NAG FL Interface
g01daf (normal_scores_exact)
1
Purpose
g01daf computes a set of Normal scores, i.e., the expected values of an ordered set of independent observations from a Normal distribution with mean $0.0$ and standard deviation $1.0$.
2
Specification
Fortran Interface
Integer, Intent (In) 
:: 
n, iw 
Integer, Intent (Inout) 
:: 
ifail 
Real (Kind=nag_wp), Intent (In) 
:: 
etol 
Real (Kind=nag_wp), Intent (Out) 
:: 
pp(n), errest, work(iw) 

C Header Interface
#include <nag.h>
void 
g01daf_ (const Integer *n, double pp[], const double *etol, double *errest, double work[], const Integer *iw, Integer *ifail) 

C++ Header Interface
#include <nag.h> extern "C" {
void 
g01daf_ (const Integer &n, double pp[], const double &etol, double &errest, double work[], const Integer &iw, Integer &ifail) 
}

The routine may be called by the names g01daf or nagf_stat_normal_scores_exact.
3
Description
If a sample of
$n$ observations from any distribution (which may be denoted by
${x}_{1},{x}_{2},\dots ,{x}_{n}$), is sorted into ascending order, the
$r$th smallest value in the sample is often referred to as the
$r$th ‘
order statistic’, sometimes denoted by
${x}_{\left(r\right)}$ (see
Kendall and Stuart (1969)).
The order statistics therefore have the property
(If
$n=2r+1$,
${x}_{r+1}$ is the sample median.)
For samples originating from a known distribution, the distribution of each order statistic in a sample of given size may be determined. In particular, the expected values of the order statistics may be found by integration. If the sample arises from a Normal distribution, the expected values of the order statistics are referred to as the ‘Normal scores’. The Normal scores provide a set of reference values against which the order statistics of an actual data sample of the same size may be compared, to provide an indication of Normality for the sample.
Normal scores have other applications; for instance, they are sometimes used as alternatives to ranks in nonparametric testing procedures.
g01daf computes the
$r$th Normal score for a given sample size
$n$ as
where
and
$\beta $ denotes the complete beta function.
The routine attempts to evaluate the scores so that the estimated error in each score is less than the value
etol specified by you. All integrations are performed in parallel and arranged so as to give good speed and reasonable accuracy.
4
References
Kendall M G and Stuart A (1969) The Advanced Theory of Statistics (Volume 1) (3rd Edition) Griffin
5
Arguments

1:
$\mathbf{n}$ – Integer
Input

On entry: $n$, the size of the set.
Constraint:
${\mathbf{n}}>0$.

2:
$\mathbf{pp}\left({\mathbf{n}}\right)$ – Real (Kind=nag_wp) array
Output

On exit: the Normal scores.
${\mathbf{pp}}\left(\mathit{i}\right)$ contains the value $E\left({x}_{\left(\mathit{i}\right)}\right)$, for $\mathit{i}=1,2,\dots ,n$.

3:
$\mathbf{etol}$ – Real (Kind=nag_wp)
Input

On entry: the maximum value for the estimated absolute error in the computed scores.
Constraint:
${\mathbf{etol}}>0.0$.

4:
$\mathbf{errest}$ – Real (Kind=nag_wp)
Output

On exit: a computed estimate of the maximum error in the computed scores (see
Section 7).

5:
$\mathbf{work}\left({\mathbf{iw}}\right)$ – Real (Kind=nag_wp) array
Workspace

6:
$\mathbf{iw}$ – Integer
Input

On entry: the dimension of the array
work as declared in the (sub)program from which
g01daf is called.
Constraints:
 if ${\mathbf{n}}\text{is even}$, ${\mathbf{iw}}\ge 3\times {\mathbf{n}}/2$;
 otherwise ${\mathbf{iw}}\ge 3\times \left({\mathbf{n}}1\right)/2$.

7:
$\mathbf{ifail}$ – Integer
Input/Output

On entry:
ifail must be set to
$0$,
$1\text{or}1$. If you are unfamiliar with this argument you should refer to
Section 4 in the Introduction to the NAG Library FL Interface for details.
For environments where it might be inappropriate to halt program execution when an error is detected, the value
$1\text{or}1$ is recommended. If the output of error messages is undesirable, then the value
$1$ is recommended. Otherwise, if you are not familiar with this argument, the recommended value is
$0$.
When the value $\mathbf{1}\text{or}\mathbf{1}$ is used it is essential to test the value of ifail on exit.
On exit:
${\mathbf{ifail}}={\mathbf{0}}$ unless the routine detects an error or a warning has been flagged (see
Section 6).
6
Error Indicators and Warnings
If on entry
${\mathbf{ifail}}=0$ or
$1$, explanatory error messages are output on the current error message unit (as defined by
x04aaf).
Errors or warnings detected by the routine:
 ${\mathbf{ifail}}=1$

On entry, ${\mathbf{n}}=\u2329\mathit{\text{value}}\u232a$.
Constraint: ${\mathbf{n}}>0$.
 ${\mathbf{ifail}}=2$

On entry, ${\mathbf{etol}}=\u2329\mathit{\text{value}}\u232a$.
Constraint: ${\mathbf{etol}}>0.0$.
 ${\mathbf{ifail}}=3$

The routine was unable to estimate the scores with estimated
error less than
etol. The best result obtained is returned together with the associated value of
errest.
 ${\mathbf{ifail}}=4$

On entry,
iw is too small. Minimum size required:
$\u2329\mathit{\text{value}}\u232a$.
 ${\mathbf{ifail}}=99$
An unexpected error has been triggered by this routine. Please
contact
NAG.
See
Section 7 in the Introduction to the NAG Library FL Interface for further information.
 ${\mathbf{ifail}}=399$
Your licence key may have expired or may not have been installed correctly.
See
Section 8 in the Introduction to the NAG Library FL Interface for further information.
 ${\mathbf{ifail}}=999$
Dynamic memory allocation failed.
See
Section 9 in the Introduction to the NAG Library FL Interface for further information.
7
Accuracy
Errors are introduced by evaluation of the functions
$d{G}_{r}$ and errors in the numerical integration process. Errors are also introduced by the approximation of the true infinite range of integration by a finite range
$\left[a,b\right]$ but
$a$ and
$b$ are chosen so that this effect is of lower order than that of the other two factors. In order to estimate the maximum error the functions
$d{G}_{r}$ are also integrated over the range
$\left[a,b\right]$.
g01daf returns the estimated maximum error as
8
Parallelism and Performance
g01daf is not threaded in any implementation.
The time taken by
g01daf depends on
etol and
n. For a given value of
etol the timing varies approximately linearly with
n.
10
Example
The program below generates the Normal scores for samples of size $5$, $10$, $15$, and prints the scores and the computed error estimates.
10.1
Program Text
10.2
Program Data
None.
10.3
Program Results
This shows a QQ plot for a randomly generated set of data. The normal scores have been calculated using
g01daf and the sample quantiles obtained by sorting the observed data using
m01caf. A reference line at
$y=x$ is also shown.