For one-way ANOVA, there is 1 metric data as the dependent variable and 1 categorical independent variable. For 2-way ANOVA, there is 1 metric continuous dependent variable and 2 multicategory independent variables and so on. Suppose I have 1 metric dependent variable and many socio-demographic multi-category independent variables, then what will be the name of the ANOVA? we cannot say 15-way ANOVA or 20-way ANOVA for that.
Why not? However, having more predictor variables in the model allows more and higher-order interaction terms, and in models with a larger number of predictors not all interactions may be relevant or needed. So unless you don't specify a model with either all possible or no interaction terms, it will be required that you describe your model in sufficient detail anyway.
@Jochen Wilhelm can I name the table as Univariate analysis of variance of Quality of life along sociodemographic variables?
Jochen Wilhelm, I rarely disagree with you! But in the statistical vernacular I speak:
- Univariate means one dependent variable (DV)
- Multivariate means 2 or more DVs
- Univariable means one explanatory variable
- Multivariable means 2 or more explanatory variables
Here is one resource that takes this view.
Article Multivariate or Multivariable Regression?
Cheers,
Bruce
Bruce Weaver , yes, you are right, thank you!
I deleted my answer to not add further confusion.
Pathak Abhijit , given Bruce's correction on my view I'd guess your proposal is ok. What do you think, Bruce?
I have a question
Which type of within subjected correlation is used in repeated ANOVA? independent, exchangeable, autoregressive, or like random intercept of mixed models?
Hi Jochen Wilhelm. Yes, I think it could be called a univariate ANOVA model, or simply an ANOVA model. But I think there are other (potentially) more important questions. E.g.,
1) I believe quality of life (QOL) the DV. If so, how was it measured? (I'm wondering if it could/should be considered ordinal, for example.)
2) What is the sample size?
3) How many explanatory variables are we talking about? (I think all explanatory variables are "factor" variables, right?)
4) Will the model include main effects only, or also some interactions? (Presumably it cannot be the full factorial model, given that there are "many" explanatory variables.)
5) Where does the analysis fall on a continuum from completely exploratory at one end to completely confirmatory at the other end?
6) How do you intend to deal with the obvious multiplicity problem?
Cheers,
Bruce
@Bruce Weaver, sir, QOL was measured through QOL BREF scale divided into 4 scales i.e. Physical Health, Psychological Well-being, Social Relations and Environment at 5 point Likert Scale. The aggregae metric score of each respondent taken and transformed as per QOL-100 items original scale through instructions. 400 sample size. 10 or more sociodemographic variables like parents of children with different disabilities group, Caste, residence type, type of family, earning member etc. Question no 4,5,6 I don't understand. Thank you
if you have one DV with three IVs with each two categories u can write like Three way ANOVA or 2*2*2 factorial design...
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress