Causality/Research Designs Flashcards
What functions do research designs have
- Exploratory data collection/ analysis
2. Hypothesis testing
Exploratory data collect /analysis
- classifying behavior for your study
- identifying potentially variables
- identifying possible relationship between behavior and these variables
- used in the early stage of a research
Hypothesis testing
looking for possible explanations for your observed relationship
What kind of relationships can you have ?
- Causal relationships
2. Correlational relationships
Causal relationships
one variable influences the other (is the cause for the other)
-direct/indirect influence
uni-direct (A—>B)
bi-direct (AB)
Correlational relationship
Two variables are associated
one change in a variable comes along with the other variable
- existing relationship ; But you can not make statements about causality
- —> covary
gives the basis for your explanations, but you have to test further if they are valid enough
e.g. baseball games/mosquitos: both increase in spring and fall in autumn —> not causing each other.
What is main characteristic of the correlational research
- non experimental design
that changes in one variable comes along with changes in the other variable = existing relationship
—> covary
- BUT you cannot see if they cause each other
- no manipulation of the variables
What are the problems with correlational designs?
- Third-Variable problem
2. Directional problem
Third variable problem
there could be a correlational relationship but only cause d by an unobserved third variable
- seemingly strong relationship but neither of the variables cause each other
Directional problem
that you can’t distinguish what causes what
When do you use correlational research?
- Collecting data at a early stage of research
- Inability to manipulate your variables
- unethical - Relating naturally occurring variables
What is the main Focus in the experimental design ?
that you have a high degree of control over your independent variables
-making predictions about causality
What are the two main characteristics in experimental design ?
- Manipulation of independent variable
2. Control over extraneous variables
independent variable
value is set/chosen by the experimenter (independent of the participant’s behavior)
Dependent variable
variable which value you will measure and observe
What are the levels of independent variable?
set values of the independent variable
What are the treatments of the experiment
specific condition for each level of
how do you manipulate the independent variable
exposing participants to at least two levels of the variable
What do you want to figure out with your manipulation?
That if you make changes in one variable this will cause changes in the other variable
Experimental group
Group that gets the treatment
Control group
Groups that doesn’t get the treatment
- Bassline for behavior comparison
What are extraneous variables
all variables that may affect your dependent variable (behavior) but are not part of your study
What is the problem with extraneous variables ?
may produce uncontrolled changes in your dependent variable
- difficult determination of effects of the independent variable = Lower external validity
- differences across the levels of the independent variable —> may seem that the independent variable have cause a change ( but in real not)
How can you control extraneous variables ?
- holding constant
2. Randomize across treatments = Radom assignment
What are advantages of the experimental design?
- you can make predictions and identify causal relationships
What are disadvantages of the experimental design?
no use if you cannot manipulate the variables
- increase in control —> decrease of external validity
What kinds of non-experimental designs do we have ?
- descriptive design
2. Correlational design
Descriptive design
Simple describing
used as a preliminary step tp understand a phenomenon
Not used to find associations between variables = one perspective
What is another form of experimental design ?
Demonstration
What is a demonstration?
group is exposed to only one treatment conditions
- measuring resulting behavior
! Not making predictions about causal relationships
Aim: To see what happens under specific conditions
and no effect of manipulations
What is the difference of a Quasi-Experiment to a true experiment?
+ looking for causal relationship
- no direct manipulation
What is the method of a quasi experiment ?
isolating the causal influence
not manipulating the IV but selecting cases when the variable varies naturally
What are advantages of a quasi experiment ?
good for testing generality
IV clear visible
- reducing error variance
= dividing data into groups of participants with similar response
What are the downsides of a quasi experiment ?
misinterpretations
– ! can never say if it is causal
What is a quasi variable ?
similar to an independent variable
difference: no manipulation
When do you use a quasi variable ?
when you do not random sampling
What are Developmental Designs ?
evaluating behavioral changes related to changes in peoples chronological age
- special quasi experiment
age = quasi variable - centre of interest is not to predict causal relationship
What types of developmental designs are available ?
- Cross Sectional Design
2. Longitudinal Design
What is a Cross Sectional design
if you take participants from different age groups
groups - based on chronological age
not measuring same participant in different ages
+ getting developmental data in short period of time
-Generation effect
What is the generation effect ?
influence on results by generational differences (e.g. education)
-observed effects= effects of age (confounding variable
damaging the internal validity
Longitudinal Design
a single group is followed over longer periods of time
Advantages of longitudinal designs
no generation effect
tracking developmental changes/behavior
Downsides of longitudinal designs
- Cross generational effect
- Participant/subject mortality
- Multiple observation effect
Cross sectional effect
findings cannot be applied to other generations
Multiple Observation effect
threat to internal validity
- Reactive testing
- history effect
What is the history effect ?
that other factors can come together with age (memory loss)
factors = confounding variable –> age = quasi-independent variable
what is a confounding variable
variable that varies along with IV
What is the problem with confounding variables
damaging the internal validity
What are solutions for to avoid confounding variables
- random sampling
- blind techniques- avoiding experimenter bias
- considering extraneous variables
- Good knowledge of literature
What is the characteristic of a quasi-independent variable
e.g. age - cannot be changed itself
no manipulation
