hello every one
Can the independent variable be a constant
I want to Run MANOVA
I have the independent variable that is constant a budget during the year
and the opinion on the impact of the budget on quality of education (1 DV), on research (2DV), diplomatic rate (3dv), Employment rate(4DV) are the dependent variable
i Run the MANOVA but it looks like i have a lot of errors even if i tried to normalize Data,
SPSS gives me this message: this Box's Test of Equality of Covariance Matrices is not computed because there are less than two nonempty cells.
i tried to work with the discrim but i get the same errors
i get my data frow a survey it means that i can not change it , what other methods to analyse my data
I am trying to analyse the impact of quantified budgeting (historical budgeting, because it is the only method that is applicated in mycountry,the independent variable is historical budgeting>>its the budget>>> its continuous variable) on some universities performance indicators.
Thank you for the help
A constant cannot be used as an independent variable by definition because a constant cannot covary/correlate with the dependent variable (DV) and hence cannot account for any variability in the DV.
Now why are you trying to use manova? SPSS is telling you that what you are doing makes no sense. David Booth
Hello Fatima,
As both Drs. Geiser and Booth note, unless you're sampling cases over time (so that budget total can vary) or over institutions or locales (so budget totals can vary), you'll have to abandon the idea of using budget total as a variable; as you note, it's a constant for your data set.
So, you'll either have to sample both the IV (budget) and the DVs at a smaller unit level (for example, at the municipality/county/parish level instead of at a state level, or at a state level instead of at the national level), or sample over time (year 1, year 2, etc.) in order to be able for address such a research question as your query posts.
If this doesn't make sense, perhaps you should consult with a statistician or more experienced researcher at your institution.
Good luck with your work.
thank you very much for help
it means that I can not do hypothesis testing with my data just descriptive analysis
Hi Fatima Ezzahra Kabba
I cannot add more to the previous answers.
However, for your future research, consider that something that is constant doesn't have much information for research. The constant can only inform about the position of a process / phenomenon.
Your statement "it means that I can not do hypothesis testing with my data just descriptive analysis" doesn't make sense. Suppose for item1 everyone answered 7 on a 7-point scale. I can test H0: mean(item1) = 2, and reject this (assuming your sample isn't small and your critical alpha isn't minute). Further, if your a priori beliefs about the distribution of this variables are 95\% 7 and 5\% 6, this might just change this a bit (again, depending things like sample size). Overall the point raised above though, that this at least for this sample this is not a variable (since it doesn't vary) is valid. If this "variable" is important for your research it is good you found this result in your pilot study so you can make adjustments for your main study.
Hello Fatima Ezzahra Kabba. David Morse wrote:
"...unless you're sampling cases over time (so that budget total can vary) or over institutions or locales (so budget totals can vary), you'll have to abandon the idea of using budget total as a variable".
After reading your question, I wondered if you do in fact have data from multiple universities, and that the annual budget is constant for each university, but varies across universities. Is that correct? It would really help to clear things up if you uploaded a small dataset with 15 or 20 rows showing the key variables for your analysis. CSV is a nice format, as it is easily accessible to everyone. HTH.
PS- Regarding MANOVA, you may find this nice article by Francis Huang interesting and informative.
Article MANOVA: A Procedure Whose Time Has Passed?
To do hypothesis testing, you need variation. In your case, maybe some universities use historical budgeting and some don't, and in that case you can test how things differ depending on whether they use historical budgeting. We can't help you without knowing what you are predicting, and how you expect your independent variable(s) to have an impact.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS