Back to the Top
hello
I have performed a study in which i need to check to see if the drug
administered gave a statistically significant difference between day
0 and day 7 . In order to do this i used SPSS to plot a Q-Q plot to
test the normaility of the distribution of the difference. On looking
at the SPSS graph i note that it plots observed values which i assume
it has taken from my data against expected values. Does anyone know
where these expected values of the normal distribution come from ?
I also note that differences of 0 are not plotted. Does anyone know
why this is ?
Paul
Back to the Top
Dear Paul,
The normal QQ plot graphically compares the distribution of the
difference between day 0 and day 7 (x-axis) to the normal
distribution (y-axis). If correctly plotted, the plot should show
both the actual data (day 0 - day 7)and the expected data (which
comes from the normal distribution) allowing you to visually
determine whether data are normally distributed. The closeness of the
data points to the 45 degree line quantifies the degree of normality
or lack thereof.
Regards,
Nav
--
Nav Coelho, M.Sc.
Biostatistician, Pharmacokinetics and Statistics
Biovail Contract Research
Tel: (416) 752-3636 Ext. 369
Email: navdeep.coelho.-a-.biovail.com
URL: www.biovail-cro.com
Back to the Top
hello Nav
Thanks for your reply . What i do not understand about these Q-Q
plots is how does SPSS calculate the expected value of the normal
distribution . Does it take my values and then predict what they
would be if they were normally distributed and then this forms the
basis of the 45 degree line ?
Back to the Top
The following message was posted to: PharmPK
Hi there.
Kolmogorov-Smirnov (KS) and Lilliefors tests of normality are based
on calculation of theoretical
cumulative frequencies at each experimental point one provides, in
this particular case the
differences between two measurements. For such calculation
experimental average and standard
deviation are calculated. These values are used for calculation of
cumulative normal distribution at
each point. Maximal difference between experimental cumulative
frequency and theoretical cumulative
frequency represents KS or Lilliefors statistic. Adherence of
experimental data to any other
theoretical distribution could be analyzed this way.
I think that calculated values used in Q-Q plot are theoretical
cumulative frequencies. Still, I do
not use this package and therefore I can not be certain.
Hope this helps.
Zeljko Debeljak, PhD student
Osijek Clinical Hospital
CROATIA
Back to the Top
Hi Paul,
You are correct as QQ plot plots the actual data along the y-axis,
and the values that the data would have if they were exactly the
percentiles of a normal curve. Therefore, if the data is
approximately like that of a bell curve, the line should reseemble
the 45 degree line.
Hope this was helpful!
Nav
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)