Formula Sheet

Antonio Marañon, Hanqing Wu

October 2022

Chapter 8
Confidence Interval for the Mean (𝜎 Known)
𝜎
𝑋̄ ± 𝑍𝛼/2 √
𝑛
or
𝜎 𝜎
𝑋̄ − 𝑍𝛼/2 √ ≤ 𝜇 ≤ 𝑋̄ + 𝑍𝛼/2 √
𝑛 𝑛
Confidence Interval for the Mean (𝜎 Unknown)

𝑆
𝑋̄ ± 𝑡𝛼/2 √
𝑛
or
𝑆 𝑆
𝑋̄ − 𝑡𝛼/2 √ ≤ 𝜇 ≤ 𝑋̄ + 𝑡𝛼/2 √
𝑛 𝑛
Confidence Interval Estimate for the Proportion

𝑝(1 − 𝑝)
𝑝 ± 𝑍𝛼/2 √
𝑛
or
𝑝(1 − 𝑝) 𝑝(1 − 𝑝)
𝑝 − 𝑍𝛼/2 √ ≤ 𝜋 ≤ 𝑝 + 𝑍𝛼/2 √
𝑛 𝑛
Sample Size Determination for the Mean
2
𝑍𝛼/2 𝜎2
𝑛=
𝑒2
Sample Size Determination for the Proportion
2
𝑍𝛼/2 𝜋(1 − 𝜋)
𝑛=
𝑒2

1
Chapter 9
𝑍 Test for the Mean (𝜎 Known)

𝑋̄ − 𝜇
𝑍STAT =
√𝜎
𝑛

𝑡 Test for the Mean (𝜎 Unknown)

𝑋̄ − 𝜇
𝑡STAT =
√𝑆
𝑛

𝑍 Test for the Proportion

𝑝−𝜋
𝑍STAT =
√ 𝜋(1−𝜋)
𝑛

𝑍 Test for the Proportion in Terms of the Number of Events of Interest

𝑋 − 𝑛𝜋
𝑍STAT =
√𝑛𝜋(1 − 𝜋)

Chapter 10
Pooled-Variance 𝑡 Test for the Difference Between Two Means
(𝑋̄ 1 − 𝑋̄ 2 ) − (𝜇1 − 𝜇2 )
𝑡STAT =
√𝑆𝑝2 ( 𝑛1 + 1
𝑛2 )
1

Separate-Variance 𝑡 Test for the Difference Between Two Means

(𝑋̄ 1 − 𝑋̄ 2 ) − (𝜇1 − 𝜇2 )
𝑡STAT = 2 𝑆22
√ 𝑆𝑛1 + 𝑛2
1

Confidence Interval Estimate for the Difference Between the Means of Two In-
dependent Populations

1 1
(𝑋̄ 1 − 𝑋̄ 2 ) ± 𝑡𝛼/2 √𝑆𝑝2 ( + )
𝑛1 𝑛2

or

1 1
(𝑋̄ 1 − 𝑋̄ 2 ) − 𝑡𝛼/2 √𝑆𝑝2 ( + ) ≤ 𝜇 1 − 𝜇2
𝑛1 𝑛2

1 1
≤ (𝑋̄ 1 − 𝑋̄ 2 ) + 𝑡𝛼/2 √𝑆𝑝2 ( + )
𝑛1 𝑛2

2
Paired 𝑡 Test for the Mean Difference
𝐷̄ − 𝜇𝐷
𝑡STAT = 𝑆
√𝐷
𝑛

Confidence Interval Estimate for the Mean Difference

𝑆
𝐷̄ ± 𝑡𝛼/2 √𝐷
𝑛
or
𝑆 𝑆
𝐷̄ − 𝑡𝛼/2 √𝐷 ≤ 𝜇𝐷 ≤ 𝐷̄ + 𝑡𝛼/2 √𝐷
𝑛 𝑛
𝑍 Test for the Difference Between Two Proportions

