CHAPTER 6 Statistical Inference & Hypothesis Testing • 6.1 - One Sample Mean μ, Variance σ 2, Proportion π • 6.2 - Two Samples Means, Variances, Proportions μ1 vs. μ2 σ12 vs. σ22 π1 vs. π2 • 6.3 - Multiple Samples Means, Variances, μ1, …, μk σ12, …, σk2 Proportions π1, …, πk CHAPTER 6 Statistical Inference & Hypothesis Testing • 6.1 - One Sample Mean μ, Variance σ 2, Proportion π • 6.2 - Two Samples Means, Variances, Proportions μ1 vs. μ2 σ12 vs. σ22 π1 vs. π2 • 6.3 - Multiple Samples Means, Variances, μ1, …, μk σ12, …, σk2 Proportions π1, …, πk Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) σ2 Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X1 X2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 X 2 ~ N 2 , 2 n2 X1 X 2 ~ ???? Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 X1 X 2 ~ N ????, ???? Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 X1 X 2 ~ N 1 2 , ???? Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 X1 X 2 ~ N 1 2 , ???? Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 12 X 1 X 2 ~ N 1 2 , n 1 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 12 2 2 X1 X 2 ~ N 1 2 , n n 1 2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 X 1 X 2 ~ N 1 2 , 12 n1 22 n2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 μ0 σ2 (“No mean difference") Test at signif level α 1 2 Classic Example: “Randomized Clinical Trial”… Pop 1 = Treatment, Pop 2 = Control Random Sample, size n1 Random Sample, size n2 Sampling Distribution =? X 1 ~ N 1 , 1 n1 Recall from section 4.1 (Discrete Models): Mean(X – Y) = Mean(X) – Mean(Y) and if X and Y are independent… Var(X – Y) = Var(X) + Var(Y) X 2 ~ N 2 , 2 n2 = 0 under H0 2 2 1 X 1 X 2 ~ N 1 2 , 2 n n 1 2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 σ2 (“No mean difference") Test at signif level α 1 2 Null Distribution X 1 X 2 ~ N 0, But what if σ1 and σ2 are unknown? 2 Then use sample estimates s12 and s22 n1 n2 with Z- or t-test, if n and n are large. 1 2 2 1 s.e. X1 X 2 0 2 2 2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) σ1 POPULATION 2 X2 ~ N(μ2, σ2) Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 σ2 (“No mean difference") Test at signif level α 1 2 Null Distribution X 1 X 2 ~ N 0, s s2 n1 n2 2 1 2 s.e. X1 X 2 0 But what if σ12 and σ22 are unknown? Then use sample estimates s12 and s22 with Z- or t-test, if n1 and n2 are large. Later… (But what if n1 and n2 are small?) Example: X = “$ Cost of a certain medical service” Assume X is known to be normally distributed at each of k = 2 health care facilities (“groups”). Hospital: X1 ~ N(μ1, σ1) Clinic: X2 ~ N(μ2, σ2) • Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 (“No difference exists.") 2-sided test at significance level α = .05 • Data Sample 1: n1 = 137 Sample 2: n2 = 140 x1 630 x2 546 s12 788.5 s22 1663.0 4.2 Null Distribution X 1 X 2 N 0, N 0, 2 95% Confidence Interval for μ1 – μ2: (84 – 8.232, 84 + 8.232) = (75.768, 92.232) 788.5 1663.0 137 140 N 0, 4.2 0 95% Margin of Error = (1.96)(4.2) = 8.232 s s2 n1 n2 2 1 x1 x2 84 NOTE: >0 does not contain 0 84 0 = 20 >> 1.96 p << .05 Z-score = 4.2 Reject H0; extremely strong significant difference Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) 1 POPULATION 2 X2 ~ N(μ2, σ2) 2 1 Sample size n1 2 1 s (“No mean difference") Test at signif level α 2 X 1 X 2 ~ N 0, 12 n1 unknown 12 and 22 Sample size n2 Null Distribution 22 n2 Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 s22 large n1 and n2 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) 1 POPULATION 2 X2 ~ N(μ2, σ2) 2 1 Sample size n1 s12 (“No mean difference") Test at signif level α 2 X 1 X 2 ~ N 0, 2 n1 n2 2 1 unknown 12 and 22 Sample size n2 Null Distribution Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 small large n11 and n22 IF the two populations are equivariant, i.e., 2 s22 H0 : 2 1 2 2 then conduct a t-test on the “pooled” samples. H 0 : 12 2 2 H A: 2 2 1 s12 2 s22 H0 : 2 2 1 2 H 0 : 12 2 2 H A: 2 2 1 s12 2 s22 Test Statistic s12 F 2 s2 Sampling Distribution =? Working Rule of Thumb Acceptance Region for H0 ¼<F<4 Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) POPULATION 2 X2 ~ N(μ2, σ2) 1 2 1 (“No mean difference") Test at signif level α unknown 12 and 22 2 Null Distribution X 1 X 2 ~ N 0, 12 n1 small n1 and n2 22 n2 Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 IF equal variances H0 : 12 22 is accepted, then estimate their common value with a “pooled” sample variance. 