PSYC 1001 Lecture Notes - Lecture 2: Confounding, Blind Experiment
27 Sep 2016
School
Department
Course
Professor
Random Assignment: each participant has an equally likely chance of being assigned to any of
the groups in the study
You could never measure all individual “quirks” from each participant
Balanced out through
GET NOTES FROM LAUREN!!!!
Confounding of Variables
- No way of knowing the connection between variables
Correlational design example:
- Study using questionnaires
- One about happiness
- Another about relationship status
- Confound: questionnaire ALWAYS presented in the same order
oResponse may be due to “order effect”
ie. if relationship status comes first: cause your answers on happiness can become bias
because you’re thinking about your good relationship
oYou’re having a bad day but the relationship questionnaire brings up positive
feelings
ie. happiness comes first: bias
oAwesome things happening to you, great mood –
oFeel bad that you’re talking about how happy you are that you say it’s not “that
bad” on your relationship status
Counterbalancing: varying the order evenly (or randomly)
oControls for the confounding variable of “order” (order effect)
oSolution to this specific confound
Designing an Experiment
Problem: Servers want to maximize tips
- What factor might improve tipping?
oPersonal introduction (name)
Hypothesis: When the server introduces themselves by name (as part of their
introduction/welcome), customer tips (as a percentage of the total bill) will increase
Design the study:
- Independent Variable: whether or not the server introduces themselves
- Dependent Variable: the value of the tip as a percentage of the total bill
Control for restaurant
- Fine if it’s one restaurant
- Introduce or not WITHIN each restaurant if comparing between more than one
restaurant
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Random assignment: each participant has an equally likely chance of being assigned to any of the groups in the study. You could never measure all individual quirks from each participant. No way of knowing the connection between variables. Counterbalancing: varying the order evenly (or randomly: controls for the confounding variable of order (order effect, solution to this specific confound. What factor might improve tipping: personal introduction (name) Hypothesis: when the server introduces themselves by name (as part of their introduction/welcome), customer tips (as a percentage of the total bill) will increase. Independent variable: whether or not the server introduces themselves. Dependent variable: the value of the tip as a percentage of the total bill. Introduce or not within each restaurant if comparing between more than one restaurant. Randomize, flip a coin to decide whether you introduce or not. Some are good tippers; recognize how hard it is to be a server.
