🟪 1.

Introduction (5 marks)

Sample size estimation is a scientific process used in research to

determine the minimum number of participants or units required to
achieve accurate, reliable, and valid results while maintaining
statistical power and minimizing error.

🔹 It ensures:

 Efficiency of resources

 Valid statistical inference

 Reduced Type I & Type II errors

✏️Definition (APA Style):

“Sample size estimation refers to the process of calculating the minimum
number of observations required to detect an effect of a given size with a
given degree of confidence.” — (Polit & Beck, 2012)

🟪 2. Why Is Sample Size Estimation Important? (5 marks)

Use the mnemonic: "VAPES"

 V – Validity of results

 A – Avoids under/over-sampling

 P – Power maintained

 E – Ethical responsibility

 S – Saves time and resources

🟪 3. Key Concepts Before Estimation (10 marks)

Understanding these is crucial before estimating sample size:

Concept Description

Population size
Total number of individuals in the target group
(N)

Confidence level
Degree of certainty (commonly 95% or 99%)
(Z)
Concept Description

Margin of error
Precision of the estimate (e.g., ±5%)
(E)

Standard
Variability within the population
deviation (σ)

Ability to detect an effect if it exists (usually

Power (1−β)
80% or 90%)

Effect size (d) Magnitude of difference you're trying to detect

🧠 Memory trick: “ZEDS-P” — Z-score, Error, Deviation, Sample Size, Power

🟪 4. Sample Size Estimation Process (20 marks)

Break into 6 major steps using the mnemonic "DEEPER":

✅ D – Define research objective

 Type of study (e.g., prevalence, experimental, correlational)

 One-tailed or two-tailed test?

✅ E – Estimate effect size (d)

 Use previous research or pilot studies

 Cohen’s d conventions:

o 0.2 (small), 0.5 (medium), 0.8 (large)

✅ E – Estimate standard deviation (σ)

 Use known data or conduct pilot study

✅ P – Power level (1–β)

 Usually 0.80 (80%) or higher

 Power tables (e.g., Cohen’s Power Primer) or software (G*Power)

✅ E – Error margin and confidence level

 Confidence interval: Z = 1.96 (95%), Z = 2.58 (99%)

 Error margin (±5%) usually standard

✅ R – Run the calculation

 Use statistical formula or software

🟪 5. Common Formulae (10 marks)

For Proportions (Cochran’s formula):

n=Z2⋅p(1−p)e2n = \frac{Z^2 \cdot p(1-p)}{e^2}n=e2Z2⋅p(1−p)

Where:

 nnn = required sample size

 ZZZ = z-score (e.g., 1.96 for 95%)

 ppp = estimated proportion (often 0.5 if unknown)

 eee = margin of error

For Means:

n=(Z⋅σE)2n = \left(\frac{Z \cdot \sigma}{E}\right)^2n=(EZ⋅σ)2

 σ\sigmaσ = population SD

 EEE = margin of error

🧠 Memory trick: “Zp over e or Zσ over E squared!”

🟪 6. Adjustments

 Finite population correction if N is small:

nadj=n1+(n−1N)n_{adj} = \frac{n}{1 + \left(\frac{n - 1}{N}\right)}nadj

=1+(Nn−1)n

 Non-response rate or dropout rate:

nfinal=n(1−dropout%)n_{final} = \frac{n}{(1 - dropout\% )}nfinal

=(1−dropout%)n

🟪 7. Tools & Software

 G*Power
  Epi Info

 WHO Sample Size Calculators

 SPSS/Excel plugins

🟪 8. Conclusion

Sample size estimation is the backbone of any valid research design. A poor
estimation leads to type errors, invalid generalizations, and ethical
issues. Tools, theory, and pilot data make this process scientific and
replicable.

✅ QUESTION 2 (50 marks): What is Sample Size? What Are the Ways
to Estimate It?

🟪 1. Introduction: What is Sample Size? (5 marks)

Sample size is the number of observations or participants selected from

the population to participate in a study.

📘 According to Siegel & Castellan (1988):

“Sample size refers to the number of units selected from a population for
inclusion in a study.”

🟪 2. Why is Sample Size Important? (5 marks)

 Ensures representativeness

 Maintains power

 Reduces bias and error

 Allows valid inferences to population

🟪 3. Factors Affecting Sample Size (10 marks)

Use the mnemonic: “ZEPSI”

Factor Description

Z-score related to desired

Z – Confidence level
confidence

E – Margin of error Precision required

P – Population Standard deviation or

variability proportion

S – Sample type Probability vs non-probability

I – Intended power Higher power = bigger sample

🟪 4. Methods to Estimate Sample Size (20 marks)

A. Using Standard Formulae

As explained in Q1 (for mean/proportion)

B. Using Pilot Study

 Conduct a mini-study (N=30–50)

 Estimate variance, effect size

C. Using Previous Studies

 Borrow SD or p-values from published research

D. Using Rules of Thumb

 For regression: n≥50+8mn ≥ 50 + 8mn≥50+8m (m = predictors)

 For factor analysis: 5–10 cases per variable

E. Using Power Analysis

 With software like G*Power

 Choose:

o Effect size

o Alpha (α)

o Power (1–β)

o Type of test (t-test, ANOVA, etc.)

🟪 5. Sample Size Estimation for Different Designs (5 marks)

Study Type Suggested Method

Cross-
Cochran’s formula
sectional

Power analysis with expected effect

Clinical trial
size

Thematic saturation (10–30 interviews

Qualitative
typical)

Correlational Based on expected r values

🟪 6. Sample Size in Qualitative Research (5 marks)

📌 Not calculated statistically; based on saturation.

 Phenomenology: 5–10

 Grounded theory: 20–30

 Thematic analysis: ~15–25

 Case study: 1–10

🧠 Trick: “Small but deep” — qualitative = fewer participants, richer insights

🟪 7. Conclusion

Sample size and its estimation are central to the scientific rigor of
research. Various quantitative formulae, software tools, and pilot
methods help researchers make this choice appropriately. In qualitative
designs, saturation is the guiding principle rather than numbers

✅ UNIT X OVERVIEW: SAMPLE SIZE ESTIMATION

This unit focuses on how we determine the number of participants

needed in various statistical situations.
📚 POSSIBLE THEORY QUESTIONS

🟪 1. What is Sample Size? Why is it Important? (5–10 marks)

Key Points to Cover:

 Definition: Sample size = number of observations in a study

 Importance:

o Reduces bias

o Increases accuracy

o Ensures reliability

o Avoids under/over-sampling

 Linked with power, error margin, and confidence level

🟪 2. Factors Affecting Sample Size Estimation (10 marks)

List & explain:

1. Confidence Level (Z-value) – typically 95% (Z = 1.96)

2. Margin of Error (e) – smaller error = larger sample

3. Standard Deviation (σ) – more variability = larger sample

4. Effect Size – smaller effect = larger sample

5. Power of the study – usually 80% or 90%

6. Expected Proportion (p) – used in categorical data

7. Dropout rate – add 10–20% to the sample

🟪 3. Steps/Process of Sample Size Estimation (10–20 marks)

🪷 Step-by-step (Use this to write structured answers):

1. Define the research objective (mean/proportion/comparison)

2. Choose the acceptable margin of error (e)

3. Select desired confidence level (Z-score)

 4. Estimate population standard deviation (σ) or proportion (p)

5. Decide the statistical power

6. Use the appropriate formula or software

7. Adjust for non-response or dropout

🧠 Mnemonic: “R-E-C-A-P-S-A”
(Research – Error – Confidence – Assumption – Power – Sample size –
Adjustment)

🟪 4. Cochran’s Formula for Estimating a Proportion (5–10 marks)

n=Z2⋅p(1−p)e2n = \frac{Z^2 \cdot p(1 - p)}{e^2}n=e2Z2⋅p(1−p)

 Used when estimating proportions (e.g., % of smokers)

 Use p = 0.5 if no estimate is available

 Z = 1.96 for 95% confidence

 e = margin of error (e.g., 0.05)

🧠 Memory Trick: "Z squared p one minus p over e squared"

🟪 5. Formula for Estimating Sample Size for Mean (5–10 marks)

n=(Z⋅σE)2n = \left(\frac{Z \cdot \sigma}{E}\right)^2n=(EZ⋅σ)2

 For quantitative data (e.g., mean blood pressure)

 σ = population SD

 E = margin of error

🧠 Trick: “Z sigma over E squared”

🟪 6. Estimating Sample Size for Comparing Two Means (20–25 marks)

💡 When? → To test if two groups differ in average (e.g., CBT vs no CBT)

Formula:

n=2(Zα/2+Zβ)2⋅σ2(μ1−μ2)2n = \frac{2(Z_{\alpha/2} + Z_\beta)^2 \cdot \

sigma^2}{(μ_1 - μ_2)^2}n=(μ1−μ2)22(Zα/2+Zβ)2⋅σ2
Where:

Symb
Meaning
ol

Zα/2 Z-score for significance level

Zβ Z-score for power

σ² Pooled variance

Difference between means

μ₁ - μ₂
(effect size)

🧠 Trick: "Double Z squared sigma over difference squared!"

📌 Mention assumptions: normality, equal variances, independent samples.

🟪 7. Estimating Sample Size for Comparing Two Proportions (20–25

marks)

💡 When? → E.g., % of depression in males vs females

Formula:

n=(Zα/2+Zβ)2⋅[p1(1−p1)+p2(1−p2)](p1−p2)2n = \frac{(Z_{\alpha/2} + Z_\

beta)^2 \cdot [p_1(1 - p_1) + p_2(1 - p_2)]}{(p_1 - p_2)^2}n=(p1−p2
)2(Zα/2+Zβ)2⋅[p1(1−p1)+p2(1−p2)]

🧠 Trick: “Z combo squared times p1q1 plus p2q2 over difference

squared”

🟪 8. Short Note: Software Tools for Estimation (5 marks)

 G*Power

 OpenEpi

 Raosoft calculator

 EpiInfo

 Easy to use for complex tests

 Graphical interfaces
🟪 9. Short Note: Thumb Rules vs Formula-Based Estimation (5
marks)

Thumb Rules Formula-Based

General rules of Precise, based on

30–50/group math

Quick and easy More accurate

Not suitable for complex

Best for research
analysis

🟪 10. Short Note: Adjusting for Dropout/Attrition (5 marks)

 Add 10–20% more participants

 Especially for longitudinal studies

 Prevents underpowering due to loss