I’ll explain BS602 – Lecture 6 in very simple, exam-friendly language, like you’re
revising the night before the test. No heavy math, just what it means and how to
think about it.
1. Normal Distribution (Bell-shaped Curve)
• Many biological and medical measurements follow this shape
(height, weight, BP, lab values).
• Most values are in the middle, few are at the extremes.
Standard Normal Distribution
This is a special normal distribution where:
• Mean = 0
• Standard deviation (SD) = 1
Because of this, all values are written as Z-scores.
2. Z-Score (Very Important )
What is a Z-score?
A z-score tells you how far a value is from the mean, measured in standard
deviations.
Formula (don’t panic ):
Z=x−meanSDZ = \frac{x - \text{mean}}{\text{SD}}Z=SDx−mean
Interpretation (EXAM GOLD):
• Z = 0 → exactly at the mean
• Positive Z → above the mean
• Negative Z → below the mean
Simple Examples:
• Z = +2 → 2 SD above average
• Z = −2 → 2 SD below average (often abnormal)
3. Clinical Meaning of Z-Scores
Pediatric weight example
• Child weight far below mean → negative Z-score
• Large negative Z-scores (≤ −2) → malnutrition risk
Without z-scores, a child might be wrongly called “small but normal”.
4. Percentiles (Easy Concept )
What is a percentile?
It tells you what percentage of people are below a value.
• 50th percentile → median
• 10th percentile → 10% below, 90% above
• 90th percentile → 90% below
Example:
• Child at 5th percentile:
o Taller than only 5% of children
o Shorter than 95%
5. Z-Scores and Percentiles (Must-Know Table)
Z-score Percentile
0 50th
+1 84th
+2 97.7th
−1 16th
−2 2.3rd
Negative Z = low percentile
Positive Z = high percentile
6. Cumulative Probability (Think “Area Under the
Curve”)
Meaning:
Probability that a value is ≤ a given number
P(X≤x)P(X \le x)P(X≤x)
• Cumulative probability = percentile ÷ 100
Examples:
• 0.84 → 84th percentile
• 0.50 → 50th percentile
• 0.99 → 99th percentile
A child at 99th percentile = higher than 99% of peers
7. The 68–95–99.7 Rule (VERY COMMON MCQ)
In a normal distribution:
• 68% within ±1 SD
• 95% within ±2 SD
• 99.7% within ±3 SD
This helps decide what is normal vs abnormal.
8. Malnutrition & Z-Scores (WHO)
• Weight-for-height Z-score (WHZ) is used
• −2 to −3 → Moderate acute malnutrition
• Below −3 → severe
WHO growth charts are based on normal distribution statistics.
9. Binomial Distribution (Different from Normal)
Used when:
• Fixed number of trials (n)
• Only two outcomes (yes/no, success/failure)
• Same probability each time (p)
• Trials are independent
Medical Examples:
• Drug works or not
• Side effect or no side effect
• Disease present or absent
10. Cumulative Probability in Binomial
• Means: probability of k or fewer successes
P(X≤k)P(X \le k)P(X≤k)
Simple Meaning:
“What is the chance that at most k events happen?”
11. Statistical Hypothesis (VERY IMPORTANT)
What is it?
A statement about a population that we test using data.
Two Types:
1. Null hypothesis (H₀)
→ No difference / no effect
2. Alternative hypothesis (H₁)
→ There is a difference / effect
Example:
• H₀: Drug does NOT lower BP
• H₁: Drug DOES lower BP
12. Statistical Decisions
Reject H₀
• Evidence is strong
• Result is statistically significant
• Effect is unlikely due to chance
Does NOT mean clinically important!
Fail to Reject H₀
• Not enough evidence
• Does not prove no effect
• Study may be underpowered
You never say “accept H₀” ❌
13. Errors (Preview)
• Type I error: False positive (reject H₀ when it’s true)
• Type II error: False negative (fail to reject H₀ when false)
BS602 – Lecture 7
(Simple & Exam-Friendly Explanation)
1. Statistical Hypothesis (Quick reminder)
When we do a study, we start with assumptions:
• Null hypothesis (H₀):
→ No effect, no difference, nothing new
• Alternative hypothesis (H₁):
→ There is an effect or difference
We collect data to decide whether to reject H₀ or fail to reject H₀.
2. Why Errors Happen
We never study the whole population, only a sample.
