ch 8 review Flashcards
differences between means of the groups that are not the result of treatment
Standard error of the Difference between the 2 means
Pooled variance
combines the variability that occurs within each group and is the first indicator of the amount of error present in analysis
First indicator of the amount of error present in analysis
Pooled variance
IF scores within each group are widely different from one another, these large differences contribute to a larger _______________________________________________ and to the over all ______
Standard error of the Difference between the 2 means, error
The larger the error, the _________ the calculated t-value and the less likely that a t-value will be _____________.
smaller, significant
to calculate t we divide what by what
The difference between the mean scores divided by the error variance( a.k.a. standard error of the different between two means- why? because mathematicians are assholes)
When calculating the degrees of freedom to use for comparing a t value to the t-table we need the df. If a control group and treatment group each has 10 participants what are the degrees of freedom used?
What else do we need to know?
(n1-1) + (n2-1)= df
(10-1)+(10-1)=18
df= 18
We need to know what alpha was set at. Most likely .05 but if history repeats itself you should beware of random bullshit like .025.
IF the T-value obtained is larger than the t-table critical value what can we conclude?
If out value is larger than the tcv then it is unlikely our results occurred by chance alone and we say the results are significant. Yippee.
If results are significant how should we refer to p?
p < .05.
What are the assumptions that all parametric tests are based on:
- the data to measure the DV are interval or ratio
- the underlying populations for each group are normally distributed (30 participants usually solves this)
- The variances for the populations are roughly equivalent (homogeneity of variance)
If the variance of one population is twice as large as the other groups can you still assume homogeneity of variance?
Yes. You do not have to be concerned until one variance is 4 times larger than the other.
What is a large enough sample size to reasonably assume we have met the requirements for normal distribution?
30
If data is NOT interval or ratio what type of test should we use? give an example
non parametric statistical tests- Chi-square test
If our sample sizes are unequal can we be sure of the robustness of our statistical test?
Yes, as long as variances are relatively similar we can be assured.
Why were inferential techniques developed?
so we could use smaller groups rather than an entire population to test the effects of an IV.