One-way ANOVA: A statistical method to compare means of three or more independent groups to determine if at least one differs significantly.
- State null and alternative hypotheses
- Verify ANOVA assumptions (normality, equal variance, independence)
- Calculate group means and overall mean
- Compute sum of squares between and within groups
- Calculate mean squares and F-statistic
- Determine critical value or p-value
- Make statistical decision
- Draw practical conclusion
H₀: μ_A = μ_B = μ_C (all formulations equally effective)
H₁: At least one mean differs
Formulation A: (15+18+12+16+14)/5 = 15
Formulation B: (22+25+20+23+21)/5 = 22.2
Formulation C: (19+17+21+18+20)/5 = 19
Overall mean: (15 + 22.2 + 19)/3 = 18.73
SSB = 5(15-18.73)² + 5(22.2-18.73)² + 5(19-18.73)²
SSB = 5(13.83) + 5(12.04) + 5(0.07) = 69.15 + 60.2 + 0.35 = 129.7
Formulation A: (15-15)² + (18-15)² + (12-15)² + (16-15)² + (14-15)² = 0 + 9 + 9 + 1 + 1 = 20
Formulation B: (22-22.2)² + (25-22.2)² + (20-22.2)² + (23-22.2)² + (21-22.2)² = 0.04 + 7.84 + 4.84 + 0.64 + 1.44 = 14.8
Formulation C: (19-19)² + (17-19)² + (21-19)² + (18-19)² + (20-19)² = 0 + 4 + 4 + 1 + 1 = 10
SSW = 20 + 14.8 + 10 = 44.8
df_between = k - 1 = 3 - 1 = 2
df_within = N - k = 15 - 3 = 12
MSB = SSB/df_between = 129.7/2 = 64.85
MSW = SSW/df_within = 44.8/12 = 3.73
F = MSB/MSW = 64.85/3.73 = 17.39
For α = 0.05, df = (2, 12)
F_critical = 3.89
Since F = 17.39 > 3.89, we reject H₀
We reject the null hypothesis. There are significant differences in blood pressure reduction effectiveness among the three drug formulations at α = 0.05.
• Decision rule: Reject H₀ if F > F_critical
• ANOVA assumptions: Verify normality and equal variance
• Statistical significance: Large F-value indicates differences
• Formulation B appears most effective (mean = 22.2)
• Significant differences justify further post-hoc analysis
• Results support drug development decisions
Educational ANOVA: Application of ANOVA to compare educational interventions, teaching methods, or curriculum effectiveness.
Traditional: (85+87+82+88+86+84)/6 = 85.33
Interactive: (90+92+88+91+89+90)/6 = 90.00
Digital: (88+91+89+93+87+92)/6 = 90.00
Overall mean: (85.33 + 90.00 + 90.00)/3 = 88.44
SSB = 6(85.33-88.44)² + 6(90.00-88.44)² + 6(90.00-88.44)²
SSB = 6(9.71) + 6(2.44) + 6(2.44) = 58.26 + 14.64 + 14.64 = 87.54
Traditional: (85-85.33)² + ... + (84-85.33)² = 23.33
Interactive: (90-90)² + ... + (90-90)² = 10.00
Digital: (88-90)² + ... + (92-90)² = 22.00
SSW = 23.33 + 10.00 + 22.00 = 55.33
df_between = 3 - 1 = 2
df_within = 18 - 3 = 15
MSB = 87.54/2 = 43.77
MSW = 55.33/15 = 3.69
F = 43.77/3.69 = 11.86
For α = 0.01, df = (2, 15)
F_critical = 6.36
Since F = 11.86 > 6.36, we reject H₀
There are significant differences in reading comprehension effectiveness among the three teaching methods.
We reject the null hypothesis. There are significant differences in reading comprehension effectiveness among the three teaching methods at α = 0.01. Both Interactive and Digital methods appear superior to Traditional.
• Stricter significance: α = 0.01 requires stronger evidence
• Practical implications: Results inform educational policy
• Sample size effect: Equal sample sizes simplify calculations
• Interactive and Digital methods show similar effectiveness
• Both outperform Traditional method significantly
• Supports modernization of teaching approaches
Industrial ANOVA: Application of ANOVA in manufacturing and quality control to compare process effectiveness, defect rates, or efficiency metrics.