Because samples vary by chance, mistakes can happen.
These mistakes are:
• Type I error
• Type II error
3. Type I Error (α) – False Positive
Definition (VERY IMPORTANT):
A Type I error happens when:
You reject H₀ even though H₀ is actually true
In simple words:
You say “there is a disease/effect” when there is none.
Medical Meaning:
• False positive
Easy Medical Examples:
• HIV test is positive, but patient is actually HIV-negative
• CT scan suggests lung cancer → biopsy is benign
• Study says a drug works → in reality, it does not
Memory Trick :
Type I = “I’m fooled”
You think something exists, but it doesn’t.
4. Type II Error (β) – False Negative
Definition:
A Type II error happens when:
You fail to reject H₀ even though H₀ is false
In simple words:
You miss a real disease or effect.
Medical Meaning:
• False negative
Easy Medical Examples:
• Cancer present, but mammogram is normal
• HIV patient tests negative early in infection
• Stroke patient has normal early CT scan
• TB patient with weak immunity → negative skin test
Memory Trick :
Type II = “I missed it”
5. Type I vs Type II (EXAM TABLE)
Feature Type I Error (α) Type II Error (β)
Common name False positive False negative
What happens Detects something that isn’t there Misses something that is there
Diagnostic test Healthy patient labeled sick Sick patient labeled healthy
Risk Anxiety, unnecessary treatment Delayed diagnosis, disease progression
Controlled by Significance level (α) Study power (1 − β)
6. Which Error Is Worse?
Depends on the situation
Type I error is worse when:
• Approving a drug that does not work
• Saying a treatment is effective when it isn’t
Type II error is worse when:
• Missing a life-saving treatment
• Missing a dangerous disease
In screening tests, we often worry more about Type II errors
In drug approval, we worry more about Type I errors
7. Significance Level (α)
What is α?
• α = probability of making a Type I error
• Risk of a false positive
Common values:
• α = 0.05 → 5% risk (most common in medicine)
• α = 0.01 → stricter (safer)
• α = 0.10 → more lenient (exploratory studies)
Simple Meaning:
If α = 0.05:
You accept a 5% chance of being wrong when claiming an effect exists.
α is chosen BEFORE the study, not after.
8. Type II Error (β) and Study Power
What is β?
• β = probability of missing a real effect
• Probability of a false negative
Study Power (VERY IMPORTANT )
Power=1−β\text{Power} = 1 - \betaPower=1−β
Meaning of Power:
Probability that the study detects a real effect if it truly exists.
Typical Target:
• Power = 80%
• Means β = 20%
9. What Increases Study Power?
Factor Effect on Power
Larger sample size ↑ Power
Bigger effect size ↑ Power
Less variability ↑ Power
Higher α ↑ Power
Low power = dangerous → you may miss real treatments.
10. Confidence Level (CL)
Relationship:
Confidence Level=1−α\text{Confidence Level} = 1 - \alphaConfidence Level=1−α
Examples:
• α = 0.05 → CL = 95%
• α = 0.01 → CL = 99%
Meaning:
How confident we are that the confidence interval contains the true value.
11. Confidence Interval (CI) – Big Concept
What is a CI?
A range of values that likely contains the true population value.
What CI gives you:
• Best estimate (sample mean)
• Uncertainty range
• Precision of the study
Key Rules:
• Narrow CI → precise estimate
• Wide CI → more uncertainty
• CI includes 0 → NOT statistically significant
• CI does not include 0 → statistically significant
12. Very Important CI Meaning ❌ (Common
Mistake)
❌ Wrong:
“There is a 95% chance the true mean is inside this CI”
✅ Correct:
If we repeat the study many times, 95% of such intervals will contain the true
mean.
13. Simple CI Calculation (Concept Only)
Steps:
1. Calculate Standard Error (SE)
SE=snSE = \frac{s}{\sqrt{n}}SE=ns
2. Find Z value (1.96 for 95%)
3. Margin of Error = Z × SE
4. CI = mean ± margin of error
You don’t need to memorize math, just interpretation.
14. Final Take-Home Summary
• Type I error (α) → false positive
• Type II error (β) → false negative
• Power (1−β) → ability to detect real effects
• Confidence level = 1 − α
• Confidence interval shows range + precision
• CI excluding 0 = statistically significant
BS602 – Lecture 8
Simple Explanation (What you REALLY need to
understand)
1. What are Statistical Tests?
Statistical tests help you answer one basic question:
Is the difference I see REAL, or just due to chance?
