Scribd
Upload
2 views15 pages

Statistical Sample Size Calculations

The document outlines various statistical formulas and exercises for determining sample sizes for different types of tests, including one-sample mean tests, two-sample t-tests, paired t-tests, ANOVA, proportion tests, chi-square tests, correlation, regression, logistic regression, survival analysis, factor analysis, and surveys. Each section provides a formula, context for use, and an example exercise with the corresponding answer. The document serves as a comprehensive guide for researchers to estimate necessary sample sizes based on specific parameters and desired confidence levels.

Uploaded by

Eng. George Sarkiss
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
2 views15 pages

Statistical Sample Size Calculations

The document outlines various statistical formulas and exercises for determining sample sizes for different types of tests, including one-sample mean tests, two-sample t-tests, paired t-tests, ANOVA, proportion tests, chi-square tests, correlation, regression, logistic regression, survival analysis, factor analysis, and surveys. Each section provides a formula, context for use, and an example exercise with the corresponding answer. The document serves as a comprehensive guide for researchers to estimate necessary sample sizes based on specific parameters and desired confidence levels.

Uploaded by

Eng. George Sarkiss
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd

 Formula:

n = ((Zα/2 × σ) / E)²
 Used to estimate the
population mean when σ is

ONE- 
known.
E = margin of error, Z = z-

SAMPLE 
value for confidence level.
Exercise: Estimate how

MEAN TEST many samples are needed


to estimate the average
blood pressure (σ=6)
within ±2 units at 95%
confidence.
Answer: n = 35
 Formula:
n = 2((Z1−α/2 + Z1−β) ×
σ / Δ)²
TWO-SAMPLE  Compares means of two
independent groups.
T-TEST  Needs desired power
(INDEPENDENT (1−β) and detectable
difference (Δ).
)  Exercise: σ=15, Δ=10,
α=0.05, power=0.8.
Answer: n ≈ 36 per group.
 Formula:
n = ((Z1−α/2 + Z1−β) ×
σd / Δd)²

PAIRED T-  Used for before/after or


matched designs; σd =
TEST 
std. dev. of differences.
Exercise: σd=10, Δd=8,
α=0.05, power=0.8.
Answer: n ≈ 25.
 Formula:
n = 2(Z1−α/2 + Z1−β)² /
ONE-WAY (k × f²)
 Tests mean differences
ANOVA (K among ≥3 groups. Effect
size f from Cohen’s f.
GROUPS)  Exercise: k=3, f=0.25,
α=0.05, power=0.8.
Answer: n ≈ 52 per group.
 Formula:
n = Z²α/2 × p(1−p) / E²
ONE-  Estimates a population
proportion (e.g., % of
PROPORTIO people vaccinated).

N TEST  Exercise: p=0.4, E=0.05,


95% CI.
Answer: n ≈ 369.
 Formula:
n = ((Z1−α/2√2p(1−p) +
Z1−β√(p1(1−p1)+p2(1−p
TWO- 2)))²) / (p1−p2)²
 Compares proportions
PROPORTIO between two groups.

N TEST  Exercise: p1=0.3,


p2=0.5, α=0.05,
power=0.8.
Answer: n ≈ 88 per group.
 Formula:
n = χ²crit / w²

CHI-
 Used for testing
association between
categorical variables.
SQUARE  Cohen’s w: 0.1=small,

TEST
0.3=medium, 0.5=large.
 Exercise: w=0.3, df=2,
α=0.05.
Answer: n ≈ 67.
 Formula:
n = ((Z1−α/2 + Z1−β)²) /
(0.5[ln((1+r)/(1−r))]²)
CORRELATIO  Tests if a linear
N (PEARSON relationship exists
between two variables.
R)  Exercise: r=0.3, α=0.05,
power=0.8.
Answer: n ≈ 84.
 Formula:
n = ((Z1−β + Z1−α/2)²
SIMPLE (1−R²)) / R² + 2
 Determines sample size
REGRESSIO to detect an effect in a
simple regression.
N  Exercise: R²=0.25,
α=0.05, power=0.8.
Answer: n ≈ 30.
 Formula:
n = ((Z1−β + Z1−α/2)²) /
MULTIPLE f² + k + 1
 Used for multiple
REGRESSIO predictors; rule of thumb
n ≥ 50 + 8k.
N  Exercise: f²=0.15, k=5,
α=0.05, power=0.8.
Answer: n ≈ 92.
 Rule of Thumb:
 At least 10 events per
LOGISTIC 
predictor variable.
Used for binary
REGRESSIO outcomes (e.g., disease
yes/no).
N  Exercise: Event
rate=0.3, predictors=8.
Answer: n ≈ 267.
 Formula:

SURVIVAL n = ((Z1−α/2 + Z1−β)² /


(ln HR)²) × E

ANALYSIS  Estimates required


events (E = event rate
(COX 
× n) for survival studies.
Exercise: HR=0.7, event
MODEL) rate=0.6.
Answer: ≈120 events.
 Rule of Thumb:
FACTOR  5–10 subjects per item.

ANALYSIS /  Ensures reliable factor


loadings.

PCA  Exercise: 20 items → n =


100–200.
 Formula:
n = Z²p(1−p) / E²
SURVEYS  Used when population
size unknown; apply
(UNKNOWN correction if finite.

POPULATION)  Exercise: 95% CI,


margin=5%.
Answer: n ≈ 384.
 Guideline:
 10–30 participants.
PILOT  Used to estimate σ or p
for planning the main
STUDY 
study.
Exercise: Choose n=20
participants.

You might also like