Quantitative study designs experimental (public health surveillance)

causality

criteria:

• the variables must be associated

• the cause must precede the effect in temporality time

• the relationship cannot be explained by no confounding third variable

Internal validity

trust the relationship between your variables within your study

• necessary before we can infer causation

• will depend on study design, implementation, and data analysis

additional threats to internal validity

• selection bias

• history

• testing

• maturation

• attrition

• instrumental changes

• compensatory rivalry/demoralization

• ambiguity about direction

selection bias

the control and the intervention groups aren’t as similar as they appear to be

• participant motivation level to change can be a big factor

• you can control for the variable you think of, there could be others

randomization

• creates equivalent groups based on probability theory

• controls for selection bias

• controls for confounding

• only relevant to experimental study designs

• not random sampling

history

• events that happen beyond the control of the researcher that can affect results

  • catastrophic events

  • political events

  • personal tragedies successes

testing

• repeatedly taking a test will make you better at the test

maturation

simply getting older, more experienced

• children

• life stages

• professional development

attrition

• AKA loss to follow up

• participants drop out before study completion

  • also form of selection bias if one group has higher attrition than the other

  • studies should plan on attrition when estimating sample sizes

  • incentives offered throughout long-term studies may reduce attrition

Instrument changes

• if the pre-test and post-test aren’t the same, are they really measuring the same thing?

• If the questionnaire or method for measurement used to collect data changes over time the researcher has to make assumptions about how to condense data

Compensatory rivalry and demoralization

• If groups are aware of each other…

  • They might become competitive, and one group might change more than the other (rivalry)

  • the control group might feel resentful and may have less commitment to remain in the study (demoralization)

Ambiguity about the direction of casual inference

• When timing of the independent variable and the dependent variable cannot be conclusively determined

  • participants who complete a drug treatment program were more likely to stay sober than those who dropped out

    • did they stay sober because they were in the program?

    • or did they complete the program because they were sober