PG851: GLM--ANOVA

This blog is for the use of students enrolled in either section of PG851. It is intended to be a forum for asking, discussing, clarifying, and helping.

Monday, November 28, 2011

here you guys I hope you find this helpful


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.

1 Comments:

Post a Comment

<< Home