Exam Study Notes on Sampling and Hypothesis Testing

Independent vs. Paired Sampling

  • Independent Sampling:

    • Two samples are taken from different populations; they do not need to be paired.
    • Sample sizes can vary (e.g., 50 from one group, 40 from another).
    • Main purpose: Compare means of two independent groups based on large sample sizes (30 or more).
    • Example: Compare sales between two different restaurants on the same night while controlling for exogenous variables like weather.

  • Paired Sampling:

    • Samples are matched or paired in a meaningful way (e.g. before and after measurements).
    • Each observation from one sample has a corresponding observation from another sample.
    • Ensures exogenous factors impact both samples equally.
    • Requires equal paired observations (e.g., comparing meal prices before and after a promotion at the same restaurant).
Key Concepts for Analysis

  • Mean Comparison: Using means for comparisons is crucial.

    • For independent sampling, calculate the mean for both groups separately and compare.
    • For paired sampling, determine the mean of the differences of paired observations.

  • Standard Deviation:

    • For two samples, use the combined standard deviation when calculating the test statistic.
    • Ensure to square the standard errors accurately to avoid calculation errors.
Confidence Intervals (CIs)
  • Construct a confidence interval to assess the difference between two sample means.
    • CI Formula: For independent samples, utilize the means and standard deviations of both samples to calculate.
    • If your CI contains zero, it suggests that there might be no significant difference between the groups.
    • Example: A CI of $(-641.6, 136.6)$ suggests inconclusive evidence regarding price differences between U.S. and Japanese automobiles.
Hypothesis Testing
  • Testing Differences in Means:
    • Common null hypothesis to test: H0: mu1 - mu2 = 0 against an alternative that suggests a difference.
    • For one-tailed tests (e.g., looking for a difference in one direction), adjust the critical values accordingly.
    • Test Statistic: Calculate the test statistic similarly for both independent and paired samples, adjusting formulas where necessary.
Practical Reminders
  • Always clearly outline your null and alternative hypotheses.
  • Maintain attention to detail throughout calculations, particularly for standard deviations and test statistics; order of operations is crucial.
  • Ensure to keep track of sample origins to avoid mix-ups. For instance, when comparing US and Japan prices, clearly denote their respective values.
  • Practice constructing confidence intervals and conducting hypothesis tests with different sample sizes and scenarios to solidify your understanding.