Hypothesis Testing
In engineering and quality control, hypothesis testing is a statistical method used to make data-driven decisions based on sample data. It provides a formal framework to determine if observed variations (e.g., a change in material strength or manufacturing dimensions) are statistically significant or just random noise. Key concepts in hypothesis testing include: 1. Null Hypothesis (H₀): The assumption of the status quo or "no effect" (e.g., "the new alloy has the same strength as the old one"). 2. Alternative Hypothesis (H₁): The claim you are trying to prove (e.g., "the new alloy is stronger"). It can be one-tailed (directional) or two-tailed (non-directional). 3. Type I Error (α): A false positive. Rejecting a true null hypothesis (e.g., stopping a perfectly good production line because a random sample looked out of spec). 4. Type II Error (β): A false negative. Failing to reject a false null hypothesis (e.g., accepting and shipping a batch of defective parts). 5. p-value: The probability of obtaining a test statistic at least as extreme as the one observed, assuming the null hypothesis is true. If the p-value is less than the chosen significance level (α), H₀ is rejected. 6. Power (1 - β): The probability that the test correctly detects a real effect or shift in the process.
Knowledge Check
1.A CNC machine is calibrated to cut shafts to a diameter of 100 mm (μ₀). A sample of 36 shafts yields a mean of 105 mm. The population standard deviation is known to be 12 mm. What is the calculated Z-statistic?
2.In the context of industrial quality control, what is the practical consequence of committing a Type I error (α)?
3.An engineer tests a new composite material, aiming to prove it has a higher tensile strength than the standard 500 MPa. The resulting p-value is 0.03. If the significance level α = 0.05, what is the correct statistical conclusion?
4.When testing the mean yield strength of an aluminum alloy using a small sample (n=12), the population standard deviation (σ) is unknown and is estimated using the sample standard deviation (s). Which probability distribution should be used to determine the critical value?
5.A sample of 25 resistors has an average resistance of 52 ohms and a sample standard deviation of 8 ohms. The target specification is 50 ohms. What is the calculated t-statistic?
6.An engineer wants to ensure that a statistical test has a high probability of successfully detecting a real, physical shift in manufacturing tolerances when one actually occurs. Which statistical metric are they trying to maximize?
7.A manufacturer claims a new motor design has a higher efficiency than the current 85% industry standard. To statistically validate this specific claim, how should the alternative hypothesis (H₁) be structured?
8.A two-tailed hypothesis test for a population mean is conducted at a 5% significance level (α = 0.05). The null hypothesis H₀: μ = 10 mm is rejected. What must mathematically be true about the corresponding 95% confidence interval for the mean?
9.Which of the following is the correct technical interpretation of a p-value in an engineering test?
10.Assuming the sample mean and standard deviation remain constant, how does quadrupling the sample size (e.g., from n=10 to n=40) affect the standard error and the magnitude of the test statistic?