Prof Ismail A Suleimana

PCP 404 : BIOSTATISTICS AND RESEARCH METHODOLOGY Two Credit Units

Outline

• Data presentation;
• inferential statistics:
• Hypothesis testing.
• Parametric and non-parametric tests:
• Student T test (paired and independent); Z score test; ANOVA; Chi square test and their non-
parametric equivalent.
• Correlation and Regression.

Introduction
Biostatistics and Research Methodology are essential components of the pharmacy curriculum,
equipping students with the skills to design studies, collecting, organizing, summarizing,
presenting, analyze data, and make evidence-based decisions in pharmaceutical practice and
research.

Biostatistics involves the application of statistical techniques to biological and health-related

fields. For pharmacy students, understanding biostatistics is crucial for interpreting clinical trial
data, assessing drug efficacy, and ensuring the safety of pharmaceutical interventions. Key topics
include measures of central tendency (mean, median, mode), measures of dispersion (range,
standard deviation), probability distributions, hypothesis testing, correlation, and regression
analysis. These concepts enable pharmacists to critically evaluate research findings and
contribute to clinical decision-making.

Research Methodology focuses on the systematic processes employed in conducting scientific

research. For pharmacy students, this encompasses understanding various research designs (such
as experimental, observational, and clinical trials), data collection methods, sampling techniques,
and ethical considerations in research. A solid grasp of research methodology allows pharmacists
to design robust studies, contribute to the development of new therapies, and ensure that research
findings are valid and applicable to patient care.

Integrating biostatistics with research methodology enables pharmacy students to not only
conduct and analyze research effectively but also to apply these findings in clinical settings,
thereby enhancing patient outcomes and advancing the field of pharmacy.
Data Presentation

Effective data presentation is essential in biostatistics to summarize, interpret, and communicate

research findings clearly and accurately. Proper presentation methods facilitate understanding and
enable informed decision-making.

Methods of Data Presentation:

1. Textual Presentation:
o Involves describing data using text, often in paragraph form.

o Useful for providing detailed explanations or highlighting specific data points.

o Example: "The study observed that 60% of participants responded positively to the
treatment, while 40% showed no significant improvement."

2. Tabular Presentation:
o Utilizes tables to organize data into rows and columns, allowing for systematic
arrangement and easy comparison.
o Effective for presenting large datasets or complex information succinctly.

o Example: A table displaying the frequency distribution of blood pressure categories

among patients.

3. Graphical Presentation:
o Employs visual elements like charts and graphs to represent data, making patterns
and trends more discernible.
o Common types include:

▪ Bar Charts: Depict categorical data with rectangular bars proportional to

the values they represent.

▪ Histograms: Show frequency distributions of continuous data, with

adjacent bars indicating data intervals.

▪ Pie Charts: Illustrate proportions of a whole, with slices representing

different categories.

▪ Line Graphs: Display data points connected by lines, useful for showing
trends over time.

▪ Scatter Plots: Present relationships between two continuous variables,

with data points plotted on a Cartesian plane.

Considerations for Effective Data Presentation:

• Clarity: Ensure that the presentation is straightforward and free from ambiguity.
• Accuracy: Represent data precisely without distortion or misrepresentation.

• Relevance: Include only pertinent data that supports the study's objectives.
 • Consistency: Maintain uniformity in formatting, scales, and labeling across all data
presentations.

Inferential statistics

Inferential statistics is a branch of statistics that allows researchers to make conclusions and
predictions about a population based on data collected from a sample. This approach enables
analysts to extend findings beyond the immediate dataset, providing insights that inform decisions
about larger groups or phenomena.

Key Concepts in Inferential Statistics:

1. Population and Sample:
o Population: The complete set of all possible observations or individuals of interest.

o Sample: A subset of the population selected for analysis.

2. Sampling Methods:

o Techniques used to select a representative subset from the population, such as

simple random sampling, stratified sampling, and cluster sampling.

3. Estimation:

o Point Estimation: Provides a single value estimate of a population parameter (e.g.,

sample mean as an estimate of population mean).

o Interval Estimation: Offers a range of values (confidence interval) within which the
population parameter is expected to lie.

4. Hypothesis Testing:

o A method to assess evidence provided by sample data in relation to a specific claim

about a population parameter.

o Involves formulating null and alternative hypotheses, selecting an appropriate test,

and making decisions based on p-values or confidence intervals.

5. Regression Analysis:

o Examines the relationship between dependent and independent variables to make

predictions or understand associations.