In medicine, we use them to decide:
• Is this patient group different from normal?
• Is drug A better than placebo?
• Is this change meaningful or random?
2. Big Picture: Types of Statistical Tests
A) Parametric tests (most important for this lecture)
Used when:
• Data are numerical & continuous
• Data are approximately normal
• Variability is reasonable
• Observations are independent
Examples:
• t-test
• z-test
• ANOVA
• Regression
Lecture 8 focuses mainly on t-tests and z-tests
B) Non-parametric tests (you’ll study later)
Used when:
• Data are NOT normal
• Data are ranks or ordinal
• Many outliers
Examples:
• Mann–Whitney
• Wilcoxon
• Spearman
3. One-Sample Tests (VERY IMPORTANT)
Used when you have:
ONE group, and you compare it to a known value
Typical medical question:
“Is my patients’ mean different from the normal/reference value?”
Examples:
• Mean glucose vs normal (5.6 mmol/L)
• Mean BP vs 120 mmHg
• Mean cholesterol vs 200 mg/dL
4. One-Sample t-Test (MOST COMMON )
When do you use it?
Use one-sample t-test when:
• One group only
• Population SD is unknown (almost always true)
• Sample size is small or moderate
This is the default test in medical research
Hypotheses (EXAM FAVORITE)
• H₀: μ = μ₀ (no difference)
• H₁:
o μ ≠ μ₀ → two-tailed
o μ > μ₀ → right-tailed
o μ < μ₀ → left-tailed
What does the t-value mean?
It tells you how far your sample mean is from the reference value, in standard
errors.
• t ≈ 0 → very close → no difference
• |t| large → far away → likely real difference
Standard Error (SE) – SIMPLE IDEA
• SD → variability of individuals
• SE → variability of the mean
Smaller SE = more precise estimate
5. Decision Making (CRITICAL )
Using p-value (most important)
• p < α (0.05) → reject H₀ → significant
• p ≥ α → fail to reject H₀ → not significant
This is how SPSS / R / exams work
Example (Fasting Glucose)
• Sample mean = 6.03
• Normal = 5.6
• p-value = 0.000088
p < 0.05 → Reject H₀
Mean glucose is significantly higher
6. One-Tailed vs Two-Tailed Tests
Test Question
Two-tailed “Is it different?”
One-tailed (>) “Is it greater?”
One-tailed (<) “Is it smaller?”
If your direction is wrong → p-value becomes large → NOT significant
(this is why planning matters!)
7. One-Sample z-Test (RARE )
When do you use it?
Use z-test only when:
• Population SD (σ) is known
• Sample size is large (≥30)
In real medicine: rare
In exams: they love asking the condition
Interpretation is SAME as t-test
• p < 0.05 → significant
• CI excluding reference value → significant
8. Two-Sample z-Test
Used when:
You compare TWO independent groups
Examples:
• Drug vs placebo
• Male vs female
• Treatment A vs treatment B
Conditions (EXAM )
• Two independent samples
• Continuous data
• Population SDs known
• Large sample sizes
Hypotheses
• H₀: μ₁ = μ₂
• H₁:
o μ₁ ≠ μ₂
o μ₁ > μ₂
o μ₁ < μ₂
Interpretation (Same rules!)
• p < 0.05 → groups are significantly different
• CI includes 0 → not significant
• CI excludes 0 → significant
9. Confidence Interval (Quick Reminder)
• Shows range of plausible values
• Narrow CI → precise
• Wide CI → uncertainty
CI includes 0 → no difference
CI excludes 0 → significant difference
10. Very Important Exam Comparison
Test When to Use
One-sample t-test 1 group vs known value, σ unknown
One-sample z-test 1 group vs known value, σ known
Two-sample z-test 2 independent groups, σ known
Test When to Use
(Later) Two-sample t-test 2 groups, σ unknown
11. Take-Home Summary (MEMORIZE THIS )
• Statistical tests decide chance vs real
• t-test → σ unknown (most common)
• z-test → σ known (rare)
• One-sample → compare to a reference
• Two-sample → compare two groups
• p-value < 0.05 → statistically significant
• CI excluding reference/0 → confirms significance
📘 BS602 – Lecture 9 (Simple Master-Level Explanation)
1⃣ What is the p-value? (MOST IMPORTANT)
Definition (simple):
The p-value tells you how surprising your data are IF the null hypothesis is true.
