Hi everbody,

I am currently writing my bachelor thesis and i have several one tailed Hypothesis which I test with MANOVAS. When testing with SPSS you only get two tailed p-values. Can i divide those p-values to get a one tailed values? Is that possible for MANOVAS or only for t-tests?

Kind regards!

You can (re)compute the significance levels for a one-tailed test for the parameter estimates just as for a t test, but if you have factors with more than two levels, what would the one-sided hypothesis mean?

Also, you need to ensure that the one-sided hypothesis is not being used just as a way to get a significant result.

Thanks for your answer. For me it's not about getting a significant result. I had the Hypothesis before testung with the data. I'd like to measure the quality of life (I know, a difficult thing to do) and therefore I have many independet variables (like sleep quality, etc.). Together all the variables should represent the QoL. So I have a Hypothesis (QoL) and lots of sub-Hypothesis (Sleep quality, life satisfication, ...)

I will offer you these quotes from some people knowledgeable about MANOVA.

I like that advice. My own observations suggest that more often than not,

people use MANOVA as a precursor to multiple ANOVAs. They seem to believe

that a significant MANOVA buys protection against Type I error in the

subsequent ANOVAs in the same way that a significant omnibus F-test buys

protection for the subsequent t-tests when one uses Fisher's LSD (with 3

groups). But as Huberty & Morris clearly say, it just ain't so! For them,

it boils down to this: If you have a multivariate question, use MANOVA; but

if you have a series of univariate questions, use a series of ANOVAs. (In

the latter case, you might consider some kind of adjustment for

multiplicity--but that's another can of worms!)

https://pdfs.semanticscholar.org/ee6c/77c99c8e4530d0cccaedf85ed525fb22a02d.pdf

Rich Ulrich wrote

> As an expert in Manova, my advice has always been,

> "Avoid it when you can."

*sorry, I meant I have many dependet variables

*sorry, I meant I have many dependet variables

You might want to consider building a QoL scale and then using that as your dependent variable rather than taking a multivariate approach. You might look at the Dimension Reduction and Scale choices under Analyze, although some of them might not be included in your license.