Inhoud
Week 2................................................................................................................... 2
Week 3................................................................................................................... 9
Week 4................................................................................................................. 13
Week 5................................................................................................................. 19
Week 7................................................................................................................. 20
,Week 2
Experiment is good since it uses randomization: Arbitrarily assigning each participant to one condition
of the experiment
Experiments are a great way to determine causality, which is usually what we’re interested in. Does a
change in X cause a change in Y?
How to create a good manipulation
1) Try hard to make the manipulation “clean”
- only manipulate your construct, hold everything else constant.
- That is, avoid “confounds”
2) Try to make the manipulation strong (especially if just doing 1-2 studies)
3) Test if it works
- Do a pretest
- Measure a “manipulation check” (in pretest, and in main study)
• - Steffel and Williams: Choice difficulty vs. How easy is it to read?
How to create a good pretest
1) Make it just like the study, same manipulation, intro, etc.
2) Measure the manipulation check
3) Can try multiple manipulations and see which works best
4) Can also measure potential confounds
With experiments, it is good practice to
include manipulation checks
If possible, there is a clear benefit to doing this in a pretest:
1) You quickly and cheaply test if your manipulation “works” as intended
2) Your measure of the manipulation/confound check cannot affect your DV when
(only) tested in the pretest
e.g., I have a manipulation and want to test how believable people think the
manipulation is. However, by asking people whether they believe my story,
they suddenly focus on this aspect.
In the context of research and experimental design, confounds (or confounding variables) are factors
that can influence the outcome of an experiment or study in addition to the variables that are being
directly studied. Confounds can make it difficult to determine whether the observed effects are truly
due to the variables of interest or if they are influenced by these extraneous factors. Essentially,
confounds can create uncertainty about the causal relationships between variables.
, Key Characteristics of Confounds
1. Correlation with Independent Variable: Confounds are related to the independent variable
(the variable that is being manipulated or categorized by the researcher).
2. Influence on Dependent Variable: Confounds also affect the dependent variable (the variable
that is being measured or observed).
3. Confounding Effect: Because confounds are related to both the independent and dependent
variables, they can obscure or distort the true relationship between these variables.
A big plus of doing experiments is that it gets rid of a lot of confounds
With true random assignment,
1) There is no self-selection to conditions,
e.g., research suggested that people who drank alcohol regularly were healthier
But this is because older people go out (and drink) less, and older people are typically less healthy.
So why not simply control for age?
because it might also (partially) be due to lonelier people going out for drinks
less (and lonelier people tend to be less healthy)
because people with less money might go out less (and are less healthy)
Does age have a linear effect? Or curvilinear? Or ... ?
With true random assignment,
1) There is no self-selection to conditions,
2) There are less possible time/location/sample effects
e.g., if we ask a group of people to taste chocolate, and then see that those who
tasted chocolate buy more of it, there is a confound:
those who like chocolate are more likely to taste it as well as buy it.
Not only run a manipulation check but also a confound check
In the context of research, particularly in experimental studies, a "manipulation" refers to the
deliberate alteration or control of an independent variable by the researcher to observe the effect on
a dependent variable. This process is a cornerstone of experimental design and is used to establish
causal relationships between variables.
Confound Check
Purpose: A confound check is conducted to identify and control for extraneous variables that could
potentially influence the relationship between the independent and dependent variables.
Confounding variables can obscure the true effects of the independent variable on the dependent
variable, leading to inaccurate conclusions.
Process:
1. Identifying Confounds: Researchers assess potential confounding variables that could be
associated with both the independent and dependent variables.
Week 2................................................................................................................... 2
Week 3................................................................................................................... 9
Week 4................................................................................................................. 13
Week 5................................................................................................................. 19
Week 7................................................................................................................. 20
,Week 2
Experiment is good since it uses randomization: Arbitrarily assigning each participant to one condition
of the experiment
Experiments are a great way to determine causality, which is usually what we’re interested in. Does a
change in X cause a change in Y?
How to create a good manipulation
1) Try hard to make the manipulation “clean”
- only manipulate your construct, hold everything else constant.
- That is, avoid “confounds”
2) Try to make the manipulation strong (especially if just doing 1-2 studies)
3) Test if it works
- Do a pretest
- Measure a “manipulation check” (in pretest, and in main study)
• - Steffel and Williams: Choice difficulty vs. How easy is it to read?
How to create a good pretest
1) Make it just like the study, same manipulation, intro, etc.
2) Measure the manipulation check
3) Can try multiple manipulations and see which works best
4) Can also measure potential confounds
With experiments, it is good practice to
include manipulation checks
If possible, there is a clear benefit to doing this in a pretest:
1) You quickly and cheaply test if your manipulation “works” as intended
2) Your measure of the manipulation/confound check cannot affect your DV when
(only) tested in the pretest
e.g., I have a manipulation and want to test how believable people think the
manipulation is. However, by asking people whether they believe my story,
they suddenly focus on this aspect.
In the context of research and experimental design, confounds (or confounding variables) are factors
that can influence the outcome of an experiment or study in addition to the variables that are being
directly studied. Confounds can make it difficult to determine whether the observed effects are truly
due to the variables of interest or if they are influenced by these extraneous factors. Essentially,
confounds can create uncertainty about the causal relationships between variables.
, Key Characteristics of Confounds
1. Correlation with Independent Variable: Confounds are related to the independent variable
(the variable that is being manipulated or categorized by the researcher).
2. Influence on Dependent Variable: Confounds also affect the dependent variable (the variable
that is being measured or observed).
3. Confounding Effect: Because confounds are related to both the independent and dependent
variables, they can obscure or distort the true relationship between these variables.
A big plus of doing experiments is that it gets rid of a lot of confounds
With true random assignment,
1) There is no self-selection to conditions,
e.g., research suggested that people who drank alcohol regularly were healthier
But this is because older people go out (and drink) less, and older people are typically less healthy.
So why not simply control for age?
because it might also (partially) be due to lonelier people going out for drinks
less (and lonelier people tend to be less healthy)
because people with less money might go out less (and are less healthy)
Does age have a linear effect? Or curvilinear? Or ... ?
With true random assignment,
1) There is no self-selection to conditions,
2) There are less possible time/location/sample effects
e.g., if we ask a group of people to taste chocolate, and then see that those who
tasted chocolate buy more of it, there is a confound:
those who like chocolate are more likely to taste it as well as buy it.
Not only run a manipulation check but also a confound check
In the context of research, particularly in experimental studies, a "manipulation" refers to the
deliberate alteration or control of an independent variable by the researcher to observe the effect on
a dependent variable. This process is a cornerstone of experimental design and is used to establish
causal relationships between variables.
Confound Check
Purpose: A confound check is conducted to identify and control for extraneous variables that could
potentially influence the relationship between the independent and dependent variables.
Confounding variables can obscure the true effects of the independent variable on the dependent
variable, leading to inaccurate conclusions.
Process:
1. Identifying Confounds: Researchers assess potential confounding variables that could be
associated with both the independent and dependent variables.
