Methods Week 5 - Research Designs (1/2)

internal validity

the extent to which one can assume that a causal relationship exists between variables

how to determine temporal precedence

figure out a baseline or manipulate x variable

threats to internal validity

a term used to describe aspects of a study that leave open that possibility of an alternative explanation

how to limit selection threats

random assignment 

what is selection threat

comparing groups based on pre-existing or non-random criterion may be biased because the groups may differ systematically from one another in ways other than the key independent variable

what is considered a false experiments

any design that has a treatment condition but not a comparison group

what is a comparison group

ex. comparing a group who is taking a new drug to a group that is taking the golden standard. NOT A CONTROL because they are receiving smt

what is internal validity history threat?

History threat: other events may have happened between the

pretest and the posttest that explain the outcome. (ex. covid)

what is internal validity maturation threat?

Maturation threat: the participant themselves may have

changed between the pretest and the posttest. (ex. the changes a person has between the ages of 16 and 30)

what is internal validity testing/practice threat?

Testing / practice threat: people can improve just based on

prior exposure to a particular testing procedure / style. (ex. getting use to the type of test)

what is internal validity instrumentation threat?

Instrumentation threat: the outcome measurement tool itself

might change between pretest and the posttest.

(ex. different DSM mental illness criteria)

what is internal validity regression to the mean threat?

Regression to the mean: extremely high or low scores at pretest may be a statistical blip that stabilize at posttest (or vice versa). (ex. third variable that changes your outcome that has nothing to do with the study)

what is internal validity attrition threat?

Attrition: people may drop out of the study between the

pretest and the posttest.

necessities for an experiment

manipulation

random assignment

experimental control

what are some conditions in an experiment?

manipulating time (treatment after 1 month, 6 months, 12 months) or treatment type (treatment 1, 2, 3,4) or both (each month has 4 sub groups)

3 types of control

no treatment control condition: no intervention

placebo control condition: fake treatment

treatment-as-usual control condition: standard or alternative treatment (golden standard)

why use and active control?

placebo effect

reactivity: participants change bc they know their being watched

demand characteristics: participants expectations of the study biases their responses

random assignment approaches

Simple random assignment: a random process is used to assign a large number of participants to the conditions.

Block randomization: randomization occurs in blocks (e.g., ABCABCABC). (Ensures equal sample sizes).

Matched-groups design: participants are matched on important traits and then randomly assigned to conditions. (Use with small samples or potentially strong confounds).

design confounded constructs

Confounded constructs: the researcher unintentionally manipulates more than just the independent variable.

Observer biases / expectancy effects: the researcher’s expectations vary across the levels of the IV.

Demand characteristics / placebo effects: the participant’s expectations vary across the levels of the IV.

solution for controlling confounded constructs

researchers can use carefully constructed comparison groups and strict experimental controls to isolate causal mechanisms and rule out confounds 

Condition 1: arousal + pain

Condition 2: pain, no arousal

Condition 3: arousal, no pain

Condition 4: no arousal, no pain