Lecture 11: Causation and Longitudinal Design

imply

a suggestion, hint, or insinuation; an indirect statement

infer

drawing a conclusion from facts etc.

implications are

embedded in reports (studies)

we make inferences

from research (evidence)

causation

if we understand cause-effect relationships, we can manipulate them (develop interventions/treatments and make more precise predictions)

most research aims to understand/investigate

some sort of causation among variables

consider the elements of (scientific) methods

operational definitions, constraint/control, statistical significance (power, effect size)

consider the design of investigations

single/multiple groups?; experimental/other?; cross-sections vs. longitudinal design?

consider the data analysis

correlational, regression, or tests of differences

consider the validity of the investigation

were constructs and operational defs representative?; can results be replicated?; are results consistent with similar investigations?; are results generalizable?

observations —> consistency —> theory…………

“Laws”

research methods, design, and data analysis are used to

imply causal relations

cross-sectional designs

rich history; analyses from many studies were based on data collected only at one time point

longitudinal research designs enable us to view

relationships/differences over time because data are collected at various time points

data analysis over time

helps us control for time effects, provide a stronger basis for causal inference, and allow us to better see trends in relation to change

cons to longitudinal designs

we lose cases, they typically drop out over long periods of time; costs more to track participants (staff), incentives (multiple assessments = multiple pay); and typically results in small samples

cohort

examining a group (students, patients, children, etc.) based on desired characteristicsc

clinical trials

examining (treatments) over time

randomized clinical trial

examining treatments over time with random assignment of participants

randomized controlled trial

examining treatments over time with random assignment of participants, but with a control group

pre-test/post-test design (single group)

measuring participants before exposure to the IV (“pre”) and then after exposure to the IV (“post”)

pre-test/post-test design (2 groups)

providing a comparative (control) group to control for pre-test effect… but how do we know if affects are from pre-test effects of the treatment (IV)

Solomon 4 group design

Controls for: pretest (practice) effects; Confounding effects: treatment/practice effects; Highest level of scientific rigor/high constraint