Which statistical tests are appropriate for:
- experimental designs
- correlational designs
- t-tests and ANOVA
- correlation or regression
Inferential stats address the issue of how….
accurately a sample represents a population
What are the 4 main assumptions of parametric tests?
Independence of observations.
Interval-or ratio-level data.
Normal distribution of data.
Homogeneity of variance
What are the 3 ways to handle missing data?
- Case-wise: deleting the problematic case (respondent)
- List-wise: deleting the problematic variable
- Imputation: statistical calculation of what the response may be
Hypothesis testing is performed to….
differentiate between real systematic patterns and random chance occurrences
What 3 factors affect the probability of making a type 2 error?
- The sample size (larger N reduces %)
- The level of significance used for the test (larger alpha reduces %)
- The actual value of the population parameter (further the true value is from the null values, the smaller the %)
*can control 1 and 2, can’t control 3
What two things increase the power of a test?
increase in sample size
increase in difference between the true pop value and the null hypothesis value
The standard error provides a measure of….
how much difference is
reasonable to expect between a statistic and a parameter
Type 1 errors are ….
Type 2 errors are…
- false positives (finding a sig result when there isn’t one. % = alpha)
- false neg (NOT finding a sig results when there is one. % is hard to know)
What 3 things influence the outcome of a test
- size difference between sample mean and original population mean
(large discrep increase chance of sig res) - variability of the scores (more var = larger std error)
- number of scores in the sample
(larger n = smaller std error)
What are the 3 measures of effect size?
- Cohen’s d
- Percentage of variance
- Confidence intervals
There are key requirements for multiple regression that need to be checker before running an analysis.
There are on the lesson 9 tab
In multiple regression models, smaller residuals are associate with…
higher R^2 values and low Sy.x values (standard error estimate/standard deviation of y, given x)
APA reporting/interpretation of multiple reg is on
slide 28 of week 9
assumptions for mult reg on slide…
47 of week 9
One of the assumptions in mult. reg. is that there must not be large degrees of….
multicollinearity
Durbin-Watson test should be between 1.5-2.5
*results closer to 0 or 4 indicates greater degrees of pos and neg autocorrelation than is good
VIF scores should be below…. or tolerance above…
4
.25
Correcting for violated assumptions on slide…
67 of week 6
