.TH RANKSORT 1 "November 28, 1986" "\(co 1980 Gary Perlman" "|STAT" "UNIX User's Manual"
.SH NAME
ranksort \- rank order columns
.SH SYNOPSIS
.B ranksort
[-r] [-l lines]
.SH DESCRIPTION
.I ranksort
reads the lines from the standard input and rank orders each column.
Blank lines are ignored.
Ties share the ranks, so if the 13th and 14th ranked numbers
are equal, then each will be assigned the rank 13.5.
.PP
.I ranksort
is used on data when the assumptions for procedures like the normal theory
Pearson correlation coefficient are suspected to be false.
Under such conditions, the Spearman rho rank order correlation coefficient
is more appropriate.
The Spearman rho is equal to the Pearson calculation on the data
converted to ranks.
The normal theory (F or t) significance test of the correlation coefficient
is a good approximation to the two tailed test of the rank order
correlation when the number of pairs is greater than 10.
For smaller samples, a table should be consulted.
.SH OPTIONS
.de OP
.TP
.B -\\$1 \\$2
..
.OP l lines
Set the maximum number of lines to be read.
.OP r
Reverse the order of the rankings.
By default,
rank orderings correlate with input data
so that smaller numbers get smaller rank order values.
.SH EXAMPLE
.nf
.ta .25i +.5i +.5i +.5i +.5i +.5i +.5i +.5i +.5i +.5i
	Input				Output
	0	5	0	2		1	6	1	5.5	
	1	6	10	3		2	7	2	7.5	
	2	7	20	3		3	8	3	7.5	
	3	8	30	4		4	9	4	9.5	
	4	9	40	4		5	10	5	9.5	
	5	0	50	0		6	1	6	1.5	
	6	1	60	0		7	2	7	1.5	
	7	2	70	1		8	3	8	3.5	
	8	3	80	1		9	4	9	3.5	
	9	4	90	2		10	5	10	5.5	
.fi
.SH LIMITS
Use the -L option to determine the program limits.
.SH "SEE ALSO
rankrel(1) and rankind(1) perform rank order statistics.