Research design fundamentals in epidemiology and public health
Measurement validity and reliability
Error and bias in population health research
Study design approaches:
Cross-sectional
Case-control
Cohort
Experimental
Qualitative methods and their integration with quantitative approaches
Ethical considerations in population health research
1A Matter of Measurement
Primary vs. Secondary Data
Primary data:
- Data collected specifically for the purpose of the study.
Secondary data:
- Data collected for other purposes but reorganized and reanalyzed.
Examples:
- Health insurance claims data
- Employment records
- National health surveys
2Primary vs. Secondary Data
Primary data
Collected specifically for a new study
Controlled by the researcher
Tailored to answer specific research questions
Secondary data
Pre-existing data collected for another purpose
Often large-scale and readily available
May require cleaning or transformation
While primary data offers control and precision, secondary data can save time and resources.
3Levels of Measurement
Level
Description
Example
Operations
Nominal
Categories with no ranking
Blood type, Sex
Equality/inequality
Ordinal
Ordered categories
Health self-rating: excellent, good, fair, poor
Greater than/less than
Interval
Equal units, no true zero
Temperature in °C
Addition/subtraction
Ratio
Equal units with true zero
Weight, Blood pressure
Multiplication/division
Understanding measurement levels is crucial for selecting appropriate statistical analyses. A variable can always be reduced to a lower level of measurement (continuous to categorical), but not elevated (categorical to continuous).
4Ecological Studies and Fallacy
Unit of analysis: Group (e.g., city-level data)
Examples:
Community fluoride levels and dental caries
Countries’ smoking rates and lung cancer rates
Ecological fallacy: Attributing group-level associations to individuals
Example:
Classrooms with more women had higher average grades
But individual-level analysis showed men had higher grades in each classroom
Classroom A
Classroom B
Classroom C
F 70
F 65
F 65
F 70
F 70
F 70
F 70
F 70
F 80
F 75
F 75
F 80
F 70
F 80
M 70
F 80
F 85
M 75
F 80
M 80
M 75
F 80
M 80
M 80
M 95
M 85
M 85
M 100
M 90
M 90
Class Mean
F 74, M 98, FM: 79
F 74, M 84, FM: 78
F 74, M 79, FM: 77
5Variables and Levels of Measurement
Categorical variables:
Dichotomous (e.g., male/female)
Polytomous (e.g., blood type)
Nominal (no implied order)
Ordinal (ranked, e.g., “good” > “fair”)
Continuous variables:
Interval scale (e.g., temperature in Celsius)
Ratio scale (e.g., body weight, height)
Note: Continuous variables can be converted to categorical, but not vice versa.
6Types of Research Design
Key concepts to distinguish studies:
Purpose: Descriptive vs. analytical
Investigator control: Observational vs. interventional
Directionality: Forward vs. backward
Sample selection: Based on exposure, disease, or neither