(𝑝1 − 𝑝2 ) − (𝜋1 − 𝜋2 )
𝑍STAT =
̄ − 𝑝)̄ ( 𝑛1 +
√𝑝(1 1
𝑛2 )
1

Confidence Interval Estimate for the Difference Between Two Proportions

𝑝1 (1 − 𝑝1 ) 𝑝2 (1 − 𝑝2 )
(𝑝1 − 𝑝2 ) ± 𝑍𝛼/2 √( + )
𝑛1 𝑛2

or
𝑝1 (1 − 𝑝1 ) 𝑝2 (1 − 𝑝2 )
(𝑝1 − 𝑝2 ) − 𝑍𝛼/2 √ + ≤ (𝜋1 − 𝜋2 )
𝑛1 𝑛2

𝑝1 (1 − 𝑝1 ) 𝑝2 (1 − 𝑝2 )
≤ (𝑝1 − 𝑝2 ) + 𝑍𝛼/2 √ +
𝑛1 𝑛2
𝐹 Test Statistic for Testing the Ratio of Two Variances

𝑆12
𝐹STAT =
𝑆22

Chapter 11
Total Variation in One-Way ANOVA

𝑐 𝑛𝑗
2
̄
SST = ∑ ∑ (𝑋𝑖𝑗 − 𝑋)
𝑗=1 𝑖=1

Among-Group Variation in One-Way ANOVA

𝑐 2
SSA = ∑ 𝑛𝑗 (𝑋̄ 𝑗 − 𝑋)
̄
𝑗=1

3
Within-Group Variation in One-Way ANOVA
𝑐 𝑛𝑗
2
SSW = ∑ ∑ (𝑋𝑖𝑗 − 𝑋̄ 𝑗 )
𝑗=1 𝑖=1

Mean Squares in One-Way ANOVA

SSA
MSA =
𝑐−1
SSW
MSW =
𝑛−𝑐
One-Way ANOVA 𝐹STAT Test Statistic

MSA
𝐹STAT =
MSW
Critical Range for the Tukey-Kramer Procedure

𝑀 𝑆𝑊 1 1
Critical range = 𝑄𝛼 √ ( + )
2 𝑛𝑗 𝑛𝑗′

Chapter 13
Simple Linear Regression Model

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋𝑖 + 𝜀i

Simple Linear Regression Equation: The Prediction Line

𝑌𝑖̂ = 𝑏0 + 𝑏1 𝑋𝑖

Computational Formula for the Slope, 𝑏1

𝑆𝑆𝑋𝑌
𝑏1 =
𝑆𝑆𝑋
Computational Formula for the 𝑌 Intercept, 𝑏0

𝑏0 = 𝑌 ̄ − 𝑏1 𝑋̄

Measures of Variation in Regression

SST = SSR + SSE

Total Sum of Squares (SST)

𝑛
2
SST = Total sum of squares = ∑ (𝑌𝑖 − 𝑌 ̄ )
𝑖=1

4
Regression Sum of Squares (SSR)
𝑛
2
SSR = Explained variation or regression sum of squares = ∑ (𝑌𝑖̂ − 𝑌 ̄ )
𝑖=1

Error Sum of Squares (SSE)

𝑛
2
SSE = Unexplained variation or error sum of squares = ∑ (𝑌𝑖 − 𝑌𝑖̂ )
𝑖=1

Coefficient of Determination
Regression sum of squares 𝑆𝑆𝑅
𝑟2 = =
Total sum of squares 𝑆𝑆𝑇
Computational Formula for SST
𝑛 𝑛 𝑛 2
2 (∑𝑖=1 𝑌𝑖 )
SST = ∑ (𝑌𝑖 − 𝑌 ̄ ) = ∑ 𝑌𝑖2 −
𝑖=1 𝑖=1
𝑛

Computational Formula for SSR

𝑛
2
SSR = ∑ (𝑌𝑖̂ − 𝑌 ̄ )
𝑖=1
𝑛 𝑛 𝑛 2
(∑𝑖=1 𝑌𝑖 )
= 𝑏0 ∑ 𝑌𝑖 + 𝑏1 ∑ 𝑋𝑖 𝑌𝑖 −
𝑖=1 𝑖=1
𝑛