Applications of Inferential Statistics:

• Healthcare: Determining the effectiveness of new treatments by analyzing clinical trial
data.
• Business: Forecasting market trends and consumer behavior based on sample surveys.

• Social Sciences: Investigating societal trends and behaviors through observational

studies.
Utilizing inferential statistics, researchers can make informed decisions and predictions about a
population without examining every individual, saving time and resources while maintaining
accuracy.

Hypothesis testing
Hypothesis testing is a fundamental statistical method used to make inferences about a population
based on sample data. It involves formulating two opposing hypotheses—the null hypothesis (H₀)
and the alternative hypothesis (H₁)—and using sample data to determine which hypothesis is
supported by the evidence.
Key Components:
1. Null Hypothesis (H₀): This is the default assumption that there is no effect or difference in
the population. For example, stating that a new drug has no effect on blood pressure
compared to a placebo.
2. Alternative Hypothesis (H₁ or Ha): This hypothesis suggests that there is an effect or
difference. Continuing the example, it would posit that the new drug does affect blood
pressure.

Steps in Hypothesis Testing:

1. State the Hypotheses: Clearly define H₀ and H₁.
2. Choose the Significance Level (α): Commonly set at 0.05, this threshold determines the
probability of rejecting H₀ when it is actually true (Type I error).

3. Select the Appropriate Test: Depending on the data and research question, choose a
statistical test (e.g., t-test, chi-square test).

4. Calculate the Test Statistic: Compute the statistic that summarizes the data in the context
of the chosen test.

5. Determine the p-value: The p-value indicates the probability of obtaining results at least as
extreme as the observed ones, assuming H₀ is true.

6. Make a Decision: If the p-value is less than or equal to α, reject H₀; otherwise, do not reject
H₀.

Example:

Suppose researchers want to test if a new medication lowers blood pressure more effectively than
an existing one.

• H₀: There is no difference in effectiveness between the new and existing medications.

• H₁: The new medication is more effective than the existing one.

After conducting a study and analyzing the data, they calculate a p-value of 0.03. With a
significance level of 0.05, they reject H₀ and conclude that the new medication is statistically
significantly more effective.

Considerations:
 • Type I Error (False Positive): Rejecting H₀ when it is true.

• Type II Error (False Negative): Failing to reject H₀ when H₁ is true.

Hypothesis testing provides a structured framework for making data-driven decisions and
assessing the strength of evidence in research findings.

Parametric and non-parametric tests

Parametric and non-parametric tests are two broad categories of statistical tests used to analyze
data. The choice between them depends on the nature of the data, distribution assumptions, and
sample size.

1. Parametric Tests

These tests assume that the data follows a specific distribution, typically a normal distribution.
They require that the data meets certain conditions, such as homogeneity of variance and interval
or ratio level of measurement.

Examples of Parametric Tests:

• t-Test: Compares the means of two groups (independent or paired samples).

• Analysis of Variance (ANOVA): Compares means across three or more groups.

• Pearson’s Correlation Coefficient: Measures the strength and direction of the linear
relationship between two variables.

• Linear Regression: Examines the relationship between dependent and independent

variables.

Advantages of Parametric Tests:

• More powerful when assumptions are met.

• Can detect smaller differences due to efficiency.

• Provide more detailed insights about effect sizes.

Disadvantages of Parametric Tests:

• Require normality and other assumptions to be met.

• Sensitive to outliers and skewed distributions.

2. Non-Parametric Tests

These tests do not assume a specific distribution for the data, making them useful when normality
cannot be assumed or when dealing with ordinal or non-normally distributed interval data.

Examples of Non-Parametric Tests:

• Mann-Whitney U Test: Alternative to the independent t-test, compares two independent

groups.
 • Wilcoxon Signed-Rank Test: Alternative to the paired t-test, compares two related groups.

• Kruskal-Wallis Test: Alternative to ANOVA, used for comparing three or more groups.

• Spearman’s Rank Correlation: Non-parametric alternative to Pearson’s correlation.

• Chi-Square Test: Used for categorical data to test associations between variables.

Advantages of Non-Parametric Tests:

• Do not require normality.

• Can handle small sample sizes and skewed data.

• Suitable for ordinal and categorical data.

Disadvantages of Non-Parametric Tests:

• Less powerful when parametric assumptions hold.

• Do not provide as much detailed information about effect size.

