Module 4 Practice Questions Flashcards
What is the primary reason for using an ANOVA over a t-test?
To test mean differences that exist for an independent variable with more than 2 levels
What is the difference between a factor and levels of a factor? Give an example of a single and two factor experiment and what test should be used for these experiments.
A factor is also known as an independent variable and the levels of a factor are the conditions that make up that independent variable.
An experiment that has one factor (one IV) may assess the impact that the IV has on some outcome variable. For example, looking at whether a particular treatment has an impact on memory performance. Treatment conditions could be drug A, drug B, and no drug.
A one way ANOVA would be used for this experiment since it consists of one IV with 3 levels.
An experiment that has two factors (two IVs) may assess the impact that these two IVs in combination have on some outcome variable. For example, looking at whether fertilizer type and planting density have an impact on crop yield. A factorial ANOVA would be used for this experiment because it consists of more than one IV.
What does a one-way ANOVA test? Although it tests _____ its calculations are based on what? This metric that is used for ANOVAs can be divided into two types…
A one-way ANOVA tests for mean differences when there is one independent variable involved with more than two levels.
Although it tests for mean differences its calculations are based on variance.
This metric of variance can be divided into two types: between variance and within variance
If you draw a score from each of two conditions in an ANOVA what are the differences in the potential sources of variance compared to two scores from the same condition in a between-subjects one way ANOVA?
When drawing a score from each of two conditions in an ANOVA the potential sources of variance are: treatment effect, individual differences, and experimental error
When drawing two scores from the same condition in an ANOVA the potential sources of variance are: individual differences and experimental error
How do the sources of variance in a one-way between subjects ANOVA relate to the F statistic? Essentially, what do the denominator and numerator measure in the F-test?
The numerator of the F-test for a one-way between subjects ANOVA measures between treatment variance (treatment effect + individual differences + experimental error)
The denominator of the F-test for a one-way between subjects ANOVA measures within treatment variance (individual differences + experimental error)
What do we use the numerator and denominator df of the F value for in ANOVA?
For a given df numerator and df denominator we can know the distribution of F values that would be obtained from sample data if the null hypothesis were true
What is an omnibus test?
An omnibus test is one that tests a general research question
How does an omnibus test present a challenge for one of the MAGIC criteria? Which one and why?
Presents a challenge for “Articulation”
This is because as results get more complex there will be more ways in which they can be articulated. A general conclusion will not suffice
What are the two general approaches to resolving the “Articulation” challenge? Define them and give examples.
Using follow up tests:
A posteriori (post hoc)
- Not based on prior planning nor driven by theoretical hypotheses
- Tukey’s HSD (tests all pairwise comparisons with a strong control for family wise error)
A priori
- Based on prior planning and driven by theoretical hypotheses
- Allows us to test more specific comparisons
- Planned contrast
What are some common post hoc tests?
LSD, Bonferroni Adjustment, Tukey HSD
What is the most common a priori test?
Planned contrast
What is the rule of contrast weights?
They must sum to zero
How do you control family wise errors when doing follow up tests? What is the consequence of not controlling for family wise error? What is the consequence of controlling too stringently for family wise error?
We control for family wise error by adjusting alpha. We can do this while conducting follow up tests such as Tukey HSD and Bonferroni adjustment.
Not controlling for family wise error means a greater likelihood of making a type I error
However, the more we hold down family wise error, the more power also decreases and the likelihood of making a type II error increases.
What are the two standardized effect sizes used in one-way ANOVAs
Eta squared
- Reflects proportion of variance accounted for by group
Cohen’s f
- Reflects ratio of effect SS to error SS
What are the assumptions of between subjects one way ANOVA?
- Independence of observations
- Distribution of outcome variable is normally distributed for each level of the IV
- Homogeneity (equality) of variance in the outcome variable across levels of the IV
What are the assumptions of within subjects one way ANOVA?
- Independence of SETS of observations
- Distribution of outcome variable is normally distributed for each level of the IV
- Homogeneity (equality) of COVARIANCE in which the relative standing of each participant is the same in each condition
Similarities and differences for one-way between and within ANOVAs
Similarities:
- Both partition total variability into two components: between treatment variability and within treatment variability
Differences:
- Within ANOVAs have less sources of error given that individual differences are a constant
Both within and between ANOVAs share the same advantages and disadvantages of independent and repeated measures t-tests…
Repeated measures have more power
- individual differences are more controlled
Repeated measures are more economical
- need less participants to achieve appropriate power
Repeated measures are susceptible to carry over effects whereas independent tests are not
Repeated measures are susceptible to demand characteristics whereas independent tests are not