Computational Formula for SSE

𝑛 𝑛 𝑛 𝑛
2
SSE = ∑ (𝑌𝑖 − 𝑌𝑖̂ ) = ∑ 𝑌𝑖2 − 𝑏0 ∑ 𝑌𝑖 − 𝑏1 ∑ 𝑋𝑖 𝑌𝑖
𝑖=1 𝑖=1 𝑖=1 𝑖=1

Standard Error of the Estimate

√ 𝑛 2
𝑆𝑆𝐸 √ ∑ (𝑌𝑖 − 𝑌𝑖̂ )
√ 𝑖=1
𝑆𝑌 𝑋 = =
𝑛−2 ⎷ 𝑛−2
Residual
𝑒𝑖 = 𝑌𝑖 − 𝑌𝑖̂
Durbin-Watson Statistic
𝑛 2
∑𝑖=2 (𝑒𝑖 − 𝑒𝑖−1 )
𝐷= 𝑛
∑𝑖=1 𝑒2𝑖

Testing a Hypothesis for a Population Slope, 𝛽1 , Using the 𝑡 Test

𝑏1 − 𝛽 1
𝑡STAT =
𝑆𝑏1

5
Testing a Hypothesis for a Population Slope, 𝛽1 , Using the 𝐹 Test
𝑀 𝑆𝑅
𝐹STAT =
𝑀 𝑆𝐸
Confidence Interval Estimate of the Slope, 𝛽1

𝑏1 ± 𝑡𝛼/2 𝑆𝑏1
𝑏1 − 𝑡𝛼/2 𝑆𝑏1 ≤ 𝛽1 ≤ 𝑏1 + 𝑡𝛼/2 𝑆𝑏1

Testing for the Existence of Correlation

𝑟−𝜌
𝑡STAT =
2
√ 1−𝑟
𝑛−2
cov(𝑋, 𝑌 )
𝑟=
𝑆𝑋 𝑆𝑌
Confidence Interval Estimate for the Mean of 𝑌
𝑌𝑖̂ ± 𝑡𝛼/2 𝑆𝑌 𝑋 √ℎ𝑖
𝑌𝑖̂ − 𝑡𝛼/2 𝑆𝑌 𝑋 √ℎ𝑖 ≤ 𝜇𝑌 ∣𝑋=𝑋𝑖 ≤ 𝑌𝑖̂ + 𝑡𝛼/2 𝑆𝑌 𝑋 √ℎ𝑖

Prediction Interval for an Individual Response, 𝑌

𝑌𝑖̂ ± 𝑡𝛼/2 𝑆𝑌 𝑋 √1 + ℎ𝑖
𝑌𝑖̂ − 𝑡𝛼/2 𝑆𝑌 𝑋 √1 + ℎ𝑖 ≤ 𝑌𝑋=𝑋𝑖 ≤ 𝑌𝑖̂ + 𝑡𝛼/2 𝑆𝑌 𝑋 √1 + ℎ𝑖

Chapter 14
Multiple Regression Model with 𝑘 Independent Variables

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋1𝑖 + 𝛽2 𝑋2𝑖 + 𝛽3 𝑋3𝑖 + … + 𝛽𝑘 𝑋𝑘𝑖 + 𝜀𝑖

