Module 10- Non Experimental Research Flashcards
Closed System
- researcher is able to control for confounds and extraneous variables
- experimental research
- prioritizes internal validity
open System
- occurring in the real world with real people
- prioritizes more external validity
- observing the issue in it’s natural state
- influenced by many factors outside the researchers control
Reasons for doing field research
- open system
- test external validity
- study events that only occur in the field
- generalizability
Disadvantages of field research
- lack of internal validity bc not conducting a true experimental design
- no manipulation of the IV
- cannot place controls and control for confounding variables
- little control over participation assignment and selection
in absence of a true experimental design
- cannot make definitive causal statements bc of the absence of the aspects of a true experimental design
when we want tentative causal statements use,
Quasi Experimental Designs
- helps make internal validity as high as possible
Quasi Experimental Design
- make tentative causal statements
- approximates aspects of the experimental design and certain control aspects
- goal is to get as close to a causal conclusion as you can but cannot manipulate the IV or control participant assignment
2 types of quasi experimental designs
- non equivalent before after design
- time series design
for quasi experimental designs what is F equal to at the beginning of the study
- F will not equal 1
- this is because we cannot control assignment in this design therefore we know groups will differ on important aspects
- cannot control for confounds
- cannot randomly assign participants to groups
- groups differ prior to the study
The F ratio will have
- large between group differences prior to IV
- F is larger than 1
- makes it hard to interpret group differences after the IV bc the groups differed pre IV
the more similar the groups are at the beginning
- the more likely the F will be closer to 1 and more it will mimic a real experiment
Non equivalent Before after design
- pre measure the DV before the treatment
- treatment= Tx
- then create a change score (post minus pre)
- contain a comparison group who does not experience the IV (not a control group bc we are not controlling for assignment and confounds like in experimental design)
Group 1> Measurement 1> Tx> Measurement 2> calc change score
- do the same for group 2, but they dont experience the IV/ treatment/ Tx
want the comparison group to be
- as similar to the treatment group as possible
Time Series Design
- single group or population is tested before and after the intervention/ IV
Basic Time Series Design
- within subjects quasi experimental design
- no adequate comparison group
- using the same group of participants
- use this when all members of the population have been exposed to the IV or intervention
Interrupted Time Series Design
- includes several pre test and post test measures of the DV
M1 M2 M3 M4 > Tx >M5 M6 M7 M8 - gives us information of the normal fluctuations of the DV
- allows us to better interpret our treatment/ intervention in order to rule our threats to Internal Validity
Multiple Time Series Design
- Between subjects time series design
- address threats to internal validity such as confounds/ alternative explanations to the DV by having a comparison group
Group 1 M1 M2 M3 M4 >. Tx >M5 M6 M7 M8
Group 2 M1 M2 M3 M4> >M5 M6 M7 M8
Correlational Design
- even less control than quasi experimental design
- no controls at all ^ cannot make causal statements
- measuring variables how they exist in the real world
- loosing internal validity
- cannot say one variable CAUSES another
- no researcher manipulation/ intervention
what questions do correlational research answer?
- answer questions about the relationship between variables
- allow us to make predictions based on the relationship bw variables
- association bw variables in the real world
Correlation
- extent to which 2 or more variables are associated/ related to one another
- degree to which variables vary together
Coefficient of determination (r^squared)
- obtained by squaring the correlation coefficient
- proportion of variance of one variable explained by knowing the other variable
- allows us to make predictions
- the higher the value, the more accurate our predictions will be
p value will tell
- the significance of the correlation
Advantages of Correlational method
- real life settings
- can yield a large amount of data
- correlations used for predictions
- extended to include many variables
- cannot prove a theory but could negate a theory
correlations have high
- external validity bc variables are in their natural state
correlations have low
- internal validity
- no controls put in place to control confounding variables
- ^ not certain if the change in the DV is due to the IV alone
correlations underlie..
- casual relationships
- 2 variables cannot be casually related unless they are correlated
- if the variables are not correlated then we automatically know that they are not causally related
correlations does not equal causation because
- there are 3 underlying possible relationships between variables
3 underlying relationships between variables in correlations
- Direct causal link: variable A causes variable B
- Direct causal link: variable B causes variable A
- Spurious correlations
Spurious correlations
- usually the most likely answer out of the 3 underlying relationships
- both variables we measured are related to a third variable
- variable A and variable B are caused by a third variable
- makes it look like there is a relationship bw variable A and B
- unknown 3rd variable
experimenter expectancy
- researchers tending to only see what they expect to see
experimenter reactivity
- researchers unconsciously influencing participants
measurement reactivity
- participants responding differently because they know they are being observed
control expectancy and reactivity by
- objective measures
- limiting researcher contact
Moderator Variable
- variable that modifies the relationship bw variables
- gender moderates the relationship be alcohol consumption and academic achievement
Differential Research Methods
- compare 2 or more groups that are differentiated on a pre existing variable
- special form of correlational research