When to Use Each Type of Test

Scenario Recommended Test Type

Data is normally distributed, with equal variances Parametric

Data is skewed or does not follow a normal distribution Non-Parametric

Sample size is large (e.g., >30 per group) Parametric

Sample size is small (e.g., <30 per group) Non-Parametric

Data is at the ordinal or nominal level Non-Parametric

Data is at the interval or ratio level Parametric

Student’s t-test (paired and independent), Z-score test, ANOVA, Chi-square test, and their
non-parametric equivalents:

1. Student’s t-test (Paired and Independent)

The t-test is a parametric test used to compare the means of two groups. It assumes that the data
follows a normal distribution and that the samples have equal variances.

(a) Independent t-test (Unpaired t-test)

• Purpose: Compares the means of two independent groups.

• Example: Comparing the average blood pressure of two different groups of patients (e.g.,
hypertensive vs. non-hypertensive).
 • Assumptions:

o Data is normally distributed.

o The two groups are independent.

o Variances are equal (can be adjusted with Welch’s correction if unequal).

(b) Paired t-test (Dependent t-test)

• Purpose: Compares the means of two related (paired) samples.

• Example: Measuring patients’ blood pressure before and after a drug intervention.

• Assumptions:

o Data is normally distributed.

o The same subjects are measured twice.

Non-Parametric Equivalent:

• Independent t-test → Mann-Whitney U test

• Paired t-test → Wilcoxon Signed-Rank test

2. Z-score Test

• Purpose: Used to compare sample and population means when the population standard
deviation (σ\sigma) is known.

• Example: Testing if the average cholesterol level in a sample differs from a known
population mean.

• Assumptions:

o Population variance is known.

o Data is normally distributed or has a large sample size (n>30n > 30).

Non-Parametric Equivalent:

• The Z-test doesn’t have a direct non-parametric equivalent, but Sign Test can be used for
single-sample comparisons.

3. Analysis of Variance (ANOVA)

ANOVA is a parametric test that compares means across three or more groups.

(a) One-Way ANOVA

• Purpose: Tests if there are significant differences between the means of three or more
independent groups.

• Example: Comparing the effectiveness of three different hypertension drugs.

 • Assumptions:

o Data is normally distributed.

o Groups are independent.

o Variances are equal (homogeneity of variance).

(b) Two-Way ANOVA

• Purpose: Examines the effect of two independent variables on a dependent variable.

• Example: Studying the effect of drug type (A, B, C) and age group (Young, Old) on blood
pressure.

Non-Parametric Equivalent:

• One-way ANOVA → Kruskal-Wallis Test

• Two-way ANOVA → Friedman Test

4. Chi-Square Test (χ² Test)

The Chi-square test is a non-parametric test used for categorical data.

(a) Chi-Square Test for Independence

• Purpose: Tests if there is an association between two categorical variables.

• Example: Checking if there is a relationship between smoking status (Yes/No) and

hypertension diagnosis (Yes/No).

• Assumptions:

o Data is categorical.

o Expected frequency in each cell should be ≥ 5.

(b) Chi-Square Goodness of Fit Test

• Purpose: Checks if a sample distribution matches an expected distribution.

• Example: Testing if the distribution of blood types (A, B, AB, O) in a sample matches the
general population distribution.

Non-Parametric Equivalent:

Chi-square is already non-parametric, but for small sample sizes, Fisher’s Exact Test is used.

Summary of Tests and Their Non-Parametric Equivalents

Parametric Test Purpose Non-Parametric Equivalent

Independent t- Compare means of two independent Mann-Whitney U test

Parametric Test Purpose Non-Parametric Equivalent

test groups

Paired t-test Compare means of related samples Wilcoxon Signed-Rank test

Compare sample mean with population

Z-score test Sign Test (for median comparisons)
mean

One-way ANOVA Compare means across multiple groups Kruskal-Wallis Test

Two-way ANOVA Compare multiple groups with two factors Friedman Test

Test association between categorical Fisher’s Exact Test (for small

Chi-square test
variables samples)

Practical examples of applying these statistical tests in pharmaceutical and clinical research:

1. Independent t-test (Unpaired t-test)

Example: Comparing Drug Efficacy

Scenario: A study compares the effect of Drug A vs. Drug B on reducing blood pressure.

• Groups: Patients receiving Drug A vs. patients receiving Drug B.

• Application: If the mean systolic blood pressure reduction differs significantly between
the two groups, the independent t-test will determine if the difference is statistically
significant.

Non-Parametric Equivalent: Mann-Whitney U Test

Used when blood pressure reduction values are not normally distributed.

2. Paired t-test (Dependent t-test)

Example: Pre- and Post-Treatment Comparison

Scenario: A clinical trial tests the effect of a new diabetes drug on fasting blood glucose levels.

