Non-parametric Tests & Research Methodology
Unit 3 shifts focus to Non-parametric tests—essential when data doesn’t follow a normal distribution (e.g., ranked clinical scores). It then comprehensively covers Research Methodology: defining the need for research, experiential design techniques, and avoiding plagiarism. A major portion dedicated to designing clinical trials, cohort/observational studies, estimating sample size, and writing study protocols. Finally, it explores various graphical data representations (Histograms, Pie Charts, Contour/Response Surface Plots) critical for results presentation.
Syllabus & Topics
- 1Non-Parametric Tests: Used when assumptions of parametric tests (like a Normal Distribution) are NOT met. Often used for ordinal (ranked) data or small sample sizes with skewed distributions. They are ‘distribution-free’. They are less powerful than parametric tests but more robust against outliers. (1) Wilcoxon Rank Sum Test: The non-parametric equivalent of the Unpaired t-test (compares two independent groups). Ranks all data from lowest to highest, regardless of group, then compares the sum of the ranks. (2) Mann-Whitney U Test: Very similar/equivalent to Wilcoxon Rank Sum, used to determine if two independent samples come from the same population. (3) Kruskal-Wallis Test: The non-parametric equivalent of One-Way ANOVA (compares 3 or more independent groups). (4) Friedman Test: The non-parametric equivalent of repeated-measures (paired) ANOVA (e.g., measuring pain scores in the SAME patients across 4 different time points).
- 2Introduction to Research Methodology: Research: A systematic, scientific inquiry to establish novel facts, solve new or existing problems, prove new ideas, or develop new theories. Need for Research: In pharmacy, to discover safer, more effective drugs; improve formulations; understand disease mechanisms; and ensure public safety. Need for Design of Experiments (DoE): Doing unstructured experiments wastes time and resources. A proper design ensures valid, objective, and accurate conclusions while minimizing experimental error and the number of trials needed. Plagiarism: The unethical practice of presenting someone else’s work, ideas, data, or words as one’s own without appropriate acknowledgment or citation. Can lead to severe academic and professional penalties.
- 3Designing the Methodology – Epidemiological Studies: (1) Observational Studies: The researcher simply observes the subjects without intervening or giving them a treatment. (e.g., observing the natural progression of a disease). (2) Cohort Studies: A type of observational study where a group (cohort) exposed to a certain factor and an unexposed group are followed FORWARD in time (prospectively) to see who develops a disease (e.g., following smokers vs. non-smokers for 20 years to see who gets lung cancer). Very strong for establishing cause-and-effect but expensive and time-consuming. (3) Case-Control Studies: Looking BACK in time (retrospectively). Taking people who ALREADY have the disease (cases) and comparing them to people who don’t (controls) to see what they were exposed to in the past (e.g., asking lung cancer patients vs. healthy people about their past smoking habits). Faster, but prone to recall bias. (4) Experimental Studies: The researcher actively intervenes (e.g., gives a drug or a placebo) and observes the outcome. Clinical trials are the gold standard.
- 4Clinical Trials & Phases: Clinical Trial: A rigorously controlled human experimental study to evaluate the safety and efficacy of a new medical intervention. Phases: Pre-clinical: Animal testing for initial safety/toxicity. Phase I (Human Pharmacology): ‘Is it safe?’ Tested on a small group (20-100) of HEALTHY volunteers. Focuses on safety, tolerability, PK/PD, and dosage range. Phase II (Exploratory): ‘Does it work?’ Tested on a larger group (100-300) of PATIENTS with the disease. Focuses on preliminary efficacy, optimal dosing, and short-term side effects. Phase III (Confirmatory): ‘Is it better?’ Tested on a large group (1000-3000) of PATIENTS across multiple centers. Usually a Randomized Controlled Trial (RCT) comparing the new drug against the current standard treatment or placebo. Provides definitive proof of efficacy/safety for regulatory approval (NDA). Phase IV (Post-Marketing Surveillance): ‘Are there long-term/rare side effects?’ Conducted after the drug is marketed. Monitors real-world effectiveness and rare adverse effects in the general population.
- 5Protocol, Sample Size & Presentation: Protocol: A detailed written plan outlining the entire study (objectives, design, methodology, statistical considerations, ethical issues). It must be approved by an Ethics Committee before starting. Sample Size Determination: Crucial calculation BEFORE starting a study. If ‘n’ is too small, the study lacks the ‘statistical power’ to detect a real difference (high risk of Type II error). If ‘n’ is too large, it wastes resources and is ethically questionable. Power of a study: The probability (usually set at 80% or 90%) that the study will correctly reject a false null hypothesis (i.e., successfully find a difference if one truly exists). Report Writing: Presenting the findings clearly, often leading to a publication or regulatory submission.
- 6Graphs in Data Presentation: Graphical representation is crucial for communicating complex statistical data clearly. (1) Histogram: A bar graph representing the frequency distribution of continuous data (bars touch each other). The area of each bar is proportional to the frequency. (2) Pie Chart: A circular chart divided into sectors, illustrating numerical proportion (useful for categorical data like percentages of different side effects). (3) Cubic Graph (3D plotting): Visualizing data across 3 dimensions (X, Y, Z axes). (4) Response Surface Plot: A 3D graph used in Design of Experiments (DoE) that maps the response (e.g., tablet hardness) against two varying formulation factors (e.g., binder concentration and compression force), looking like a topographical map or ‘surface’. (5) Contour Plot: The 2D top-down view of a response surface plot, where lines connect points of equal response (similar to elevation lines on a hiking map). Essential in optimization studies.
Learning Objectives
Exam Prep Questions
Q1. If Non-Parametric tests don’t require strong assumptions like Normal Distribution, why don’t we just use them all the time?
Non-parametric tests are considered less “powerful” than parametric tests. This means that if there truly IS a difference between two drugs, a parametric test (like a t-test) is more likely to successfully detect it. Non-parametric tests have a higher chance of falling victim to a Type II error (failing to find a difference that actually exists). Therefore, if your data meets the strict criteria (normally distributed, continuous), you should ALWAYS choose the parametric option for better statistical power.
Q2. What is the difference between an Observational Study and an Experimental Study (Clinical Trial)?
In an Observational Study (like a cohort or case-control study), the researcher acts as a passive observer. They record what people do naturally (e.g., eat a certain diet, smoke) and see what happens to their health over time.
In an Experimental Study, the researcher actively INTERVENES. They randomly assign subjects to receive a specific intervention (e.g., taking a new drug vs. a placebo) and strictly control the environment to measure the outcome.
Q3. Why do we test Phase I clinical trials on HEALTHY volunteers instead of sick patients?
The primary goal of a Phase I trial is solely to determine SAFETY, not efficacy (not whether the drug cures the disease). It establishes the maximum tolerated dose and how the body processes the drug (Pharmacokinetics). Healthy volunteers are used because their bodies do not have underlying organ damage or disease-related variables that might confuse the safety data or put an already vulnerable patient at undue risk. (Exceptions exist, such as highly toxic cancer drugs, which are tested directly in advanced cancer patients in Phase I).