Situation Confidence Interval 1 mean ( known) x z* Significance Test Conditions SRS Ho: µ = µo n Z x o n and Population is stated normal, or Large sample size (n ≥ 30), or Approx. Linear Norm Prob Plot and 1 mean ( unknown) xt * or s n t 2 dependent means (matched pairs) Ho: µ = 0 df = n-1 Population 10n SRS and x o s n population is normal, or large sample size (n ≥ 30), or moderate sample size (15≤ n≤ 30) with moderate skewness and no outliers, or small sample size (n< 15) with little skewness and no outliers. Two independent SRSs and 2 independent means ( known) x x z 1 * 2 2 1 n1 2 2 Ho: µ1 = µ2 n2 z x1 x 2 12 n1 22 n2 population is normal, or large sample size (n1,n2 ≥ 30), or moderate sample size (15≤ both n1 & n2 ≤ 30) with moderate skewness and no outliers, or small sample size (n1 and n2< 15) with little skewness and no outliers 2 independent means ( unknown) (x1 – x 2 ) t * 2 1 2 2 s s n1 n2 (use df smaller of n1–1 and n2–1) Ho: µ1 = µ2 x –x t 12 2 2 s1 s2 n1 n2 Two independent SRSs and population is normal, or large sample size (n1,n2 ≥ 30), or moderate sample size (15≤ both n1 & n2 ≤ 30) with moderate skewness and no outliers, or small sample size (n1 and n2< 15) with little skewness and no outliers 1 proportion pˆ z * 1. Population 10n 2. SRS 3. TOS: npo 10 Ho : p po pˆ (1 – pˆ ) n z p po po 1 po n CI: n1 po 10 n p 10 n1 p 10 1. Two independent SRS’s Ho : p1 p 2 pˆ1 pˆ 2 z* 2 proportions 2. Population 10n1, Population 10n2 n1 pˆ 5,n1 (1 pˆ ) 5 n2 pˆ 5,n2 (1 pˆ ) 5 p1 p2 p1 p p1 p where p x1 x 2 n1 n 2 n1 n2 n1 pˆ1 5,n1 (1 pˆ1 ) 5 CI: n2 pˆ 2 5,n2 (1 pˆ 2 ) 5 pˆ1 (1 pˆ1 ) pˆ 2 (1 pˆ 2 ) z n1 n 2 Chi-Square Ho: 3. TOS: There is no association between the row vars. and column vars. in the table. Independence: Two-Way Table 2 O E 1. SRS 2. All expected counts ≥ 1 3. No more than 20% of exp. counts are less than 5. 2 E df = (r–1)(c–1) r = rows, c = columns E Chi-Square Ho: Goodness of Fit: One-Way Table rowtot coltot tabletot There is no difference between the distributions. (….is the same as…) 2 O E 1. SRS 2. All expected counts ≥ 1 3. No more than 20% of exp. counts are less than 5. 2 E df = n-1 Regression slope (µy = Confidence Interval for the slope : + ßx) b t *SE b` with n-2 df 1 SE b 1. Mean response b t SE b relationship with x 2. For each value of x: response y varies normally repeated responses y are indep.1 stdev (unknown) about the true reg. line is constant for all x s x x 2 Repeated observations on the same individual are not allowed. y Ho: ß = 0 has a straight line