriate. However, if σ₁ and σ₂ are known and the sample sizes are large enough, the Z test is valid. In practice, since population standard deviations are rarely known, the sample standard deviations are often used in conjunction with the Z test only when the sample sizes are large enough to ensure the sample standard deviations closely estimate the population parameters.

Testing the Difference in Social Security Benefits Between Retired and Disabled Workers

Suppose we want to test whether the average difference in monthly Social Security benefits between retired workers and disabled workers exceeds $30. To formulate this hypothesis test, we define the null hypothesis as H₀: μ₁ - μ₂ ≤ $30, meaning the difference in means is less than or equal to $30. The alternative hypothesis is H₁: μ₁ - μ₂ > $30, indicating that the difference exceeds $30. This setup corresponds to a right-tailed test at the 0.05 significance level.

The data needed include the sample means (x̄₁ and x̄₂), sample standard deviations (s₁ and s₂), and sample sizes (n₁ and n₂) for each group. From these, the test statistic is calculated as:

z = ( (x̄₁ - x̄₂) - $30 ) / √(s₁²/n₁ + s₂²/n₂)

where the numerator adjusts the sample difference by the hypothesized mean difference of $30, and the denominator represents the standard error of the difference in means. If the calculated z-value exceeds the critical value (approximately 1.645 for a 0.05 level in a right-tailed test), we reject the null hypothesis, concluding that the difference in benefits is statistically greater than $30. Otherwise, we fail to reject the null hypothesis, indicating insufficient evidence to support that the difference exceeds $30.

References

  • Weerahandi, S. (2014). Statistics for Experimenters: Design, Innovation, and Discovery. Springer.
  • Kutner, M. H., Nachtsheim, C. J., Neter, J., & Li, W. (2004). Applied Linear Statistical Models. McGraw-Hill Education.
  • Wilcox, R. R. (2012). Introduction to Robust Estimation and Hypothesis Testing. Academic Press.
  • Moore, D. S., Notz, W., & Frohlich, D. (2013). Statistics: Concepts and Contexts. W. H. Freeman and Company.
  • Freedman, D., Pisani, R., & Purves, R. (2007). Statistics (4th Ed.). W. W. Norton & Company.
  • DeGroot, M. H., & Schervish, M. J. (2012). Probability and Statistics. Pearson.
  • Ott, R. L., & Longnecker, M. (2010). An Introduction to Statistical Methods and Data Analysis. Brooks/Cole.
  • Lind, D. A., Marchal, W. G., & Wathen, S. A. (2010). Basic Statistics for Business and Economics. McGraw-Hill Education.
  • Hogg, R. V., Tanis, E. A., & Zimmerman, D. (2013). Probability and Statistical Inference. Pearson.
  • Schmidt, F. L., & Hunter, J. E. (2004). Methods of Meta-Analysis: Correcting for Publication Bias. Wiley.