Research and stats design Flashcards
What is sampling error?
The difference between a sample mean and a population mean OR sampling error is the error caused by observing a sample instead of the whole population.
What is the best way to reduce sampling error?
Random sampling
What is one of the error terms of a stats procedure to account for sampling error?
standard error of mean, standard error of estimate, mean square errors, standard error of regression coefficient.
What is purposive sampling?
A deliberate process to identify population characteristics such as participants and treatments and then select a sample that embodies the desired population characteristics
What are the 2 sources of error?
Sampling error and error of measurement
What can error of measurement be attributed to?
Systematic error and random error
What is systematic error?
Errors that reduces the validity of the experiment. Usually due to wrong calibration of equipment, irregularities of research process.
What is random error?
Individual differences like mood (things u can’t control like temp) . Affects reliability of experiment.
What are the ways to control extraneous variables (EV)?
- Eliminate the EV
- Build the EV into the design
- Statistically controlled (using covariance and multiple regression)
- Matching participants (participant from group A matched with e.g. same height and weight from group B)
- Random assignment
- Blinded procedures (single, double, triple, partial - researcher only knows about the treatment right before administering)
- Deception
How many threats are there to internal validity? What are they?
13
THIS MESS (rule out during design). DREAD (stats corrected)
Testing, history, instrumentation, maturation, experimental mortality, stats regression and selection maturation interaction.
Diffusion of experimental , Rivalry, equalisation of treatments, ambiguous temporal precedence, demoralization.
What are the 7 ways to counteract DREAD?
- Use different persons for each treatment condition
- Arrange treatment conditions so contact between control and experimental minimised
- Using single, double, triple blind - no bias
- Debrief participants to assess expectations and experiences
- Not using rewards
- Clearly define treatment conditions
- Use past research to guide causal
How can we improve quasi experiment?
- adding control or comprison group
- adding pretests and posttests
- removing and reinstating treatments
- adding replications
- reversing treatments
- case matching
What is the statistical technique use in case matching (of quasi experiments) to reduce selection bias?
Propensity scores (matching)
What is regression discontinuity designs?
Using cutoff scores to assign participants to a treatment
What is:
- Bivariate correlation & simple linear regression?
- Multiple correlation & multiple linear regression?
- the equation for multiple linear regression?
- r/s between 2 variables. Regression that uses a bivariate correlation to predict the criterion variable (dependant variable)
- r/s between more than 2 variables
- Y = A + B(1)X(1) + B(2)X(2)
Y = criterion variable (DV)
A = intercept
B = regression coefficient
X = Predictor criterion (IV)
