Formula Sheet for Statistics

Confidence Interval Hypothesis Testing

Single Mean 𝜎 𝑋̅ − 𝜇
𝜇 ∈ 𝑥̅ ± 𝑧𝛼⁄2 𝑍=
(𝜎 𝑘𝑛𝑜𝑤𝑛 √𝑛 𝜎/√𝑛
Single Mean 𝑠 ̅
𝑋−𝜇
𝜇 ∈ 𝑥̅ ± 𝑡𝛼⁄2 𝑇= 𝑤𝑖𝑡ℎ 𝑣 = 𝑛 − 1
(𝜎 𝑢𝑛𝑘𝑛𝑜𝑤𝑛) √𝑛 𝑠/√𝑛
Two means 𝜎12 𝜎22 𝑥̅1 − 𝑥̅2 − (𝜇1 − 𝜇2 )
𝜇1 − 𝜇2 ∈ (𝑥̅1 − 𝑥̅2 ) ± 𝑧𝛼⁄2 √ + 𝑍=
(𝜎 𝑘𝑛𝑜𝑤𝑛) 𝑛1 𝑛2 √𝜎12 /𝑛1 + 𝜎22 /𝑛2
𝑥̅1 − 𝑥̅2 − (𝜇1 − 𝜇2 )
1 1 𝑇=
Two Means 𝜇1 − 𝜇2 ∈ (𝑥̅1 − 𝑥̅2 ) ± 𝑡𝛼⁄2 𝑠𝑝 √ + 1 1
𝑛1 𝑛2 𝑠𝑝 √𝑛 + 𝑛
𝜎12 = 𝜎22 1 2
unknown (𝑛1 − 1)𝑆12 + (𝑛2 − 1)𝑆22
𝑆𝑝2 = 𝑣 = 𝑛1 + 𝑛2 − 2
𝑛1 + 𝑛2 − 2
𝑥̅1 − 𝑥̅2 − (𝜇1 − 𝜇2 )
𝑇=
Two Means √𝑠12 /𝑛1 + 𝑠22 /𝑛2
𝑠12 𝑠22
𝜎12 ≠ 𝜎22 𝜇1 − 𝜇2 ∈ (𝑥̅1 − 𝑥̅2 ) ± 𝑡𝛼⁄2 √ + (𝑠12 /𝑛1 + 𝑠22 /𝑛2 )2
unknown 𝑛1 𝑛2 𝑣= 2
(𝑠1 /𝑛1 )2 (𝑠22 /𝑛2 )2
𝑛1 − 1 + 𝑛2 − 1
(𝑛 − 1)𝑠 2 2
(𝑛 − 1)𝑠 2 (𝑛 − 1)𝑠 2
Variance < 𝜎 < 𝜒2 = 𝑤𝑖𝑡ℎ 𝑣 = 𝑛 − 1
𝜒𝛼2⁄2 2
𝜒1−𝛼 ⁄2 𝜎2
𝑠12 1 𝜎12 𝑠12 𝜎22 𝑠12
Two Variances < < 𝑓𝛼⁄2 (𝑣1 , 𝑣2 ) 𝑓(𝑣1 = 𝑛1 − 1, 𝑣2 = 𝑛2 − 1) = 2 2
𝑠22 𝑓𝛼⁄2 (𝑣1 , 𝑣2 ) 𝜎22 𝑠22 𝜎1 𝑠2
𝑝̂ − 𝑝0
Proportion 𝑝 ∈ 𝑝̂ ± 𝑧𝛼⁄2 √𝑝̂ 𝑞̂/𝑛 𝑧=
√𝑝0 𝑞0 ⁄𝑛
𝑝̂1 𝑞̂1 𝑝̂ 2 𝑞̂2 𝑝̂1 − 𝑝̂2
Two Proportions 𝑝1 − 𝑝2 ∈ (𝑝̂1 − 𝑝̂2 ) ± 𝑧𝛼⁄2 √ +
𝑛1 𝑛2 √𝑝̂ 𝑞̂(1⁄𝑛1 + 1⁄𝑛2 )
𝑧𝛼/2 𝜎 2 𝑧𝛼/2 2
Sample Size 𝑛=( ) for mean, n = 𝑝̂ (1 − 𝑝̂ ) ( ) for proportion
𝑒 𝑒
Prediction Intervals 𝑥0 ∈ 𝑥̅ ± 𝑧𝛼⁄2 𝜎√1 + 1⁄𝑛 (𝜎 𝑘𝑛𝑜𝑤𝑛) 𝑥0 ∈ 𝑥̅ ± 𝑡𝛼⁄2 𝑠√1 + 1⁄𝑛 (𝜎 unknown)
2
Test statistic (𝑜𝑖 − 𝑒𝑖 )
𝜒2 = ∑ 𝑤𝑖𝑡ℎ (𝑟 − 1) × ( 𝑐 − 1) 𝑑𝑒𝑔𝑟𝑒𝑒𝑠 𝑜𝑓 𝑓𝑟𝑒𝑒𝑑𝑜𝑚
(Goodness of Fit Tests) 𝑒𝑖
𝑖
𝑛 𝑛
(𝑋𝑖 − 𝑋̅)2 𝑋−𝜇
𝑋̅ = ∑ 𝑋𝑖 /𝑛 𝑆 =∑ 2 𝑍= where X = 𝑁(𝜇, 𝜎)
𝑛−1 𝜎
𝑖=1 𝑖=1
(𝑛 − 1)𝑆 2 𝑋̅−𝜇
Chi square RV with v=n-1: 𝜒 2 (𝑛 − 1) = T RV with v=n-1 T=
𝜎2 𝑠/√𝑛

F RV with v1=n1-1 and v2=n2-1 d.o.f 𝐹 = (𝜎22 𝑆12 )/(𝜎12 𝑆22 )

Regression
𝑆𝑥𝑦 = ∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ ) (𝑦𝑖 − 𝑦̅) 𝑆𝑦𝑦 = ∑𝑛𝑖=1(𝑦𝑖 − 𝑦̅)2 𝑆𝑥𝑥 = ∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ )2 𝑏0 = 𝑦̅ − 𝑏1 𝑥̅
SST = SSR + SSE 𝑆𝑆𝑇 = ∑𝑛𝑖=1(𝑦𝑖 − 𝑦̅)2 𝑆𝑆𝑅 = ∑𝑛𝑖=1(𝑦̂𝑖 − 𝑦̅)2 𝑏1 = 𝑆𝑥𝑦 /𝑆𝑥𝑥
𝑆𝑆𝐸 𝑆𝑆𝐸 𝑆𝑆𝑅 𝑠2 𝑠2 𝑋𝑖2
𝑆𝑆𝐸 = 𝑆𝑦𝑦 − 𝑏1 𝑆𝑥𝑦 𝑆2 = 𝑅 2 = 1 − 𝑆𝑆𝑇 , 𝑓 = 𝑠𝑏21 = 𝑠𝑏20 = ∑
𝑛 − (𝑘 + 1) 𝑠2 𝑆𝑥𝑥 𝑆𝑥𝑥 𝑛
𝑖