Transcript Chapter 8
Section 8.5 Testing a claim about a mean (σ unknown) Objective For a population with mean µ (with σ unknown), use a sample to test a claim about the mean. Testing a mean (when σ known) uses the t-distribution 1 Notation 2 Requirements (1) The population standard deviation σ is unknown (2) One or both of the following: The population is normally distributed or The sample size n > 30 3 Test Statistic Denoted t (as in t-score) since the test uses the t-distribution. 4 Example 1 People have died in boat accidents because an obsolete estimate of the mean weight (of 166.3 lb.) was used. A random sample of n = 40 men yielded the mean x = 172.55 lb. and standard deviation s = 26.33 lb. Do not assume the population standard deviation is known. Test the claim that men have a mean weight greater than 166.3 lb. using 90% confidence. What we know: µ0 = 166.3 n = 40 x = 172.55 Claim: µ > 166.3 using α = 0.1 s = 26.33 Note: Conditions for performing test are satisfied since n >30 5 Example 1 Using Critical Regions What we know: µ0 = 166.3 n = 40 x = 172.55 Claim: µ > 166.3 using α = 0.1 s = 26.33 H0 : µ = 166.3 H1 : µ > 166.3 right-tailed test Test statistic: tα = 1.304 t = 1.501 Critical value: (df = 39) t in critical region Initial Conclusion: Since t in critical region, Reject H0 Final Conclusion: Accept the claim that the mean weight is greater than 166.3 lb. 6 Calculating P-value for a Mean (σ unknown) Stat → T statistics → One sample → with summary 7 Calculating P-value for a Mean (σ unknown) Enter the Sample mean (x) Sample std. dev. (s) Sample size (n) Then hit Next 8 Calculating P-value for a Mean (σ unknown) Select Enter the Select Hypothesis Test Null:mean (µ0) Alternative (“<“, “>”, or “≠”) Then hit Calculate 9 Calculating P-value for a Mean (σ unknown) The resulting table shows both the test statistic (t) and the P-value Test statistic (t) P-value 10 Example 1 Using the P-value What we know: H0 : µ = 166.3 H1 : µ > 166.3 Using StatCrunch µ0 = 166.3 n = 40 x = 172.55 Claim: µ > 166.3 using α = 0.1 s = 26.33 Stat → T statistics→ One sample → With summary Sample mean: 172.55 Sample std. dev.: 37.8 Sample size: 40 ● Hypothesis Test Null: proportion= Alternative 166.3 > P-value = 0.0707 Initial Conclusion: Since P-value < α, Reject H0 Final Conclusion: Accept the claim that the mean weight is greater than 166.3 lb. 11 P-Values A useful interpretation of the P-value: it is observed level of significance Thus, the value 1 – P-value is interpreted as observed level of confidence Recall: “Confidence Level” = 1 – “Significance Level” Note: Only useful if we reject H0 If H0 accepted, the observed significance and confidence are not useful. 12 P-Values From Example 1: P-value = 0.0707 1 – P-value = 0.9293 Thus, we can say conclude the following: The claim holds under 0.0707 significance. or equivalently… We are 92.93% confident the claim holds 13 Example 2 Loaded Die When a fair die (with equally likely outcomes 1-6) is rolled many times, the mean valued rolled should be 3.5 Your suspicious a die being used at a casino is loaded (that is, it’s mean is a value other than 3.5) You record the values for 100 rolls and end up with a mean of 3.87 and standard deviation 1.31 Using a confidence level of 99%, does the claim that the dice are loaded? What we know: µ0 = 3.5 n = 100 x = 3.87 Claim: µ ≠ 3.5 using α = 0.01 s = 1.31 Note: Conditions for performing test are satisfied since n >30 14 Example 2 Using Critical Regions What we know: µ0 = 3.5 n = 100 Claim: µ ≠ 3.5 using x = 3.87 α = 0.01 s = 1.31 H0 : µ = 3.5 H1 : µ ≠ 3.5 two-tailed test Test statistic: zα = -2.626 zα = 2.626 z = 3.058 Critical value: (df = 99) t in critical region Initial Conclusion: Since P-value < α, Reject H0 Final Conclusion: Accept the claim the die is loaded. 15 Example 2 Using the P-value What we know: H0 : µ = 3.5 H1 : µ ≠ 3.5 Using StatCrunch µ0 = 3.5 n = 100 Claim: µ ≠ 3.5 using x = 3.87 α = 0.01 s = 1.31 Stat → T statistics→ One sample → With summary ● Hypothesis Test Sample mean: 3.87 Sample std. dev.: 1.31 Null: proportion= Sample size: 100 Alternative 3.5 ≠ P-value = 0.0057 Initial Conclusion: Since P-value < α, Reject H0 Final Conclusion: Accept the claim the die is loaded. We are 99.43% confidence the die are loaded 16 17 Section 8.6 Testing a claim about a standard deviation Objective For a population with standard deviation σ, use a sample too test a claim about the standard deviation. Tests of a standard deviation use the c2-distribution 18 Notation 19 Notation 20 Requirements (1) The sample is a simple random sample (2) The population is normally distributed Very strict condition!!! 