chapter 9 one way anova Flashcards
We can use an ________ when we have more than two groups of participants or when participants experience more than_____ treatment ________.
ANOVA, 2 treatment conditions
A between groups ANOVA is an _______ test of probability that we can use with _______ independent variable having more than _____ levels.
A between groups ANOVA is an INFERENTIAL test of probability that we can use with ONE independent variable having more than TWO levels.
A __________ ________ ANOVA helps us determine if groups differ as a function of the _____ ____ ____. We conduct additional ______ tests to determine _______ the difference________.
A BETWEEN GROUPS ANOVA helps us determine if groups differ as a function of the LEVEL OF IV. We conduct additional STATISTICAL tests to determine WHERE the difference EXISTS.
We conduct Post hoc Analysis to ________specific ______________ when we have more than ____ groups..
We conduct POST HOC ANALYSIS to ISOLATE specific DIFFERENCES when we have more than TWO groups..
Which analysis of variance (ANOVA) is more sensitive to to detecting a statistically significant outcome? Why?
Repeated Measures ANOVA- because there is less error.
When we find a statistically significant outcome what other tests/measures should we perform?
Post Hoc and measure effect size.
Statistical significance of variance is determined by calculating a ______ of variance ______ treatments (treatment variance) to variance ________ treatments (error variance).
Statistical significance of variance is determined by calculating a RATIO of variance BETWEEN treatments (treatment variance) to variance WITHIN treatments (error variance).
When results of a statistical test are significant we conclude…
it is unlikely that the difference between the treatments occurred by chance.
Why not just use the t-test for comparisons of more than two groups? What kind of test reduces this problem.
Repeated use of the t-test can result in an increase in TYPE 1 error.
What is a type 1 error?
a statistically significant difference occurred by chance, rather than due to treatment of DV.
When we set alpha at .05 what are we saying?
We will not tolerate more than a 5% chance that variance is due to chance.
What type of ANOVA has one independent variable with multiple levels or conditions?
A One Way ANOVA
ANOVA is a family of statistical tests that compares ______ _______ to asses whether ________ across _____ are reliable.
ANOVA is a family of statistical tests that compares GROUP MEANS to asses whether DIFFERENCES across MEANS are reliable.
________ _______ ANOVA: an ANOVA in which a single individual does not provide data for more than one _____ or ______.
BETWEEN GROUPS ANOVA: an ANOVA in which a single individual does not provide data for more than one CONDITION or GROUP.
ANOVA = what/what
ANOVA = Treatment variance/error variance
When calculating ANOVA, if the calculated ratio is large, then the treatment effect is ______ compared to ______ ________ __________. IF this is the case we can conclude that treatment LIKELY made a difference and that these is a statistically significant outcome.
When calculating ANOVA, if the calculated ratio is large, then the treatment effect is LARGE compared to MEASUREMENT ERROR. IF this is the case we can conclude that treatment LIKELY made a difference and that these is a statistically significant outcome.
The F statistic is named in honour of ________ who advanced the use of ANOVA by suggesting probability is at the heart of good inferential statistics
Fisher
The most important assumption that needs to be met before conduction an ANOVA.
The DV must be measured on either an interval or ratio scale
The assumption that variability across the treatment conditions will be similar is_______________ and must be met before an ANOVA can be conducted.
Homogeneity of Variance: see page 206 for ways to state this using statistical notation because this shit program will not allow changes of font.
If we want the standard deviation for each condition to be similar we are referring to..
Homogeneity of Variance
One way of determining if homogeneity of variances (HOV) exist is by looking at the ______ of the graphs of each condition. If they are all different you have ________ of variance. What else can this graph tell us?
Kurtosis, heterogeneity ( think heterosexual people like opposite or different sex- we WANT HOMOS)
It can also tell us to what degree we have normal distribution of error .
The Hartley F-Max test is one way of testing
Homogeneity of Variance
The Levene’s Test of Equality of variances is one way of testing
Homogeneity of Variance
The F-statistic is the ratio of ______ to ______variance used to detect _____ _______.
The F-statistic is the ratio of TREATMENT to ERROR variance used to detect STATISTICAL SIGNIFICANCE.
If the variances are not equal among the groups at what point would the assumption be violated?
If one variance is four times larger than another then we have not met this assumption required to complete an ANOVA.
If assumptions are not met for an ANOVA what statistical tests would you consider
Non parametric tests
Two assumptions that must absolutely be met before conducting a ANOVA
- Dv must be interval or ration
- Homogenity of variances must be assumed (cannot have one more than 4 times larger than another)
Independence of observations refers to:
ensuring that observations or data in each treatment condition are independent. That is to say, no person is measured in more than one condition.
When conducting and ANOVA we are interested in calculating three measures of variability. They are…
- total variance
- error variance
- variance between groups
First step in calculating an ANOVA is to calculate the…
Total variance or SStotal
pg 208
Second step in calculating an ANOVA is to calculate the..
* hint S squared
approximation of an average deviation from the mean. (Divide SStotal by dftotal)
df total = all of the scores from all conditions minus 1.
pg 208
Variability between groups reflects
the effect of the treatments
Variability within groups reflects
individual differences among individuals in a single treatment, or error variance
variability between individuals participating in the same condition. Also used as the __________denominator for ANOVA
Error variance
When tabulating data for a source table, we want a ______ difference between groups and a ___________ difference within groups.
Large, small
If you use three levels of an IV how many comparisons are possible? What if you use four levels?
3, 6
Tukeys Post Hoc test is also referred to as the …
Honesty Significant Difference
The HSD or Tukey’s Post Hoc test requires what kind of sample sizes?
equal.
what reflects the amount of error present across all conditions?
MSerror
When calculating the Tukey HSD what three values do we need:
q, MSerror and n (only the sample size NOT the population)
What four things do we need to calculate q for the HSD?
k: number of treatments
an alpha (ex: .05)
and dferror
q table- also called the studentized range statistic table.
What is the benefit to using a Scheffe Test instead of Tukey’s
Does not require equal groups
what is the estimate of the proportion of variability explained by the independent variable- or strength of association
effect size