• Before Treatment: Blood glucose levels are measured.

• After Treatment: Blood glucose levels are measured again.

• Application: A paired t-test will assess if the difference in glucose levels before and after
treatment is statistically significant.

Non-Parametric Equivalent: Wilcoxon Signed-Rank Test

Used when blood glucose levels do not follow a normal distribution.

3. Z-score Test
Example: Comparing a Sample to Population Data

Scenario: A pharmaceutical company tests a new cholesterol-lowering drug. The general

population’s average LDL cholesterol is 120 mg/dL.

• Application: A Z-test determines if the mean LDL cholesterol level in the study sample
significantly differs from the general population.

Non-Parametric Equivalent: Sign Test

Used when data is not normally distributed.

4. One-Way ANOVA

Example: Comparing Multiple Drug Formulations

Scenario: A pharmaceutical company develops three formulations of a pain relief drug and wants
to compare their effectiveness.

• Groups: Patients receiving Formulation A, Formulation B, and Formulation C.

• Application: ANOVA determines if the mean pain reduction score significantly differs
among the three groups.

Non-Parametric Equivalent: Kruskal-Wallis Test

Used when pain scores are skewed or ordinal.

5. Two-Way ANOVA

Example: Investigating Drug Effect Across Age Groups

Scenario: Researchers study how a hypertension drug’s effectiveness varies based on:

1. Dosage level (Low, Medium, High)

2. Age group (Young, Middle-aged, Elderly)

• Application: Two-way ANOVA assesses both factors and their interaction to see if they
influence blood pressure reduction.

Non-Parametric Equivalent: Friedman Test

Used when data is not normally distributed.

6. Chi-Square Test for Independence

Example: Assessing Drug Adverse Reactions

Scenario: A hospital collects data on adverse effects from a new antibiotic and categorizes them
as:

1. Present (Yes/No)

2. Patient Gender (Male/Female)

 • Application: The Chi-square test determines if adverse effects are associated with gender.

Non-Parametric Alternative: Fisher’s Exact Test

Used if sample sizes are very small.

7. Chi-Square Goodness of Fit Test

Example: Checking Genetic Predisposition for Drug Response

Scenario: A genetic study tests whether the observed distribution of a drug metabolism gene
variant in a population matches the expected distribution.

• Application: A Chi-square goodness-of-fit test compares expected vs. observed

frequencies of gene variants.

Conclusion

Statistical Test Clinical Application Example

Independent t- Compare two independent

Drug A vs. Drug B on BP reduction
test groups

Pre- and post-treatment

Paired t-test Blood glucose before and after a diabetes drug
comparison

Comparing a cholesterol drug’s effect to the

Z-score test Sample vs. population mean
general population

Compare three or more drug

One-way ANOVA Pain relief drug formulations
formulations

Investigate drug effects across

Two-way ANOVA Hypertension drug effects across age groups
factors

Relationship between
Chi-square test Adverse drug reactions and gender
categorical variables

Step-by-step guides on how to perform these statistical tests in SPSS for pharmaceutical and
clinical research data analysis.

1. Independent t-test in SPSS and R

Scenario: Comparing the reduction in blood pressure between Drug A and Drug B.

SPSS Procedure:
 1. Load Data:

o Ensure data is structured with one column for Blood Pressure Reduction and one
for Drug Group (A/B).

2. Go to: Analyze → Compare Means → Independent-Samples T Test.

3. Select:

o Test Variable: Blood Pressure Reduction.

o Grouping Variable: Drug Group (A/B).

o Click Define Groups, enter 1 for Drug A and 2 for Drug B.

4. Click OK.

5. Interpret Output:

o Look at the p-value (Sig. 2-tailed) under Levene's Test for Equality of Variances.

o If p < 0.05, there's a significant difference.

2. Paired t-test in SPSS

Scenario: Measuring blood glucose before and after treatment.

SPSS Procedure:

1. Go to: Analyze → Compare Means → Paired-Samples T Test.

2. Select:

o Pre-Treatment Glucose as Variable 1.

o Post-Treatment Glucose as Variable 2.

3. Click OK.

4. Interpret Output:

o Check Sig. (2-tailed) value under Paired Differences for significance.

3. Z-score Test in SPSS and R

Scenario: Comparing a sample’s LDL cholesterol to the population mean.

