Hi,
I don't understand why I am not obtaining the results of my ANCOVA... Dots are displayed.
Dependent variable: number of products selected (continuous)
IV: number of healthy products selected / number of unhealthy products selected (continuous)
Covariates: Nutrition Knowledge / Motivation (continuous)
Thank you for your help!
Hard to say exactly - but it looks like there is no variation in some of your variables.
I'd start by looking at your descriptive statistics for each variable (M and SD) and then look at correlations to see if any of them are perfectly collinear.
Your error sum of squares is zero. The F-statistic uses this as the denominator resulting in division by zero. This implies that some of the independent variables sum exactly the dependent variable. Is number of products selected = Total Healthy + Total Unhealthy (+ some constant)?
Best regards,
Timo Törmäkangas
Gerontology Research Center
University of Jyväskylä
Finland
Timo Törmäkangas Hi! Thank you for your help. Yes it's.
I tried to do another analysis (one-way repeated measures ANOVA) with covariates but I'm unsure how should I analyze the results. Can u help on this?
As Timo mentioned, the problem was that the model was perfectly specified by number of healthy products selected / number of unhealthy products selected since they are perfectly dependent on each other with regard to the dependent variable, which is simply the sum of both. So, what should this predictors add as information to your model?!?!
For your second model: this model seems more plausible, BUT with the information given about the dependent variable, it sounds as if you have a count variable and not a continous variable, since values below zero and fractions are not possible (only integers). In this case, the ordinary least squares regression is not the correct statistical model (in most cases), since the residuals will most often not follow a normal distribution (check this point) and the distribution of the means is possibly not linear. Typically, suitable models are from the generalized linear models family, e.g. poisson or negative binomial regression with other link functions as the identity link!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software