Config 1: Mean = 2.4, n = 5
Config 2: Mean = 4.8, n = 5
Config 3: Mean = 3.0, n = 5
Overall mean: (2.4 + 4.8 + 3.0)/3 = 3.4
SSB = 5(2.4-3.4)² + 5(4.8-3.4)² + 5(3.0-3.4)²
SSB = 5(1) + 5(1.96) + 5(0.16) = 5 + 9.8 + 0.8 = 15.6
Config 1: (2-2.4)² + (3-2.4)² + (1-2.4)² + (4-2.4)² + (2-2.4)² = 0.16 + 0.36 + 1.96 + 2.56 + 0.16 = 5.2
Config 2: (5-4.8)² + (4-4.8)² + (6-4.8)² + (5-4.8)² + (4-4.8)² = 0.04 + 0.64 + 1.44 + 0.04 + 0.64 = 2.8
Config 3: (3-3.0)² + (2-3.0)² + (4-3.0)² + (3-3.0)² + (3-3.0)² = 0 + 1 + 1 + 0 + 0 = 2.0
SSW = 5.2 + 2.8 + 2.0 = 10.0
df_between = 3 - 1 = 2
df_within = 15 - 3 = 12
MSB = 15.6/2 = 7.8
MSW = 10.0/12 = 0.83
F = 7.8/0.83 = 9.40
For α = 0.05, df = (2, 12)
F_critical = 3.89
Since F = 9.40 > 3.89, we reject H₀
Configuration 2 has the highest defect rate (mean = 4.8)
Configuration 1 appears most effective (mean = 2.4)
Significant differences suggest process optimization opportunities
We reject the null hypothesis. There are significant differences in defect rates among the three assembly line configurations at α = 0.05. Configuration 1 is the most effective, while Configuration 2 has the highest defect rate.
• Quality control: Lower defect rates indicate better performance
• Cost implications: Statistical differences may have significant economic impact
• Process optimization: Results guide operational improvements
• Configuration 1 should be adopted company-wide
• Configuration 2 needs immediate process review
• Statistical evidence supports management decisions
Experimental design: The framework for organizing data collection to enable valid statistical analysis.
Factor: An independent variable in an experiment that is manipulated to observe its effect.
Level: The different values or categories of a factor.
Treatment: A specific combination of factor levels.
Block: A group of similar experimental units.
Randomization: Assigning treatments to experimental units randomly to reduce bias.
Replication: Repeating treatments to increase precision and estimate error.
Blocking: Grouping similar units to control for extraneous variation.
- Problem identification: Define the research question and objectives
- Factor selection: Identify independent variables to test
- Level determination: Specify treatment conditions
- Design selection: Choose appropriate experimental design
- Sample size planning: Determine adequate sample sizes
- Data collection: Execute experiment according to design
- Assumption verification: Check ANOVA assumptions
- ANOVA computation: Calculate all ANOVA components
- Statistical inference: Test hypotheses and make decisions
- Practical interpretation: Connect results to research objectives
- Implementation: Apply findings to real-world context
• Medical: Drug efficacy studies, treatment comparisons, clinical trials
• Psychology: Behavioral studies, therapy comparisons, cognitive experiments
• Agriculture: Crop variety testing, fertilizer comparisons, pest control methods
• Education: Teaching method evaluations, curriculum comparisons, assessment tools
• Industry: Quality control, process optimization, product testing
• Control for confounding variables
• Ensure randomization and blinding when appropriate
• Consider ethical implications in human studies
• Plan for adequate sample size and power
Agricultural ANOVA: Application of ANOVA in farming and agricultural research to compare crop yields, fertilizer effectiveness, or farming techniques.
Type A: (120+125+118+122)/4 = 121.25
Type B: (130+135+128+132)/4 = 131.25
Type C: (125+128+123+126)/4 = 125.5
Type D: (115+118+112+116)/4 = 115.25
Overall mean: (121.25 + 131.25 + 125.5 + 115.25)/4 = 123.31
SSB = 4(121.25-123.31)² + 4(131.25-123.31)² + 4(125.5-123.31)² + 4(115.25-123.31)²
SSB = 4(4.25) + 4(63.20) + 4(4.79) + 4(65.03) = 17.0 + 252.8 + 19.16 + 260.12 = 549.08
Type A: (120-121.25)² + (125-121.25)² + (118-121.25)² + (122-121.25)² = 1.56 + 14.06 + 10.56 + 0.56 = 26.74
Type B: (130-131.25)² + (135-131.25)² + (128-131.25)² + (132-131.25)² = 1.56 + 14.06 + 10.56 + 0.56 = 26.74
Type C: (125-125.5)² + (128-125.5)² + (123-125.5)² + (126-125.5)² = 0.25 + 6.25 + 6.25 + 0.25 = 13.0
Type D: (115-115.25)² + (118-115.25)² + (112-115.25)² + (116-115.25)² = 0.06 + 7.56 + 10.56 + 0.56 = 18.74
SSW = 26.74 + 26.74 + 13.0 + 18.74 = 85.22
df_between = 4 - 1 = 3
df_within = 16 - 4 = 12
MSB = 549.08/3 = 183.03
MSW = 85.22/12 = 7.10
F = 183.03/7.10 = 25.78
For α = 0.05, df = (3, 12)
F_critical = 3.49
Since F = 25.78 > 3.49, we reject H₀
Type B fertilizer shows highest yield (mean = 131.25)
Type D shows lowest yield (mean = 115.25)
Significant differences justify changing fertilizer practices
We reject the null hypothesis. There are significant differences in corn yield among the four fertilizer types at α = 0.05. Type B fertilizer appears most effective while Type D is least effective.
