----- Forwarded message from Antigoni Kaliontzopoulou <[email protected]> -----

Date: Sat, 9 Mar 2013 08:17:53 -0500
From: Antigoni Kaliontzopoulou <[email protected]>
Reply-To: Antigoni Kaliontzopoulou <[email protected]>
Subject: Re: Compare Vector Directions?
To: [email protected]

Dear Marlijn,

if I understand correctly what you are trying to do, there are several points here:

Why is a 30 degrees angle showing a significant P value (0.0001), but
a 150 degrees angle a P value of 1 ? I thought 150 degree angle is
actually like 30 degrees, but then with the vectors having opposite
signs?


Whether you get a significant result or not when comparing vector directions highly depends a) on the number of observations examined for each comparison and b) on the dispersion of these observations around the vectors. As such, you cannot extrapolate numerical patterns from one comparison to the other. You need to trust the p-values you obtain.

So that vectors with a 150 degree angle between them would still be
more similar to each other (but with opposite signs) than vectors that
have an angle of 90 degrees between them?


Same as above, due to sampling issues etc, your analysis seems to indicate that in one case (factor A) you cannot statistically support a difference between the examined vectors (vector of shape-CS and vector of shape-factorA), while in the other (factor B) you can.

In my opinion, the relation between Factor B and shape is very similar
to centroid size effects, as is Factor A, although in opposite
direction.
All other factors have a different relation to shape.


Hum... What I see from the results you attached is that factors A and E have a similar effect as size (vector directions are not distinguishable), while factors B, C and D do not have the same effect as size. However, from these results you cannot say if different factors have the same effect BETWEEN THEM. For that, you would need to run the same kind of analysis, but comparing factor A to B, B to C and so on.

Also, I am not sure what other analyses you have done until now, but maybe starting from a more general framework (like an ANCOVA-type analyses to examine interactions between factors and size) would be useful to get a first, more general image. Finally, do not forget that vectors also have a magnitude, other than a direction, which may (or may not) be interesting for your study.

A methodological suggestion would be to also run the same analyses using the tools included in the geomorph R library (trajectory.analisis function), although I assume you should obtain the same results.

I hope this helps
Antigoni

On Fri, Mar 8, 2013 at 2:57 AM, <[email protected]> wrote:

----- Forwarded message from Marlijn Noback <[email protected]> -----

     Date: Thu, 7 Mar 2013 05:52:18 -0500
      From: Marlijn Noback <[email protected]>
      Reply-To: Marlijn Noback <[email protected]>
      Subject: Compare Vector Directions?
      To: [email protected]

Dear all,

With enthusiasm I have tried the feature of vector direction
comparison within MorphoJ. But there is something I do not understand.
I sincerly hope you can help me with this issue. I am new to this
analysis, so I am not sure how to interpret my data.

I am trying to see if some of the environmental factors I am analyzing
have similar effects on cranial shape as do changes in cranial size.

I compared the following vectors:
Regression of shape on factor A, B, C, D and E.
Regression of shape on centroid size.

And it gave me the following results:

Angles (in degrees)

CentSize1
Fac.A  148.621
Fac.B   32.194
Fac.C   64.088
Fac.D   76.127
Fac.E  130.435

P-values (parametric)

CentSize1
Fac.A   1.00000
Fac.B   <.00001
Fac.C   0.00433
Fac.D   0.08267
Fac.E   0.99999

Why is a 30 degrees angle showing a significant P value (0.0001), but
a 150 degrees angle a P value of 1 ? I thought 150 degree angle is
actually like 30 degrees, but then with the vectors having opposite
signs?
So that vectors with a 150 degree angle between them would still be
more similar to each other (but with opposite signs) than vectors that
have an angle of 90 degrees between them?

In my opinion, the relation between Factor B and shape is very similar
to centroid size effects, as is Factor A, although in opposite
direction.
All other factors have a different relation to shape.

Is this the correct interpretation?

I would very much appreciate your help in this.

With kind regards,

Marlijn Noback

marlijn.noback ( at ) ifu.uni-tuebingen.de

----- End forwarded message -----




--


Antigoni Kaliontzopoulou

CIBIO, Centro de Investigação em Biodiversidade e Recursos Genéticos
Campus Agrário de Vairão, 4485-661 Vairão
PORTUGAL
Department of Ecology, Evolution, and Organismal Biology
Iowa State University, Ames,
Iowa 50011, USA

tel: +351 91 3086188
mail to: [email protected]
            [email protected]



----- End forwarded message -----



Reply via email to