Research Design (First Half)

Validity

whether our measurement tool is measuring what it’s supposed to

Face Validity

Test that measures what its supposed to at face value

Content Validity

Test that covers all relevant aspects of the concept

Criterion Validity

Test that correlates with a known standard

Concurrent Validity

Comparison with a gold-standard test simultaneously

Predictive Validity

Test is able to predict future performance 

Construct Validity

measures the theoretical concept it claims to 

Reliability 

Consistency; Test is reliable if it produces the similar results when repeated under the same conditions

Test-Retest reliability

Measures consistency over time

Alternate Forms reliability

Compares 2 different versions of the same test; useful for avoiding test-retest bias

Internal Consistency Reliability

measures consistency within a test

Inter-Rater Reliability

measures agreement between different testers

Odds Ratios 

• A measure of association between exposure and outcome

• OR = 1 - No association

• OR > 1 - exposure increase the odds of outcome

• OR < 1 - Exposure decreases the odds of outcome 

Sampling Bias

sample selected isn’t representative of the whole population

Selection bias

certain groups are more likely to be included than others

Volunteer bias

people who choose to participate may differ from those who don’t

Convenience sampling bias

sampling based on ease of access rather than randomness

Attrition/Loss of data Bias

When data is missing or participants drop out of a study

Assessment bias

when the way outcomes are measured is influenced by factors unrelated to the variables studied 

Observer bias

Person assessing outcome is influenced by knowledge of the participant group’s expectations

Recall bias

Participants are asked to remember past events and their recall is influenced by their current conditions/beliefs

Response bias

participants answering in ways they believe are expected or desirable.

Instrument bias

Tools/methods used to measure outcomes are inconsistent

Blinding

keeps assessors unaware of group assignment

Qual: ethnographic

immersion into cultures or communities to understand norms, values, and behaviors

Qual: Grounded Theory

generate theory from data rather than testing existing ones

Qual: Phenomenological

focuses on the lived experiences of individuals

Qual: Interpretive inquiry

understanding behavior in natural settings

Qual: Critical theory

focus on power, inequality, and social justice

Quan: Cross sectional study

where data is collected from many individuals at a single point in time to examine relationships between variables.

Quan: Longitudinal study

repeatedly studies the same individuals over an extended period of time

Quan: Case control study

observational study where Individuals with a condition are compared to individuals without the condition

Quan: Cohort study

observational study where a group of people with a specific condition are followed over a period of time

Quan: Pre-Post study

Study where an intervention is given to one group and the outcome is compared before and after treatment

Prospective controlled trials

Pre and post with comparison group; 1 group gets treatment, the other doesn’t; groups made in a non-randomized sequence

Randomized controlled trials

Randomly placed into treatment and comparison

Conventional RCT

Participants placed into two or more groups

Stratified RCT

sample size placed into two subgroups then studied conventionally

Crossover RCT

participants gets both treatments in a sequence, with a washout period in between

Cluster RCT

entire groups are randomized