Can someone explain to me what an Multivariate analysis of variance (MANOVA) does in terms of statistics and studies?

2.75K views

Can someone explain to me what an Multivariate analysis of variance (MANOVA) does in terms of statistics and studies?

In: Mathematics

Anonymous 0 Comments

Let’s start with the basics. Univariate would be a single variable, such as my dog’s weight: I weighed him several times and saw that he weighed 52, 53, 51, 54, and 53 pounds again. You could do some basic statistical analysis on this list, such averages (mean, median, and mode), standard deviation, etc.

It gets more interesting if you pair the list of weights against the list of dates when he was weighed. This is a bivariate (two variable) analysis. You might look for trends: is he gaining weight? Losing weight? Is there no relationship at all between weight and time?

Things get more interesting if you combine weight and time with other data, such as the amount of food he receives every day, how often we go for a morning run, weather data (temperature, precipitation, clouds), how much I’m travelling for work (he stays with my aunt while I’m gone). This multivariate data set might reveal some interesting things: he gains one pound per week that he stays with my aunt, regardless of how much I tell her to feed him. We exercise less when the weather is bad, but this does not seem to have an effect on his weight.

Multivariate analysis is useful in engineering, when you’re trying to optimize a complicated system. First you need to figure out how all the variables are related, then you can work to optimize for cost, performance, reliability, etc.