2 pooled pooled s (n1 1) s12 (n2 1) s22 n1 n2 2 The pooled variance is a weighted average of s12 and s22, using the degrees of freedom as the weights. Consider two independent populations…and a random variable X, normally distributed in each. POPULATION 1 X1 ~ N(μ1, σ1) POPULATION 2 X2 ~ N(μ2, σ2) 1 2 1 12 n1 2 pooled2 pooled 1 s s.e. s IF equal variances (“No mean difference") Test at signif level α unknown 12 and 22 2 Null Distribution X 1 X 2 ~ N 0, n H0 : 12 22 small n1 and n2 22 n2 1spooled1 n1 n2n2 2 IF equal variances s then use Satterwaithe Test, Welch Test, etc. SEE LECTURE NOTES AND TEXTBOOK. H0 : 12 22 is accepted, then estimate their common value with a “pooled” sample variance. 2 pooled pooled is rejected, Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 (n1 1) s12 (n2 1) s22 n1 n2 2 The pooled variance is a weighted average of s12 and s22, using the degrees of freedom as the weights. Example: Y = “$ Cost of a certain medical service” Assume Y is known to be normally distributed at each of k = 2 health care facilities (“groups”). Hospital: Y1 ~ N(μ1, σ1) Clinic: Y2 ~ N(μ2, σ2) • Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 (“No difference exists.") 2-sided test at significance level α = .05 • Data: Sample 1 = {667, 653, 614, 612, 604}; n1 = 5 • Analysis via T-test (if equivariance holds): “Group Means” y1 “Group 2 Variances” s1 s2 = SS/df Pooled Variance 667 653 614 612 604 5 630))2 (604630)2 (667 630 51 SS1 630 Point estimates y2 788.5 s 2 2 df1 593 525 520 3 546 546))2 (520546 546))2 (593 546 31 y yi / n NOTE: y1 y2 84 >0 1663 2.11 4 1663 F 788.5 SS2 ( n11)( 1)788.5 s1 ()n 1)s1)( (3 2 2 2 1663 ) spooled (5 1080 1) n(5 1 n 2 2 (31) 2 Sample 2 = {593, 525, 520}; n2 = 3 2 df2 The pooled variance is a weighted average of the group variances, using the degrees of freedom as the weights. Example: Y = “$ Cost of a certain medical service” Assume Y is known to be normally distributed at each of k = 2 health care facilities (“groups”). Hospital: Y1 ~ N(μ1, σ1) Clinic: Y2 ~ N(μ2, σ2) • Null Hypothesis H0: μ1 = μ2, i.e., μ1 – μ2 = 0 (“No difference exists.") 2-sided test at significance level α = .05 • Data: Sample 1 = {667, 653, 614, 612, 604}; n1 = 5 Sample 2 = {593, 525, 520}; n2 = 3 • Analysis via T-test (if equivariance holds): Point estimates “Group Means” y1 “Group 2 Variances” s1 s2 = SS/df Pooled Variance 667 653 614 612 604 5 630))2 (604630)2 (667 630 51 630 788.5 s 2 2 546 546))2 (520546 546))2 (593 546 31 NOTE: y1 y2 84 >0 1663 2.11 4 1663 F 788.5 SS = 6480 ) (31)(1663) 2 spooled (51)( 788.5 1080 (51) (31) df = 6 Standard Error y2 593 525 520 3 y yi / n 11 1 1 s.e. 1080 24 s 5n1 3n2 2 pooled The pooled variance is a weighted average of the group variances, using the degrees of freedom as the weights. p-value = 2P(Y1 Y2 84) 2 P T6 84240 2 P T6 3.5 > 2 * (1 - pt(3.5, 6)) Reject H0 at α = .05 stat signif, Hosp > Clinic [1] 0.01282634 R code: > y1 = c(667, 653, 614, 612, 604) > y2 = c(593, 525, 520) > > t.test(y1, y2, var.equal = T) Formal Conclusion Two Sample t-test p-value < α = .05 Reject H0 at this level. data: y1 and y2 t = 3.5, df = 6, p-value = 0.01283 alternative hypothesis: true difference in means is not equal to 0 95 percent confidence interval: 25.27412 142.72588 Interpretation sample estimates: mean of x mean of y The samples provide evidence that the 630 546 difference between mean costs is (moderately) statistically significant, at the 5% level, with the hospital being higher than the clinic (by an average of $84). NEXT UP… PAIRED MEANS page 6.2-7, etc.