WEEK 2 Flashcards
Evan’s-Stoddart Model
- social + physical + genetic environment, then the disease
- we treat the disease, but forgetting to consider the “environments” will undo previous work.
Eco -Social Model
- best for research
How is data usually collected in Health Sciences?
sampling, except for census.
Census
include everyone.
Population Levels
- target population (general public your study applies to)
- source population (share specific characteristics)
- sample population (people you approach to take part)
- study population (people who agree to take part)
Probability Sampling
probability of a sampling unit to be selected.
What are the different probability sampling types?
- simple random sampling
- stratified sampling
- cluster sampling
- multistage sampling
Simple Random Sampling
- no bias
- completely random
- may not get an accurate representation of the target population.
Stratified Sampling
- random samples from distinct groups created from shared /similar characteristics. (geography, sex)
- people are randomly selected from the groups.
Cluster Sampling
- natural clusters are formed
- all units in a cluster are used.
- randomly select the clusters, not the people within the clusters.
Multistage Sampling
- primary units are selected
- the secondary units are selected from the primary units.
- no bias
What are the non-probability-based sampling methods?
- convenience sampling
- purposive sampling
Convenience Sampling
- recruitment is based on accommodating access for individuals, schools, and communities.
- does not produce a representative sample.
What are 2 errors that can occur in sampling?
- external validity
- sampling bias
Purposive Sampling
- recruitment of people based on the information they can provide.
- key informants: chosen so they have expertise relevant to the study topic.
Vulnerable Populations
- limited ability to make decisions about volunteering to participate
- young children, individuals with health issues, prison, socially marginalized people, and pregnant women.
- should not be selected unless necessary
Sample size importance?
- you want the right amount so resources arent wasted and results remain valid.
- smaller error = larger sample size
- The effect of exposure on outcome = smaller sample size.
What are the types of error?
- Type 1 alpha error: there is a relationship within the study population that does not exist in the source (irl)
- Type 2 beta error: there is no relationship within the study population that does exist in the source (irl)
- both errors have both the null and alternative hypothesis as correct.
power
1-beta error
- determines the ability to detect a true relationship
Data collection in health
- interviewing
- questionnaire
- survey instruments
- sampling tissue
- blood work
- direct measurement of physical functioning
- imaging
Interviews
- verbally asking questions and recording responses
- trained interviewers = ensure accuracy and completeness
- major time commitment
- open-ended
- semi-structured = illicit responses
- can see body movements
Probing
- prompting an interviewee to expand on an answer to provide a completed response
Self-Administered Survey V. Interview
Interview:
- can be trained to ensure accuracy and completeness
- time commitment
- expensive
Self-Administered Survey:
- time and cost-efficient
- reaches a large number of people
- honest answers to sensitive questions
- problem with populations of low literacy
- people with limited access to the Internet.
Interviewer Bias
- the interviewer asks different questions for the cases and controls.
- probing only individuals they suspect of having more information
Questionnaire
- series of questions to gather data
- the first set of topics allows the researcher to determine the eligibility
- close-ended questions only allow for limited responses
(nominal = no built-in order…ordinal = inherently ordered questions) - open-ended questions allow for unlimited answers.
Things to consider in a questionnaire (7)?
- Order of questions
- Getting the correct answer
- Data recording methods
- Layout and formatting
- SKIP logic codes
- Back translation
- Pilot testing
