Re: st: RE: Percentiles: Identify of smallest observation closetto percentile value

[Rod Hunter <info@hunarch.com.au>]

Thanks Nick and Ronnie.
I still don't appear to be out of the woods.

1.
I applied Nick's procedure to my example:
. centile var1 , c(75)
--------------------------------------------------------------------
				-- Binom. Interp. --
	Variable	Obs  Percentile	Centile	[95% Conf. Interval]
				
	var1	4         75	9.75	7.45          10*

*	Lower (upper)	confidence limit held	at minimum	(maximum) of sample
-----------------------------------------------------------
. ret li
------------------------------------------------
scalars:
	r(n_cent)	=	1
	r(N)	=	4
	r(ub_1)	=	10
	r(lb_1)	=	7.450000000008672
	r(c_1)	=	9.75
macros:
	r(centiles)	:	"75"
-------------------------------------------------
. list var1 if var1 == `r(c_1)'
-----------------------------------------------
	var1
.
-----------------------------------------------
This does not give me the identity nor value of the largest observation 
within the percentage of the observations included by the 75th percentile.

2.
I tried Ronnie's approach:

. sum var1, d
-------------------------------------------
		var1
	Percentiles	Smallest
1%	7	7
5%	7	8
10%	7	9	Obs	4
25%	7.5	10	Sum of Wgt.	4

50%	8.5		Mean	8.5
		Largest	Std. Dev.	1.290994
75%	9.5	7
90%	10	8	Variance	1.666667
95%	10	9	Skewness	0
99%	10	10	Kurtosis	1.64
---------------------------------------------------
. local macro*
. list var1 if var1 == `p75'
-------------------------------------------------
	var1
-------------------------------------------------

3.
This is what I have been using:

. _pctile  var1,p(75)
. gen var1_p75=r(r1)
. tabstat var1 if var1==var1_p75, stat(max)
-------------------------------------
no observations
-------------------------------------
alternatively:
. list var1 if var1==var1_p75
---------------------------------------
 	var1
---------------------------------------

Any further suggestions?

Regards,

Rod Hunter






Nick Cox wrote:

> Rod Hunter
>
> > A simple eg. will suffice.
> >
> > Obs. Var1
> > ----------
> > 1     	10
> > 2	9
> > 3	8
> > 4	7
> > -----------
> >
> > I would like Stata to tell me the value of the observation
> > that is at or
> >  below the 75th percentile point, and to yield the identity (or
> > identities) of the observation(s).
>
> I'll change your example. I don't know about "below". ~ 75% are
> below; do you what them all?
>
> . use auto
> . centile rep78 , c(75)
>
>                                                        -- Binom.
> Interp. --
>     Variable |     Obs  Percentile      Centile        [95% Conf.
> Interval]
> -------------+--------------------------------------------------------
> -----
>        rep78 |      69         75             4               4
> 5
>
> . ret li
>
> scalars:
>             r(n_cent) =  1
>                  r(N) =  69
>               r(ub_1) =  5
>               r(lb_1) =  4
>                r(c_1) =  4
>
> macros:
>           r(centiles) : "75"
>
> . list rep78 if rep78 == `r(c_1)'
>
>      +-------+
>      | rep78 |
>      |-------|
>   5. |     4 |
>  15. |     4 |
>  24. |     4 |
>  29. |     4 |
>  30. |     4 |
>      |-------|
>  33. |     4 |
>  35. |     4 |
>  38. |     4 |
>  47. |     4 |
>  55. |     4 |
>      |-------|
>  56. |     4 |
>  58. |     4 |
>  59. |     4 |
>  62. |     4 |
>  63. |     4 |
>      |-------|
>  70. |     4 |
>  72. |     4 |
>  73. |     4 |
>      +-------+
> Nick
> n.j.cox@durham.ac.uk
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/