{!!! SPAM ???} Re: GLM repeated measures concern/question

Posted by Kornbrot, Diana on Jan 01, 2012; 9:06am
URL: http://spssx-discussion.165.s1.nabble.com/GLM-repeated-measures-concern-question-tp5104022p5112991.html

I do indeed have CONTENT knowledge
The key descriptive parameters for any visual search task are the slope and intercept of the function of reaction time (DV) on set size (IV).
Typically, as in this case,researcher are comparing these parameters in different conditions. So the inferential problem is to know whether there are reliable differences in slope between tasks/conditions. For example condition 1 may have the target further out to the side than condition 2 whaere targets are more central.
In order to get the work published in a ‘respectable’ psych journal, as questioner well knows, he will need to get the slopes and intercepts and perform inferential tests. So what he needs to know is how to do this in SPSS.
My suggestions enable him to do just that, in what I have found the simplest way
(There IS a syntax option in GLM polynomial that allows one to specify the actual set size spacing
In this case: GLM RESPONSE BY STIMULUS /CONTRAST(STIMULUS) =POLYNOMIAL(1,8,32)
But suspect means plot will still be equally spaced)
__
SPSS command synatx referenc odf includse: POLYNOMIAL Polynomial contrasts. This setting is the default for within-subjects factors.
The first degree of freedom contains the linear effect across the levels of the
factor, the second degree of freedom contains the quadratic effect, and so on. In
a balanced design, polynomial contrasts are orthogonal. By default, the levels
are assumed to be equally spaced; you can specify unequal spacing by entering a
metric consisting of one integer for each level of the factor in parentheses after
the keyword POLYNOMIAL. (All metrics that are specified cannot be equal; thus,
(1, 1, . . . 1) is not valid.) An example is as follows:
GLM RESPONSE BY STIMULUS /CONTRAST(STIMULUS) =
POLYNOMIAL(1,2,4).
Suppose that factor STIMULUS has three levels. The specified contrast indicates
that the three levels of STIMULUS are actually in the proportion 1:2:4. The
default metric is always (1, 2, . . . k), where k levels are involved. Only the
relative differences between the terms of the metric matter. (1, 2, 4) is the same
metric as (2, 3, 5) or (20, 30, 50) because, in each instance, the difference between
the second and third numbers is twice the difference between the first and second

---

Your suggestions are interesting but they change the nature of the question. (NB taking logs give spacing of 0,3, 5 which STILL isn’t equally spaced.)
Sternberg, S. (1969). The discovery of processing stages: Extensions of Donders' method. Acta Psychologica, Amsterdam, 30, 276-315.
Sternberg, S. (2001). Separate modifiability, mental modules, and the use of pure and composite measures to reveal them. Acta Psychologica, 106(1-2), 147-246.
Townsend, J. T. (1972). Some results concerning the identifiability of parallel and serial processes. British Journal of Mathematical and Statistical Psychology, 25, 168-197.
Townsend, J. T. (1992). On the proper scales for reaction time. In H.-G. Geissler, S. W. Link & J. T. Townsend (Eds.), Cognition, information processing and psychophysics:  basic issues (Vol. Lawrence Erlbaum Associates). Hove and London.
Treisman, A. (1991). Search, similarity, and integration of features between and within dimensions. Journal of Experimental Psychology: Human Perception and Performance676, 17, 652-676.

Sternberg’s key Ms have more than 900citations, Treisman’s more than 500, Townsend’s more than 2000.  This is a well researched area & the key issues are KNOWN
Best
Diana





On 31/12/2011 22:14, "Rich Ulrich" <rich-ulrich@...> wrote:

Diana seems to know something about the subject at hand.  
But I have to offer an obvious suggestion - I think the word needed
was "monotonic" and not "linear", in "reaction time is a linear function
of set size...."   -- Surely the difference between sets=1 vs 2  is much
larger than the difference between sets= 31 vs. 32  -- which is what
is implied by "linear".  My intuition says, That is not reasonable.  My
experience says that going against that intuition results in non-linear
relations that introduce artifacts in the testing and complications in the
interpretations.

I would agree that "throwing out the linear assumption is ESSENTIAL" if
she were referring to the spacing of (1,8,32).  That is not reasonable.

But she says, "Test whether log or raw set size is better linear
predictor"  and I presume that "raw" refers to (1,8,32).  

As to logs:  The log-spacing of (1,8,32) can be taken from the
respective powers of 2:  (0,3,5).  That is prettly close to linear, so
it is not unreasonable to use it, if you do not have a program that
affords the luxury of using the slightly unequal log-spacing (0,3,5).  
It is also justified by the experimenters' (presumable) intention of
providing equal spacing in their design.  

(Of course, the log is not the only power transformation that might
be considered for spacing.)

---

Professor Diana Kornbrot
email:  d.e.kornbrot@...    
web:    http://web.me.com/kornbrot/KornbrotHome.html
Work
School of Psychology
 University of Hertfordshire
 College Lane, Hatfield, Hertfordshire AL10 9AB, UK
 voice:   [phone]
   fax:     [phone]
Home
 19 Elmhurst Avenue
 London N2 0LT, UK
    voice:   [phone]
    mobile: [phone]
   fax:      [phone]