Class Notes (1,100,000)
CA (630,000)
UTSC (30,000)
Psychology (8,000)
PSYB01H3 (200)
Lecture 6

PSYB01H3 Lecture Notes - Lecture 6: Confounding, Intravenous Therapy, Diet Pepsi


Department
Psychology
Course Code
PSYB01H3
Professor
Connie Boudens
Lecture
6

This preview shows pages 1-3. to view the full 11 pages of the document.
PSYB01
Lecture 6
1
Experiments
Example to demonstrate importance of experimental design:
taste test comparing diet Pepsi and diet Coke
Pepsi was in a bottle that was opened prior to air it out, it was also warm & not fizzy
Coke was just bought minutes before the experiment and had ice in it = cold and
refreshing
problems with experiment:
o the experimenter was hinting at which one would taste better and be more
refreshing
o the identity of the products was not concealed could have influenced which one
the participant liked better
o the two products should have been presented in exactly the same fashion the
only thing that should be manipulated is the brand of the products they should
both be at the same temperature, carbonation, in the same container, etc.
Problems with establishing causality in correlational research
if you want to look at the relationship between two variables you can just collect the
information on them and look at the correlation coefficient - strong positive, strong
negative, no relationship
some other variables shown to be related to each other:
o eating breakfast and academic performance of school aged children if children
eat a good breakfast their academic performance is better
o eating disorders and watching soap operas in teenage females teenage females
who watch soap operas are more likely to have an eating disorder than teenage
females who don‟t watch soap operas
o these things are found to be correlated with each other BUT doesn‟t mean that one
things cause another
o for example, if you eat breakfast you are maybe doing better in school because
you‟re better able to pay attention because you‟re not hungry – can focus you‟re
energy on learning instead of listening to your stomach rumble
o correlational results ONLY tell you if two variables are related to each other
can‟t tell you in what way they are related or which one causes the other or if one
of them actually causes another or if there is a third variable at play
o in order to answer causal questions experiments must be done!
Direction of influence problem
o Class attendance and grades strong correlation between the number of times
people attend class and what their grades are like
o Most people will think if students go to class more often this will result in higher
grades but this may not be the case
o It could be that class attendance results in a higher grade BUT it could also be that
people who get good grades are more likely to go to class
o Higher grades makes them want to be in class more because they get good
feedback, feel good about being in the academic environment = therefore get
higher grades
o Question of directional causality which one is actually causing which

Only pages 1-3 are available for preview. Some parts have been intentionally blurred.

PSYB01
Lecture 6
2
Third variable problem
o Other problem with inferring causality from correlational results
o There is third variable that is causing a change in both of the variables you
correlated
o Example: research shows that positive mood and charitable donations are related
to each other positively correlated
May be a third variable involved causing those things
Having just gotten paid can cause a change in both of these variables
having money in hand puts them in a good mood and also makes it more
likely they will make a donation
o Example: correlation between ice cream sales and drowning
But actually a third variable temperature is causing a change in both of
these variables
When temperature is really high ice sales go up and more people go
swimming the more people go swimming the more people drown
o Sometimes can have a direction of influence problem and a third variable problem
Depressed mood and impaired sleep
Those who suffer from depressed mood have a lot of sleep disturbances
It could be that depression is causing disturbed sleep or vice versa
question of direction of influence
But it could also be that family problems ( a third variable) is causing both
impaired sleep and depressed moods
Other stresses can also cause a change in both of these variables
For example: just because every time there is a cat there is food all over the place doesn‟t
mean the cat is putting food all over the place
Three things needed to establish causality via experiments must include the following factors
Temporal order must be correct
o Cause has to come before the outcome
o In correlational research unless it has some longitudinal component where you
can tell what happened first you can‟t establish what came first
Variables have to covary
o As one changes the other must change ex. as one goes up the other one goes up
No other variable is causing the outcome
o For example, in the taste test must make sure that the only difference is the drinks
themselves
Keep all extraneous variables the same such as the temperature of the drink, how it was
presented, what the experimenter says
If all extraneous variables are controlled they have still have an influence but the
influence on all participants is the same
If there‟s anything you can‟t hold constant or control solution = random assignment; on
a random basis you assign participants to one condition vs. another via random numbers
table/pulling numbers out of a hat
o If you have two conditions you have to put participants in = put the number 1 and
2 in a hat then draw one number and put it back in the hat, draw another number

Only pages 1-3 are available for preview. Some parts have been intentionally blurred.

PSYB01
Lecture 6
3
put it back in the hat, etc. over time should draw both of the numbers equal
amount of times
o Every participants has equal chance of being put into one condition vs. another
condition
In experiments always present the independent variable first this takes care of the
temporal component independent variables causes dependent variable so it is always
presented first
Present the independent variable carefully and consistently, measure the dependent
variable rigorously
o Taste test present drinks in exactly the same way ( same temperature, cups, etc.)
and say exactly the same thing to participants and take care of anything else the
participant might react to in the experiment
o Dependent variable preference for drink be careful about the measurement and
operationalize the dependent variable in a meaningful way
Extraneous vs. confounding variable
Confounding variables are related to the independent variable they vary along with the
independent variable, to the point that its‟ hard to see their separate effects
Extraneous variables are anything other than the independent variable that could have an
effect on the dependent variable
Confounding variables are a subset of extraneous variables type of extraneous variables
All confounding variables are extraneous variables but not all extraneous variables are
confounding variables
Examples:
o Temperature of drink confounded with the type of drink one drink was warm
and one was cold
o If a man does something he has to apologize for, does the gazing behaviour
predict if he will be forgiven
Length of relationship confounded with gaze behaviour Kobe Bryant got
accused of sexually assaulting another women but denied the rumours and
said it was consensual and apologized to everyone including wife in a
press conference his wife was gazing at him at the entire time
Gazing at him with a loving expression does the amount of gazing time
predict whether a women will forgive him?
Can we estimate the probability a women is going to forgive a man based
on how much time she spends looking at him during the press conference
where he apologizes
BUT confound between length of relationship and gazing
Newer couples spend more time looking at each other compared to
couples who have been together longer - therefore gazing behaviour may
not be good way to predict if she‟ll forgive him
How experiments allow you to do this
Hold all variables constant
Use random assignment
Present i.v. first
You're Reading a Preview

Unlock to view full version