Multiple Regression Model with Two Independent Variables

𝑌𝑖 = 𝛽0 + 𝛽1 𝑋1𝑖 + 𝛽2 𝑋2𝑖 + 𝜀𝑖

Multiple Regression Equation with Two Independent Variables

𝑌𝑖̂ = 𝑏0 + 𝑏1 𝑋1𝑖 + 𝑏2 𝑋2𝑖

Coefficient of Multiple Determination

Regression sum of squares SSR
𝑟2 = =
Total sum of squares SST

Adjusted 𝑟2
2 𝑛−1
𝑟adj = 1 − [(1 − 𝑟2 ) ]
𝑛−𝑘−1

6
Overall 𝐹 Test
MSR
𝐹STAT =
MSE
Testing for the Slope in Multiple Regression
𝑏𝑗 − 𝛽 𝑗
𝑡STAT =
𝑆𝑏𝑗

Confidence Interval Estimate for the Slope

𝑏𝑗 ± 𝑡𝛼/2 𝑆𝑏𝑗

Determining the Contribution of an Independent Variable to the Regression

Model
SSR (𝑋𝑗 ∣ All 𝑋s except 𝑗) = SSR(All 𝑋s) − SSR(All 𝑋s except j)
Contribution of Variable 𝑋1 , Given That 𝑋2 Has Been Included
SSR (𝑋1 ∣ 𝑋2 ) = SSR (𝑋1 and 𝑋2 ) − SSR (𝑋2 )
Contribution of Variable 𝑋2 , Given That 𝑋1 Has Been Included
SSR (𝑋2 ∣ 𝑋1 ) = SSR (𝑋1 and 𝑋2 ) − SSR (𝑋1 )
Partial 𝐹 Test Statistic
𝑆𝑆𝑅 (𝑋𝑗 ∣ All𝑋 s except 𝑗)
𝐹STAT =
𝑀 𝑆𝐸
Relationship Between a 𝑡 Statistic and an 𝐹 Statistic
𝑡2STAT = 𝐹STAT
Coefficients of Partial Determination for a Multiple Regression Model Contain-
ing Two Independent Variables
SSR (𝑋1 ∣ 𝑋2 )
𝑟𝑌2 1.2 =
𝑆𝑆𝑇 − 𝑆𝑆𝑅 (𝑋1 and 𝑋2 ) + 𝑆𝑆𝑅 (𝑋1 ∣ 𝑋2 )
and
SSR (𝑋2 ∣ 𝑋1 )
𝑟𝑌2 2.1 =
𝑆𝑆𝑇 − 𝑆𝑆𝑅 (𝑋1 and 𝑋2 ) + SSR (𝑋2 ∣ 𝑋1 )
Coefficient of Partial Determination for a Multiple Regression Model Containing
𝑘 Independent Variables
SSR (𝑋𝑗 ∣ All 𝑋s except 𝑗)
𝑟𝑌𝑗2(All variables except 𝑗) =
SST − SSR(All 𝑋s) + SSR (𝑋𝑗 ∣ All 𝑋s except 𝑗)
Odds Ratio
probability of an event of interest
Odds ratio =
1 − probability of an event of interest

7
Chapter 15
Quadratic Regression Model
2
𝑌𝑖 = 𝛽0 + 𝛽1 𝑋1𝑖 + 𝛽2 𝑋1𝑖 + 𝜀𝑖

Quadratic Regression Equation

𝑌𝑖̂ = 𝑏0 + 𝑏1 𝑋1𝑖 + 𝑏2 𝑋1𝑖

2

Regression Model with a Square-Root Transformation

√𝑌𝑖 = 𝛽0 + 𝛽1 𝑋1𝑖 + 𝜀𝑖

Original Multiplicative Model

𝛽 𝛽
𝑌𝑖 = 𝛽0 𝑋1𝑖1 𝑋2𝑖2 𝜀𝑖

Transformed Multiplicative Model

𝛽 𝛽
log 𝑌𝑖 = log (𝛽0 𝑋1𝑖1 𝑋2𝑖2 𝜀𝑖 )
𝛽 𝛽
= log 𝛽0 + log (𝑋1𝑖1 ) + log (𝑋2𝑖2 ) + log 𝜀𝑖
= log 𝛽0 + 𝛽1 log 𝑋1𝑖 + 𝛽2 log 𝑋2𝑖 + log 𝜀𝑖

Original Exponential Model

𝑌𝑖 = 𝑒𝛽0 +𝛽1 𝑋1𝑖 +𝛽2 𝑋2𝑖 𝜀𝑖

Transformed Exponential Model

ln 𝑌𝑖 = ln (𝑒𝛽0 +𝛽1 𝑋1𝑖 +𝛽2 𝑋2𝑖 𝜀𝑖 )

= ln (𝑒𝛽0 +𝛽1 𝑋1𝑖 +𝛽2 𝑋2𝑖 ) + ln 𝜀𝑖
= 𝛽0 + 𝛽1 𝑋1𝑖 + 𝛽2 𝑋2𝑖 + ln 𝜀𝑖

Variance Inflationary Factor

1
𝑉 𝐼𝐹𝑗 =
1 − 𝑅𝑗2

𝐶𝑝 Statistic
(1 − 𝑅𝑘2 ) (𝑛 − 𝑇 )
𝐶𝑝 = − [𝑛 − 2(𝑘 + 1)]
1 − 𝑅𝑇2