sigma plot "bad arg for fdist"

[oliver l.]

hello,

i try to do regression graphs with pairs of xy
then i always receive a error message :

Bad Arg for FDist F= .. df1= ... df2= ...

both axes are log (common) and my regression function y=x^b
to receive a straight line ....

what can i do

thanks for any help

oliver

[Rich Ulrich]

On 2 May 2002 01:52:32 -0700, olech...@hotmail.com (oliver l.)
wrote:

> hello,
>
> i try to do regression graphs with pairs of xy
> then i always receive a error message :
>
> Bad Arg for FDist F= .. df1= ... df2= ...
>

[ ... ]
If you *always* receive the error message,
it sounds like your SPSS is broken; reload.
Oops. Sigma Plot. (A product SPSS sells.)
What version is it? on what sort of machine?

A google search <"Sigmaplot" error regression -bars >
does not find anything relevant in the first 50 hits.
People don't seem to have many program errors.

"Always"? - *you* might always
make the same unique error of some sort.
An error might be apparent in the
command language that you don't show;
or a clue might lie in the diagnostic values
that you don't quote (one of them probably 0 or negative).

Hope this helps.

--
Rich Ulrich, wpi...@pitt.edu
http://www.pitt.edu/~wpilib/index.html

[oliver]

hello,
thank you for your help
first of all i don't get this error message
all he time .. i have to work on regression
graphs for a test (x: velocity y: erosion rate)

i have y short exaple of numbers where it is not
working:

10,0000 6,0000e-5
11,0000 1,5900e-4
13,0000 5,1100e-4
14,0000 7,5500e-4
15,0000 2,6580e-3
16,0000 4,0180e-3
17,0000 8,1950e-3

i use the 2 parameter power regression formula y=x^b
to receive a straight liine on a double log scale ...
it works in a lot of cases - but for example with
the example above, i receive this error message ...

my system: pc / win xp + sigma plot 2001

thanks for further assistance :)

greetings from munich/ germany

oliver

"Rich Ulrich" <wpi...@pitt.edu> schrieb im Newsbeitrag
news:evm2duk7evm2rhjj4...@4ax.com...

[Rich Ulrich]

On Thu, 2 May 2002 21:48:47 +0200, "oliver" <olech...@hotmail.com>
wrote:

> hello,
> thank you for your help
> first of all i don't get this error message
> all he time .. i have to work on regression
> graphs for a test (x: velocity y: erosion rate)
>
> i have y short exaple of numbers where it is not
> working:
>
>
> 10,0000 6,0000e-5
> 11,0000 1,5900e-4
> 13,0000 5,1100e-4
> 14,0000 7,5500e-4
> 15,0000 2,6580e-3
> 16,0000 4,0180e-3
> 17,0000 8,1950e-3
>
> i use the 2 parameter power regression formula y=x^b
> to receive a straight liine on a double log scale ...
> it works in a lot of cases - but for example with
> the example above, i receive this error message ...

[ ... ]

You are saying that you quoted the error message before,
when you gave it as showing dots instead of the F
and numerator and denominator degrees of freedom?

- My guess would have to be that it was indicating:
There are ZERO cases available for analysis.

I see two ways that could happen, two things that
might cause SigmaPlot to reject the data. To get around them,
you can put periods in place of the European-style commas;
and you can move the decimal places in order to
get rid of the "E" (for exponential) notation, which a lot
of programs don't like.

If this doesn't do it, you should ask a friend to watch
you up close. Google does not show that the program
has had anything like this as a noted flaw, so you are
probably making some other easy, foolish mistake.