Wednesday, November 30, 2011
"on the margins"
Independence of errors
Tuesday, November 29, 2011
front desk- attendance sheet
Section 8 of the Review Sheet
She said something very similar to what TJ said in the review. Both random effects and repeated measures variables influence the whole model in that they change the way in which effects are tested. So if there is one random effect IV, it changes the way all IVs are evaluated (against what error term). Same goes for repeated measure.
You should be able to recognize whether there is either a repeated measure or a random effect from the data as opposed to analyses that don't include either.
Homework Question: Changing up Randomized and Fixed Effect Variables
Reporting Non-signficance
Monday, November 28, 2011
Is it data, split file, organize, by group, and the two levels of IV.
Factorial ANOVA, Advantageous and Disadvantageous
Advantageous:
1. It tests for two or more IVs thus preventing an inflated Alpha that would occur if you run separate oneway ANOVAs
2. Allows for separate evaluations of each predictor variable.
3. Allows for examinations for Joint, or interactive effects of the variables.
Disadvantages:
1. Crossing multiple variables with many levels dramatically increases the resources required.
Repeated Measures
Advantageous:
1. Fewer participants are needed
2. Greater Familiarity with research procedure (Less time recruiting and explaining)
3. Each participant serves as own control
4. More power, since variance due to participant can thus be partialed out of error term.
Disadvantageous:
1. More is required of each participant
2. Sphericity
Mixed Designs
Advantageous:
1. Both combines randomized and repeated components.
2. Mixed Designs are preferable if carryover on one variable is a concern,
but change over time is of interest.
Disadvantageous:
1. The assumptions of Normality, homogeneity of variance, independence of observation and Sphericity are all in play.
ANCOVA
Advantageous:
1. Removes variance due to extraneous variable from error term
2. DV scores are reset to “what they would have been had all participants been equal on the covariate.”
Disadvantageous:
1. A causal relationship cannot be inferred especially if it is nonexperimental.
2. Very sensitive to outliers
3. Both covariate and DV have to meet normality
4. In nonexperimental designs, sometimes the covariate is substantively related to the IV.
5. Covariate only help if it is related to the DV, if not it just burned a degree of freedom for nothing.
6. A lot of assumptions, Multicollineratiy, linearity, homogeneity of regression, reliability of covariate, as well as independence of observation, normality, homogeneity of variance
Latin Square Designs
Advantageous:
1. Allow for incomplete crossover designs while maintaining the benefits of counterbalancing.
Disadvantageous:
1. Only a subset of possible orders is used.
2. You have to do it by hand.
3. It is hard to do and requires a lot of planning, ensuring that the orders and sequences are correct.
Sunday, November 27, 2011
Missed class for Latin Squares!
Random effects DF
Reliability of the covariate
Descriptive and inferential stats
Saturday, November 26, 2011
Final Project
Article Review 5
Friday, November 25, 2011
Wednesday, November 23, 2011
Final Project
For the final project, are we supposed to use indepenent variables which yield significant results? I ran an analysis which did not produce any significant results so I am wondering if I should choose other constructs.
Tuesday, November 22, 2011
review session
Thursday, November 17, 2011
Final Project
Wednesday, November 16, 2011
Terminology
Tuesday, November 15, 2011
df for Random Effects ANOVA
Monday, November 14, 2011
Article Review: Relevance to Current Question
Friday, November 11, 2011
Tables in ANCOVA homework
Independence of Observations
Full ANCOVA Instructions
Thursday, November 10, 2011
What does this mean? Please elaborate
To what value of the covariate were scores adjusted? (B) What does it mean to adjust scores in this way? (C) Was this a reasonable value of the covariate for SPSS to use? Why or why not?
Question
ANOVA writeup
Upcoming Review Sessions
Wednesday, November 09, 2011
HW grade change
Tuesday, November 08, 2011
Covariant & Indep
submitting ANCOVA homework
homogeneity of variance for CV
ANCOVA vrs Repeated Measures
Exam 2
Friday, November 04, 2011
Repeated HW
Wednesday, November 02, 2011
Trend-Analysis
Extra Credit
Tuesday, November 01, 2011
Marginals
I know that Bonferroni, Scheffe, Tukey are all under homogenity of variance. Dunnette under heterogenity of variance.
Scheffe, Tukey and Dunnette, Equal cell sizes while Bonferroni has unequal cell sizes.
sphericity vrs. compound symmetry
Invitation to an industrious student
2 Questions: 1. Compound Symmetry 2. Post Hocs
2. In class, for post hocs for factorial design, we split the file and then looked at a post hoc run with a 1-way ANOVA, and for post hocs for mixed design, we split file, ran specifically chosen t tests, and then adjusted alpha. Why different post hocs approaches for the different types of designs?
