MANCOVA

• Description of the data : 465 female, 20- to 59-year old, English-speaking residents of the San Fernando Valley, a suburb of Los Angeles, surveyed in February 1975.

- Caseseq: case sequence number; an ID variable

- Workstat: work status; a categorical variable assessing current employment status and, if

not employed, attitude toward unemployed status (1-working women, 2-role satisfied housewives, 3-role dissatisfied housewives)

- Marital: current marital status; a categorical variable assessing current marital status (1-

not married, 2-married)

- Children: presence of children; a categorical variable assessing whether or not one has

children (0-no, 1-yes)

- Religion: religious affiliation; a categorical variable assessing religious affiliation (1-

other, 2-catholic, 3-protestant, 4-jewish)

- Race: race; a categorical variable assessing ethnic affiliation (1-caucasian, 2-other)

- Control: locus of control; measure of control ideology (large numbers=external)

- Attmar: attitudes toward current marital status; satisfaction with current marital status

(large numbers=dissatisfaction)

- Attrole: attitudes toward role of women; measure of conservative or liberal attitudes

toward role of women (large numbers=conservative attitudes)

- Sel: socioeconomic level; measure of deference accorded employment categories (large

numbers=lower income)

- Attho u se: attitudes towards housework; frequency of experiencing favorable and

unfavorable attitudes toward homemaking (large numbers=unfavorable attitudes)

- Age: age group; chronological age in 5-year categories (1=16-20, 2=21-25, 3=26-30,

4=31-35, 5=36-40, 6=41-45, 7=46-50, 8=51-55, 9-56- 60)

- Educ: years of schooling; number of years completed

• Assignment

1. Develop a model for MANOVA, choosing multiple DVs (at least 2) and one categorical IV. Examine your correlation matrix to help with your decisions. Don’t use too many DVs (have at least an n=20 in each cell)
2. Use the same model you used for MANOVA, except this time include a covariate (continuous variable) that you think may have an effect on the analysis. Again, the correlation matrix should help guide you.
3. Choose one of the dichotomous variables. Using a split file, perform the same MANCOVA detailed above for each of the two groups.
   • Note that for any of these, you are free to transform the data to create a dichotomous/categorical variable if you are interested in different groups not explicitly stated by the variables (ex: high vs low socioeconomic level as determined by some cut-off point like above/below the mean). If you have an IV with more than two levels and are coming up with cell sizes <20, consider collapsing it into a dichotomous variable (collapsing work stat into working women vs housewives for example if you are not interested in role satisfaction).

• Things to remember:
• The first thing you should do is present your model. Don’t look for the "right" model, look at your variables and how they are related; think about how they are related theoretically.
• Make sure to screen your data and test assumptions for each model (you will be testing your assumptions three separate times-for the last analysis just split the file by your IV and by your grouping variable to see if you have a sufficient number of cases in each cell, don’t worry about the other assumptions).

• Missing data, sample size in each cell, linearity (of DVs, of CV with DVs, of IV with DVs), multivariate normality (n in each cell), univariate outliers, multivariate outliers, homogeneity of variance-covariance (Box’s M, and variance ratios), if applicable homogeneity of regression (build the model to test IV-CV interaction along with other main/interaction effects), multicollinearity (correlations: of DVs, of CVs)
• Present results as if they were in the results section of a journal article (see your book for examples how to write-up results in a standard fashion). Be specific about everything you do and attach your output. You are free to explain things on your output but include them in your write-up as well.
• Make sure at the end of the analysis to interpret what your results mean. Where are there significant group differences? Are results different between groups? Was your model correct?
• It’s ok if your model isn’t significant, it doesn’t necessarily mean you’ve done something wrong. However if you have limitations in your analysis note them as possible explanations (not an equal representation of groups, possible violations of assumptions, perhaps your transformed grouping wasn’t meaningful, etc.)

Price: $29.34

Solution: The downloadable solution consists of 21 pages, 834 words and 29 charts.
Deliverable: Word Document