Experimental Design

explanatory variable

a variable that is manipulated or categorized to observe its effect on the response variable in an experiment

response variable

the outcome measured in an experiment to assess the effect of treatments

why is blind/double-blind important

Blind or double-blind designs are important because they help reduce bias, ensuring that treatment effects are observed without the influence of participant or researcher expectations

completely randomized design

when you randomly assign subjects to different treatment groups to make sure the results are not biased

sample → random assignment → group 1 & group 2

why is random assignment beneficial

a random assignment allows us to conclude a treatment causes changes in the response variable

treatment

the specific condition or thing that is applied to the subjects in an experiment— given to the treatment group

experimental unit

who or what we are assigning to a treatment

blind

when the subject doesn’t know which treatment they got

double-blind

when the subject and experimenter/researcher doesn’t know which treatment was assigned

confounding variable

something extra that you didn’t mean to include, but it affects your results

ex: you want to see if studying more makes test scores go up.
but the students who study more also sleep more.

so you don’t know if studying or sleeping caused the better scores.
→ sleep is the confounding variable.

placebo effect

when a person thinks they’re getting a real treatment, and they feel better or different, even though they got a fake treatment

block

groups of experimental units that are similar

randomized block design

separate subjects into blocks (based on variables they know might impact the results of the experiment) and then randomly assign treatments within each block

subjects → blocks → random assignment for each block → create however many groups needed → compare effectiveness for groups → combine and compare for blocks

matched pairs design

• subjects are paired (block of size 2) and then randomly assigned to a treatment

• each subject receives two treatments

  • order of the treatments must be randomized

EX: “Pair the top two students, then the next two students, and so on (25 pairs). For each pair, toss a coin. If heads, the first student in pair gets the [treatment 1]. If tails → treatment 2

random assignment and random sampling

Can determine causal relationship in population. This design is relatively rare in the real worl

random assignment but not random sampling

Can determine causal relationship in that sample only. This design is where most experiments would fit

not random assignment but random sampling

Can detect relationships in population, but cannot determine causality. This design is where many surveys and observational studies would fit

no random assignment and not random sampling

Can detect relationships in that sample only, but cannot determine causality. This design is where many unscientific surveys and polls would fit