Hi, I found that Univariate does not show the relationship between two variable but shows only the characteristics of a single variable at a time.
In contrast, ANOVA can tell if an independent variable (e.g. categorical) has significant influence on Dependent variable (ordinal).
Its Correct?
Share your views please.
Hello Anwar,
When referring to "univariate" statistical methods, most folks are describing the number of dependent (outcome) variables involved in a data analysis: one. A multivariate statistical method implies two or more dependent variables.
One-way anova has a single independent variable (IV which is categorical/nominal, as you indicate) having two or more levels, and a single, metric (DV, interval or ratio strength scale) dependent variable. One-way manova has a single IV and two or more metric DVs.
Your question is a little vague, so please pardon the explanations above if you already understand them. If you're referring to the fact that the software package SPSS has several ANOVA subprograms, one being "unianova (analyze/general linear model/univariate)" and another being "oneway (analyze/compare means/one-way anova)," this is true. However, given the same single IV and single DV, both subprograms would give the same result of the omnibus hypothesis test: Ho: mu(1) = mu(2) = ... = mu(k) (for k levels of the IV). If you have more than one IV, then only the unianova should be used, since the error terms would be different between the two programs (because oneway would push all variation attributable to IVs other than the one under investigation into the unexplained or "error" pile).
Good luck with your work!
Dear Anwar,
Univariate means the simplest form of presenting a data. Since it's "uni," it means one. To find associations, we conceptualize as "bivariate," that is the analysis involves two variables (dependent and independent variables). ANOVA is a test which is used to find the associations between a continuous dependent variable with more that two categories of an independent variable. But remember, to execute ANOVA, you need to perform a post-hoc test as well, to find where the significance between groups exist. This will give a logical interpretation of your results.
Hello Anwar,
When referring to "univariate" statistical methods, most folks are describing the number of dependent (outcome) variables involved in a data analysis: one. A multivariate statistical method implies two or more dependent variables.
One-way anova has a single independent variable (IV which is categorical/nominal, as you indicate) having two or more levels, and a single, metric (DV, interval or ratio strength scale) dependent variable. One-way manova has a single IV and two or more metric DVs.
Your question is a little vague, so please pardon the explanations above if you already understand them. If you're referring to the fact that the software package SPSS has several ANOVA subprograms, one being "unianova (analyze/general linear model/univariate)" and another being "oneway (analyze/compare means/one-way anova)," this is true. However, given the same single IV and single DV, both subprograms would give the same result of the omnibus hypothesis test: Ho: mu(1) = mu(2) = ... = mu(k) (for k levels of the IV). If you have more than one IV, then only the unianova should be used, since the error terms would be different between the two programs (because oneway would push all variation attributable to IVs other than the one under investigation into the unexplained or "error" pile).
Good luck with your work!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS