Week 2: Data Collection

the 4 types of population used to collect data

Target population, source population, sample population, study population

target population

- broadest group
- the general population that the study seeks to understand

Source population

- specific individuals from the target population from which a sample will be drawn

sample population

- the Individuals selected to participate from the source population
- If the source population is small, all might be asked to participate

study population

- The eligible members from the sample that consent to participation

sampling bias

occurs when individuals selected for the study do not represent the source population as a whole

nonrandom-sampling bias

occurs when each individual in the source population does not have an equal chance of being selected for the sample population

probability based sampling

Methods for ensuring that members of a source population have an equal likelihood of being invited to participate in a research study
different types:
- simple random sampling
- systematic sampling
- stratified sampling
- cluster sampling

Simple random sampling

Completely random selection from a population
- eg. 12 out of 36 people selected

Systematic sampling

Every nth person, usually with a random start

Stratified sampling

Random sampling from distinct groups (stratum),
eg. 3 people from each sex or geographic location

Cluster sampling

Natural clusters, rather than individuals selected
eg. eligible children within each school

Multistage sampling

1st primary sample units are selected eg. 10 clusters in a municipality, then from there every nth person is selected

Non-probability based sampling

Convenience sampling
Purposive sampling

Berkson's bias

can occur when cases and controls for a study are recruited from hospitals and therefore are more likely than the general population to have comorbid conditions

Healthy Worker Bias

workers, on average, are healthier than the general population (Occupational populations)

Exclusion Bias

applying different eligibility criteria to cases and controls

Selection Bias

occurs when the members of a study population are not representative of which they are drawn

nonresponse bias

bias introduced when a large fraction of those sampled fails to respond

Nested case-control study

- cases and controls are drawn from the population in a cohort study

eligibility criteria

The characteristics that define the population

key informants

individuals selected to participate in a qualitative study because they have expertise relevant to the study question

purposive sampling

selecting sample members to study because they possess attributes important to understanding the research topic

data saturation

sampling until no new information emerges

vulnerable populations

Groups of people with diminished autonomy who cannot participate fully in the consent process. Such groups may include children, individuals with cognitive disorders, prisoners, and marginalized populations.

confidence interval (CI)

Given a sample from a population, the CI indicates a range in which the population mean is believed to be found. Usually expressed as a 95% CI, indicating the lower and upper boundaries.
- narrow CI = more certainty about the value of a statistic
- if all people n the total population are included, no CI is needed

sample size calculator

a statistical program used to provide a recommended sample size - can find online

type 1 error (alpha)

False Positive
- a significant result when no significant result exists
- most studies use alpha = 5%

Type 2 error (beta)

False Negative
- no significant result, when it actually does exist
- usually referred to as the power of the study
solution = bigger sample size

Power

The ability of a test to detect significant differences in a population when they actually exist.
- power = 1 - Beta

Questionares

A series of questions used as a tool for gathering data

Likert Scale

a question that asks participants to rank preferences numerically, such as 1 for strong disagreement and 5 for strong agreement

habituation

error that occurs when participants completing a questionnare become so accustomed to a response, they continue without giving their true perspective

Filter or contingency questions

a question you ask the respondents to determine whether they are qualified or experienced enough to answer a subsequent one
- online a skip logic can be used to hide irrelevant questions

internal consistency

items in a survey measure various aspects of the same concept

Cronbach's alpha

the measure of internal consistency
- KR - 20

Inter-observer agreement/ inter-rater agreement

The degree to which two or more independent observers report the same observed values after measuring the same events

content validity/ logical validity

researchers agree that the content is valid

Face validity

researchers agree content is easy for study participants to understand and complete

construct validity

measures the theoretical construct of an assessment
- can be an empirical test

convergent validity

is present when two items that the underlying theory says are related, are in fact related
- the opposite = discriminant validity

Criterion or concrete validity

uses an established test to validate your own test with a similar theoretical construct

concurrent validity

participants in a pilot study complete both tests and a strong correlation between both tests is evident

Predictive validity

could be demonstrated by administering a test to incoming med students and comparing their results on the new test with those from their initial licensing exam