exam 2 Flashcards
What is the definition of operational definition?
- definition of a concept that specifies the operation required to manipulate or measure the concept.
- applies to both IV and DV
How would you choose your independent variable levels?
- three or more levels allow you to observe non-linear relationships.
- Conduct pilot studies to choose your levels.
What are things you should watch out for in DV?
- Try to use established measurements
- strive for interval or ration data
- Reliability- extent to which measures of the same phenomenon are consistent and repeatable.
How do you measure reliability?
- Calculate the correlation coefficient
- 1 test-retest reliability
- 2 inter-rater reliability
What is the validity of a measurement?
- the extent to which we are measuring what we want to measure.
How do you measure validity?
- predictive validity- degree to which the measurement predicts future behavior.
- Concurrent validity- degree to which the measurement correlates with an established measure. (use more the 1 type of DV)
If your measure is reliable, is it also valid?
- Not necessarily
If your measure is valid, is it also reliable?
- Yes: if a measure is unreliable, it must automatically be considered invalid.
What is a between subject design?
- Each participant is exposed to only ONE level of the IV
What are the advantages of BS design?
- No carry over effect
2. used for subject variables
What are the disadvantages of BS design?
- possibility of non-equivalent groups
- 2 Solutions- large group size
- random assignment to groups
What is a WS design?
- Each participant is exposed to ALL of the levels of the IV.
What are the advantages of WS design?
- observed effects cannot be due to individual difference.
- Fewer participants required
- More powerful statistics
What are the disadvantages of WS design?
- Cant be used for subject variables
2. Possible carryover effects
What are some solutions to the carryover effect?
- Randomize the order of trials
2. Counter balancing- ordering the presentation of levels so as to minimize or eliminate confounds due to order effect.
What are different examples of counterbalancing?
- Complete counterbalancing- use every possible order of conditions
- Latin Square- a procedure in which each condition appears at every position in the order equally often
When to use a BS design?
- for subject variables
- DV is a permanent change
- when you are especially concerned about participant bias.
Precautions for BS designs?
- random assignment
2. use large number of participants
When should you use a WS design?
- WHen none of the previous factors apply
2. when you measurement is very sensitive to individual differences
Precautions for WS designs?
counterbalance or randomize the order of conditions
What are matched group desings?
- BS design with non-random assignment
2. Uses same statistics as WS design
What are the advantages of MD design?
- decreases likelihood of non-equivalent data
2. greater statistical power
What are the disadvantages of MD?
- Time consuming
What are some important characteristics of experimental design?
- the number of IV
- the number of levels for each IV
- Whether each IV is WS or BS
- scale of measurement of your DV
What type of design would you use for 1IV with two levels?
- t-test
- 1 BS-> independent group t-test
- 2 WS or MD-> paired group t test
What are the advantages of single variable design?
relatively simple
what are the disadvantages of single variable design?
- difficult to determine appropriate range of IV
2. Cannot infer the shape of a function from only 2 points.
What type of design would you use for 1IV with multilevels?
Anova
- BS-> one-way BS ANOVA
- WS-> one-way WS ANOVA
what are the advantages to 1IV multilevel design?
- range of IV is less critical
2. can infer shape of functions
What are the disadvantages of 1IV multilevel design
- more participants (or time) needed
- counterbalancing is more difficult
- statistics more complicated
What is the design matrix for factorial designs?
- indicates the # of IV and the number of levels for each (3x4)
What factorial design for different variables?
- 2 IVs-> use a 2 way ANOVA
2. 3 IVs-> use a three-way ANOVA
What are the three types of ANOVA
- All factors (IVs) are BS
- All Factors are WS (or matched)
- At least one BS & one WS factor
How do you interpret factorial designs?
- Main Effects- difference between 2 or more levels of ONE of your IVs
- 1 it is possible to have a main effect for each IV in your design.
How do interpret interactions in factorial designs?
- occur between two or more IVs in your design
What are the advantages of factorial designs?
- more efficient that conducting separate experiments for each IV
- improves experimental control
- can investigate interactions between IVs
What are disadvantages of factorial designs?
- statistics become more complicated
2. results can be difficult to interpret
What are the two types of stats?
- descriptive and inferential stats.
What are descriptive stats?
- describes a single distribution of scores
2. measures of association
What are inferential stats?
used to compare two of more distributions.
What is parametric inferential stats?
- e.g., t-test, ANOVA
- used only for interval and ratio data
- assume frequency distributions for the population are normal
- less conservative/more powerful.
What are non parametric inferential stats
- e.g., chi-square, wilcoxen
- many can be used with ordinal or even nominal scales
- do not assume normal distribution
- moer conservative/less powerful.
What is the p-value?
- probability that observed difference among distributions occurred by chance.
- provided by all inferential statistics.
What is a type 1 error?
- finding a significant difference when the results are actually due to chance
What is a type 2 error?
- falling to find a significant difference when there really is one.
What is an independent t-test?
- compares mean and variance of two independent distributions
- significance depends on degree of overlap between the distributions.
when is the t value more likely to be significant?
- the difference between means is large.
2. the variances of each distribution are small.
what is a paired t-test?
- calculate the difference score for each participant for two condition.
- compute mean and variance for distribution of difference scores and compare to zero.
What is a one way BS ANOVA (3 levels)
compares two types of variability:
- 1 between groups
- 2 within groups
What is a one way WS ANOVA
- compares variability between conditions
2. variability within conditions.