It does NOT tell you:
• the probability that the disease is present ❌
• the probability that H₀ is true ❌
It DOES tell you:
• how unlikely your result is assuming no real effect exists ✅
Think like this:
• Null hypothesis (H₀) = “Nothing is happening / no difference / no disease”
• Data = what you observed
• p-value = “If H₀ were true, how often would I see data this extreme by
chance?”
Interpretation:
• p < 0.05 → very unlikely by chance → reject H₀
• p > 0.05 → could happen by chance → fail to reject H₀
Clinical analogy (easy to remember):
Fire alarm example
• H₀: No fire
• Alarm rings loudly (data)
• Small p-value → alarm rarely rings without fire → suspect fire → reject H₀
• Large p-value → alarm often false → no fire → fail to reject H₀
⭐ Exam sentence:
“The p-value measures how extreme the observed data are under the assumption that
the null hypothesis is true.”
2⃣ Link between t-value and p-value
What is the t-value?
• It tells you how far your result is from H₀, measured in standard error
units
In simple words:
• t-value = distance
• p-value = probability
Relationship:
• Large |t| → result far from H₀ → small p-value → significant
• Small |t| → result close to H₀ → large p-value → not significant
Visual idea:
• Center of t-distribution = H₀
• Your t-value is a point on the curve
• p-value = shaded area beyond that point
3⃣ Two-Sample t-Test (Independent t-test)
Purpose:
To compare means of TWO independent groups
Example:
• Drug A vs Drug B
• Male vs Female
• Treatment vs Control
Hypotheses:
• H₀: μ₁ = μ₂ (no difference)
• H₁: μ₁ ≠ μ₂ (difference exists)
Assumptions:
1. Independent samples
2. Approximately normal data
3. Equal variances → use standard t-test
4. Unequal variances → use Welch’s t-test
Interpretation of t-value:
• Large |t| → strong evidence against H₀
• Small |t| → weak evidence
• Positive t → group 1 mean > group 2 mean
• Negative t → group 1 mean < group 2 mean
4️⃣ Welch’s t-test (Unequal Variance)
✅ When to use:
• Variances are unequal
• Sample sizes are different
• Very common in medical research
Safer than standard t-test
Example (Blood Pressure):
• Mean difference = 12 mmHg
• t ≈ 2.17
• p ≈ 0.036
Interpretation:
• Difference is ~2 standard errors away from zero
• p < 0.05 → statistically significant
• Difference unlikely due to chance
⭐ Exam line:
“Welch’s t-test compares two independent means without assuming equal variances.”
5⃣ Confidence Interval (CI)
What does CI mean?
• Range of values where the true mean difference likely lies
Rule:
• If 0 is NOT inside CI → significant
• If 0 is inside CI → not significant
Example:
• CI = (0.78 , 23.21) → does NOT include 0 → significant
6⃣ One-Way ANOVA (VERY HIGH-YIELD)
Why ANOVA?
• To compare 3 or more group means
• Using multiple t-tests increases error ❌
Hypotheses:
• H₀: All means are equal
• H₁: At least one mean is different
Key idea:
ANOVA compares:
• Between-group variation
• Within-group variation
If between-group variation is much larger → groups differ
F-Statistic:
• F = Between-group variance / Within-group variance
• Large F → small p-value → reject H₀
Important:
• ANOVA tells you IF a difference exists
• NOT which group differs
• For that → post-hoc tests (e.g., Tukey)
7⃣ ANOVA Assumptions
1. Independent samples
2. Normal distribution within each group
3. Equal variances (homogeneity)
Variance check:
• Levene’s test
• p > 0.05 → assumption satisfied ✅
8️⃣ Interpretation of ANOVA Example
Results:
• p = 2.72 × 10⁻¹³ (EXTREMELY small)
• Reject H₀
• Means differ strongly
Meaning:
• Differences between drugs >> differences within each drug
• Strong clinical & statistical evidence
FINAL TAKE-HOME SUMMARY (EXAM
GOLD)
• p-value → how surprising data are if H₀ is true
• t-value → distance from H₀
• Two-sample t-test → compare 2 means
• Welch’s t-test → unequal variances
• ANOVA → compare ≥3 means
• Small p (<0.05) → reject H₀
• ANOVA significant? → use post-hoc tests