SPSS Procedure:

1. Compute the Z-score manually using: Z=Xˉ−μσ/nZ = \frac{\bar{X} - \mu}{\sigma / \sqrt{n}}

2. Use the Compute Variable function:

 o Go to Transform → Compute Variable.

o Create a new variable:

o Z = (Mean_LDL - 120) / (15 / sqrt(N))

o Where 120 is the population mean, and 15 is the standard deviation.

4. One-Way ANOVA in SPSS and R

Scenario: Comparing pain relief scores across three drug formulations.

SPSS Procedure:

1. Go to: Analyze → Compare Means → One-Way ANOVA.

2. Select:

o Dependent Variable: Pain Relief Score.

o Factor Variable: Drug Formulation (A, B, C).

3. Click OK.

4. Interpret Output:

o Look at Sig. value under ANOVA → If p < 0.05, there is a significant difference.

5. Two-Way ANOVA in SPSS and R

Scenario: Studying how drug dosage and age group affect blood pressure.

SPSS Procedure:

1. Go to: Analyze → General Linear Model → Univariate.

2. Select:

o Dependent Variable: Blood Pressure Reduction.

o Fixed Factors: Dosage Level and Age Group.

3. Click OK.

4. Interpret Output:

o Check Sig. values for main effects (Dosage & Age) and interaction effect (Dosage *
Age).

6. Chi-Square Test for Independence in SPSS and R

Scenario: Checking if adverse drug effects depend on gender.

SPSS Procedure:
 1. Go to: Analyze → Descriptive Statistics → Crosstabs.

2. Select:

o Row Variable: Adverse Reaction (Yes/No).

o Column Variable: Gender (Male/Female).

3. Click Statistics, check "Chi-square", then OK.

4. Interpret Output:

o Look at Asymp. Sig. under Chi-Square Tests. If p < 0.05, the association is
significant.

7. Chi-Square Goodness-of-Fit Test in SPSS and R

Scenario: Checking if observed distribution of genetic variants matches expected proportions.

SPSS Procedure:

1. Go to: Analyze → Nonparametric Tests → Legacy Dialogs → Chi-Square.

2. Select the variable representing genetic variants.

3. Enter expected values.

4. Click OK.

5. Interpret Output:

o Check Sig. under Chi-Square Tests. If p < 0.05, the sample distribution differs
significantly.

Final Thoughts

Test SPSS

t-test (Independent & Paired) Compare Means → T-test

Z-score Test Compute Variable

ANOVA (One-way & Two-way) General Linear Model

Chi-Square Test Crosstabs

Correlation and Regression in Pharmaceutical and Clinical Research

1. Correlation Analysis

Correlation measures the strength and direction of a relationship between two variables. It does
not imply causation.
Example in Pharmaceutical Research

• Scenario: A study investigates the correlation between dosage of a painkiller and pain
relief score.

• Expected Outcome: A positive correlation means higher doses provide greater pain relief.

Common Correlation Methods:

Method When to Use Interpretation

Pearson’s Correlation When both variables are normally Measures linear relationship
(r) distributed and continuous (values range from -1 to +1)

Spearman’s Rank When data is not normally

Measures monotonic relationship
Correlation (ρ) distributed or ordinal

When sample size is small or ranks

Kendall’s Tau Measures rank correlation
are used

SPSS Steps for Pearson’s Correlation:

1. Go to: Analyze → Correlate → Bivariate.

2. Select Variables: (e.g., Dosage & Pain Relief Score).

3. Check ‘Pearson’ and OK.

4. Interpret Output:

o r > 0 → Positive correlation.

o r < 0 → Negative correlation.

o p-value < 0.05 → Significant correlation.

2. Regression Analysis

Regression predicts one variable (dependent) from another (independent).

Example in Clinical Research

• Scenario: Predict blood pressure reduction based on antihypertensive drug dose.

• Application: Helps determine optimal dose-response relationship.

Types of Regression:

Regression Type Use Case Example

Simple Linear Predict one outcome based on Predict blood sugar levels based on insulin
Regression one predictor dose
Regression Type Use Case Example

Multiple Linear Predict one outcome using Predict cholesterol reduction using diet,
Regression multiple predictors exercise, and medication

Predict a binary outcome Predict whether a patient experiences an

Logistic Regression
(Yes/No) adverse drug reaction (Yes/No)

SPSS Steps for Simple Linear Regression:

1. Go to: Analyze → Regression → Linear.

2. Select:

o Dependent Variable: Blood Pressure Reduction.

o Independent Variable: Drug Dosage.

3. Click OK.

4. Interpret Output:

o R² value → Strength of prediction.

o p-value → Significance of predictor.