03 - research designs Flashcards
Observational - define
When occurrences or changes in the independent variable occur as a result of natural history
Experimental - define
When occurrence or change in the IV is manipulated by the investigator
Synonym of observational studies
Nonexperimental study
Investigator decides who gets exposure (not the subjects)
Exposure is deliberately manipulated, intervention
May be done on individuals or groups/populations
Experimental studies
Types of observational designs (4)
- Ecological studies
- Cohort studies
- Case-control studies
- Cross-sectional studies
Retrospective - define
Assessment of the previous occurrence of dependent variable in an attempt to reconstruct an influencing factor
Prospective (longitudinal) - define
When the DV is observed overtime
Characteristics of cohort studies (4)
- Studies of individuals
- Begin by classifying subjects on the basis of their exposure
- Subjects then followed up to determine the rate of the study outcome
- May be prospective or retrospective
Characteristics of case-control studies (3)
- Study of individuals
- Begin by classifying subjects on the basis of their outcome
- Then exposure is retrospectively ascertained for cases and control
What are cases?
Subjects with the outcome of interest
What are controls?
Subjects without the outcome of interest
Characteristics of cross-sectional studies (3)
- Studies of individuals (ex: population survey)
- Exposure and outcome status are simultaneously ascertained in all subjects
- Because exposure and outcome are measured at the same time, it is often difficult to differentiate between cause and effect
Advantages of cohort studies (2)
- Provides an absolute measure of risk
- Allows the study of multiple disease outcome
Disadvantages of cohort studies (5)
- Can be expensive
- Time consuming
- Ineffective for rare outcomes
- Loss to follow-up
- Can only assess risk factors at baseline
Advantages of case-control studies (3)
- Quick(ish) and easy
- Can be used for rare events
- Can control for certain aspect
Disadvantages of case-control studies (4)
- Cannot determine absolute risk
- Subject to recall bias
- Specifically associated with 1 factor/disease
- Temporal relationship uncertain
Advantages of cross-sectional surveys (2)
- Quick and easy
- Generates hypothesis
Disadvantages of cross-sectional surveys (3)
- No temporal relationship
- Cannot define causation
- Normally requires larger sample sizes
3 types of experimental design (3)
Pre-experimental designs
Natural/quasi-experiments
True experiments
Weakest degree of experimental control
No random assignments of participants
Control very few sources of invalidity
Pre-experimental designs
3 types of pre-experimental designs
- One-shot design
- One-group pre- and post-design
- Static group comparison
- People are in a group for natural reasons they are not randomly assigned (ex: class, sports team)
- An independent variable is introduced in a real world setting or occurs by chance
- There is no randomization so groups may not be equivalent
Natural (quasi-) experiments
3 types of quasi-experiments
- Pre and post design
- Repeated measures
- Time series
In randomized designs, the only difference between groups prior to treatment is due to ___ or __
Chance or sampling error
The estimate of sampling error is called ___
Error
The effect of the independent variable is called ___ or ___
Treatment or experimental variance
What is the minimax principle?
Demonstrates how to minimize error variance while maximizing treatment variance
Minimax principle is done through what? (3)
- Randomization
- Sample size
- Use of a sound research design
Gold standard in research
Randomization of participants + control group
True experimental designs
2 types of true experimental designs
Post-test only
Pre- and post-test design
What is the main effect - TED
Effects of each independent variables