• Agricultural significance: Yield differences translate to economic impact
• Decision support: Results guide farming practices
• Replication: Multiple plots per treatment increase reliability
• Type B fertilizer increases yield by ~10 units
• Economic analysis should follow statistical results
• Consider environmental impact of most effective fertilizer
Medical ANOVA: Application of ANOVA in clinical research to compare treatment effectiveness, drug efficacy, or therapeutic outcomes.
Medication 1: (15+18+16+14+17)/5 = 16
Medication 2: (22+20+24+21+23)/5 = 22
Medication 3: (19+17+20+18+16)/5 = 18
Overall mean: (16 + 22 + 18)/3 = 18.67
SSB = 5(16-18.67)² + 5(22-18.67)² + 5(18-18.67)²
SSB = 5(7.13) + 5(11.11) + 5(0.45) = 35.65 + 55.55 + 2.25 = 93.45
Med 1: (15-16)² + (18-16)² + (16-16)² + (14-16)² + (17-16)² = 1 + 4 + 0 + 4 + 1 = 10
Med 2: (22-22)² + (20-22)² + (24-22)² + (21-22)² + (23-22)² = 0 + 4 + 4 + 1 + 1 = 10
Med 3: (19-18)² + (17-18)² + (20-18)² + (18-18)² + (16-18)² = 1 + 1 + 4 + 0 + 4 = 10
SSW = 10 + 10 + 10 = 30
df_between = 3 - 1 = 2
df_within = 15 - 3 = 12
MSB = 93.45/2 = 46.73
MSW = 30/12 = 2.5
F = 46.73/2.5 = 18.69
For α = 0.01, df = (2, 12)
F_critical = 6.93
Since F = 18.69 > 6.93, we reject H₀
Medication 2 shows greatest cholesterol reduction (mean = 22)
Results are statistically significant at α = 0.01
Findings support clinical decision-making
We reject the null hypothesis. There are significant differences in cholesterol reduction among the three medications at α = 0.01. Medication 2 appears most effective for cholesterol reduction.
• Medical significance: Results inform patient care decisions
• Stricter significance: α = 0.01 reflects importance of medical decisions
• Regulatory implications: Results may affect drug approval
• Medication 2 reduces cholesterol by 6 units more than average
• Statistical significance supports clinical relevance
• Further research needed for safety profile
Effect size: A quantitative measure of the magnitude of the experimental effect, independent of sample size.
Power: The probability of correctly rejecting a false null hypothesis.
Randomized block design: An experimental design that controls for variability by grouping similar experimental units.
Factorial design: An experimental design that tests multiple factors simultaneously.
Interaction effect: When the effect of one factor depends on the level of another factor.
Main effect: The effect of a factor averaged over all levels of other factors.
- Design planning: Select appropriate experimental design for the research question
- Sample size calculation: Determine adequate sample size for desired power
- Randomization scheme: Plan for proper randomization to avoid bias
- Data collection protocol: Establish procedures to ensure data quality
- Assumption verification: Check all ANOVA assumptions before analysis
- Statistical analysis: Perform ANOVA and post-hoc tests if needed
- Effect size calculation: Determine practical significance
- Power analysis: Evaluate the study's ability to detect effects
- Practical interpretation: Connect statistical results to real-world implications
- Report writing: Communicate results clearly and completely
• Human subjects: Obtain proper consent and protect privacy
• Animal studies: Follow ethical guidelines for animal welfare
• Environmental impact: Consider ecological consequences of applications
• Equity: Ensure research benefits are accessible to all populations
• Transparency: Report all relevant details and limitations
• Repeated measures designs for longitudinal studies
• Nested designs for hierarchical data structures
• Split-plot designs for complex experimental constraints
• Response surface methodology for optimization