21 Test Statistic Denoted c2 (as in c2-score) since the test uses the c2 -distribution. n Sample size s Sample standard deviation σ0 Claimed standard deviation 22 Critical Values Right-tailed test “>“ Needs one critical value (right tail) Use StatCrunch: Chi-Squared Calculator 23 Critical Values Left-tailed test “<” Needs one critical value (left tail) Use StatCrunch: Chi-Squared Calculator 24 Critical Values Two-tailed test “≠“ Needs two critical values (right and left tail) Use StatCrunch: Chi-Squared Calculator 25 Example 1 Problem 14, pg 449 Statisics Test Scores Tests scores in the author’s previous statistic classes have followed a normal distribution with a standard deviation equal to 14.1. His current class has 27 tests scores with a standard deviation of 9.3. Use a 0.01 significance level to test the claim that this class has less variation than the past classes. What we know: σ0 = 14.1 n = 27 Claim: σ < 14.1 s = 9.3 using α = 0.01 Note: Test conditions are satisfied since population is normally distributed 26 Example 1 Using Critical Regions What we know: σ0 = 14.1 n = 27 s = 9.3 Claim: σ < 14.1 using α = 0.01 H0 : σ = 14.1 H1 : σ < 14.1 Left-tailed Test statistic: c2L = 12.20 Critical value: (df = 26) c2 = 11.31 c2 in critical region Initial Conclusion: Since c2 in critical region, Reject H0 Final Conclusion: Accept the claim that the new class has less variance than the past classes 27 Calculating P-value for a Variance Stat → Variance → One sample → with summary 28 Calculating P-value for a Variance Enter the Sample variance (s2) Sample size (n) NOTE: Must use Variance s2 = 9.32 = 86.49 Then hit Next 29 Calculating P-value for a Variance Select Enter the Select Hypothesis Test Null:variance (σ02) Alternative (“<“, “>”, or “≠”) σ02 = 14.12 = 198.81 Then hit Calculate 30 Calculating P-value for a Variance The resulting table shows both the test statistic (c2) and the P-value Test statistic (c2) P-value 31 Example 1 Using Critical Regions What we know: H0 : σ2 = 198.81 H1 : σ2 < 198.81 Using StatCrunch s2 = 86.49 σ02 = 198.81 σ0 = 14.1 n = 27 s = 9.3 Claim: σ < 14.1 using α = 0.01 Stat → Variance → One sample → With summary Sample variance: Sample size: 86.49 27 ● Hypothesis Test Null: proportion= Alternative 198.81 < P-value = 0.0056 Initial Conclusion: Since P-value < α (α = 0.01), Reject H0 Final Conclusion: Accept the claim that the new class has less variance than the past classes We are 99.44% confident the claim holds 32 Example 2 Problem 17, pg 449 BMI for Miss America Listed below are body mass indexes (BMI) for recent Miss America winners. In the 1920s and 1930s, distribution of the BMIs formed a normal distribution with a standard deviation of 1.34. Use a 0.01 significance level to test the claim that recent Miss America winners appear to have variation that is different from that of the 1920s and 1930s. 19.5 20.3 19.6 20.2 17.8 17.9 19.1 18.8 17.6 16.8 Using StatCrunch: s = 1.1862172 What we know: σ0 = 1.34 n = 10 Claim: σ ≠ 1.34 s = 1.186 using α = 0.01 Note: Test conditions are satisfied since population is normally distributed 33 Example 2 Using Critical Regions What we know: σ0 = 1.34 n = 10 Claim: σ ≠ 1.34 s = 1.186 using α = 0.01 H0 : σ = 1.34 H1 : σ ≠ 1.34 two-tailed 0.005 Test statistic: c2L = 2.088 Critical values: (df = 26) c2R = 26.67 c2 = 7.053 c2 not in critical region Initial Conclusion: Since c2 not in critical region, Accept H0 Final Conclusion: Reject the claim recent winners have a different variations than in the 20s and 30s Since H0 accepted, the observed significance isn’t useful. 34 Example 2 Using P-value What we know: σ0 = 1.34 n = 10 Claim: σ ≠ 1.34 H0 : σ2 = 1.796 H1 : σ2 < 1.796 Using StatCrunch s2 = 1.407 σ02 = 1.796 s = 1.186 using α = 0.01 Stat → Variance → One sample → With summary Sample variance: Sample size: 1.407 10 ● Hypothesis Test Null: proportion= 1.796 Alternative < P-value = 0.509 Initial Conclusion: Since P-value ≥ α (α = 0.01), Accept H0 Final Conclusion: Reject the claim recent winners have a different variations than in the 20s and 30s Since H0 accepted, the observed significance isn’t useful. 35