How can I 'align' multiple two-way tabulations in esttab? - stata

I am trying to prepare a table that will display two-way frequency tables of several variables. The logic is that each of the variables will be tabulated by the same binary indicator.
I would like to send the output to a tex file using the estout community-contributed family of commands. However, each cross-tabulation appears in new column.
Consider the following reproducible toy example:
sysuse auto
eststo clear
eststo: estpost tab headroom foreign, notot
eststo: estpost tab trunk foreign, notot
esttab, c(b) unstack wide collabels(N)
----------------------------------------------------------------
(1) (2)
Domestic Foreign Domestic Foreign
N N N N
----------------------------------------------------------------
1_missing_5 3 1
2 10 3
2_missing_5 4 10
3 7 6
3_missing_5 13 2
4 10 0
4_missing_5 4 0
5 1 0 0 1
6 0 1
7 3 0
8 2 3
9 3 1
10 3 2
11 4 4
12 1 2
13 4 0
14 1 3
15 2 3
16 10 2
17 8 0
18 1 0
20 6 0
21 2 0
22 1 0
23 1 0
----------------------------------------------------------------
N 74 74
----------------------------------------------------------------
Is there a way to 'align' the output so that there are only two Domestic and Foreign columns?

If you're outputting to a tex file, one solution would be to use the append option of esttab. So in your case it would be something like:
sysuse auto
eststo clear
estpost tab headroom foreign, notot
eststo tab1
estpost tab trunk foreign, notot
eststo tab2
esttab tab1 using outputfile.tex, c(b) unstack wide collabels(N) replace
esttab tab2 using outputfile.tex, c(b) unstack wide collabels(N) append
I believe there may be a more elegant solution as well, but this is generally pretty easy to implement. When appending, you'll likely have to specify a bunch of options to remove the various column headers (I believe estout's default assumes you don't want most of those headers, so it may be worth looking into estout instead of esttab).

Producing the desired output requires that you stack the results together.
First define the program append_tabs, which is a quickly modified version of appendmodels, Ben Jann's program for stacking models:
program append_tabs, eclass
version 8
syntax namelist
tempname b tmp
local i 0
foreach name of local namelist {
qui est restore `name'
foreach x in Domestic Foreign {
local ++i
mat `tmp'`i' = e(b)
mat li `tmp'`i'
mat `tmp'`i' = `tmp'`i'[1,"`x':"]
local cons = colnumb(`tmp'`i',"_cons")
if `cons'<. & `cons'>1 {
mat `tmp'`i' = `tmp'`i'[1,1..`cons'-1]
}
mat li `tmp'`i'
mat `b'`i' = `tmp'`i''
mat li `b'`i'
}
}
mat `b'D = `b'1 \ `b'3
mat `b'F = `b'2 \ `b'4
mat A = `b'D , `b'F
ereturn matrix results = A
eret local cmd "append_tabs"
end
Next run your tabulations and stack their results using append_tabs:
sysuse auto, clear
estimates store clear
estpost tabulate headroom foreign, notot
estimates store one
estpost tabulate trunk foreign, notot
estimates store two
append_tabs one two
Finally, see the results:
esttab e(results), nonumber mlabels(none) eqlabels(none) collabels("Domestic" "Foreign")
--------------------------------------
Domestic Foreign
--------------------------------------
1_missing_5 3 1
2 10 3
2_missing_5 4 10
3 7 6
3_missing_5 13 2
4 10 0
4_missing_5 4 0
5 1 0
5 0 1
6 0 1
7 3 0
8 2 3
9 3 1
10 3 2
11 4 4
12 1 2
13 4 0
14 1 3
15 2 3
16 10 2
17 8 0
18 1 0
20 6 0
21 2 0
22 1 0
23 1 0
--------------------------------------
Use the tex option in the esttab command to see the LaTeX output.

Related

How can I export a table only containing cells with >N observations?

I want to export several tables from Stata to Excel but I am only interested in tables where each cell has more than 5 observations. Is there any way I can set a condition in order to do just that?
With the community-contributed command taboutI tried to do something along the following lines:
tabout var1 if n(cell) > 5
For cells with less than 5 observations, I would also like the exported table to show another value (such as . or *) instead of 0.
You can export such a table with the community-contributed command esttab:
sysuse auto, clear
bysort foreign rep78 : generate freq = _N
quietly tabulate foreign rep78 if freq >= 5, matcell(A)
forvalues i = 1 / 2 {
forvalues j = 1 / 4 {
if A[`i', `j'] == 0 matrix A[`i', `j'] = .
}
}
esttab matrix(A)
----------------------------------------------------------------
A
c1 c2 c3 c4
----------------------------------------------------------------
r1 8 27 9 .
r2 . . 9 9
----------------------------------------------------------------
A general approach is two passes, one to calculate cell frequency in a new variable and the other to tabulate conditional on that variable.
. sysuse auto
(1978 Automobile Data)
. tab foreign rep78
| Repair Record 1978
Car type | 1 2 3 4 5 | Total
-----------+-------------------------------------------------------+----------
Domestic | 2 8 27 9 2 | 48
Foreign | 0 0 3 9 9 | 21
-----------+-------------------------------------------------------+----------
Total | 2 8 30 18 11 | 69
. table foreign rep78
----------------------------------------
| Repair Record 1978
Car type | 1 2 3 4 5
----------+-----------------------------
Domestic | 2 8 27 9 2
Foreign | 3 9 9
----------------------------------------
. bysort foreign rep78 : generate freq = _N
. table foreign rep78 if freq >= 5
----------------------------------
| Repair Record 1978
Car type | 2 3 4 5
----------+-----------------------
Domestic | 8 27 9
Foreign | 9 9
----------------------------------
More general code would support if and in conditions.

Stata alternatives for lookup

I have a large Stata dataset that contains the following variables: year, state, household_id, individual_id, partner_id, and race. Here is an example of my data:
year state household_id individual_id partner_id race
1980 CA 23 2 1 3
1980 CA 23 1 2 1
1990 NY 43 4 2 1
1990 NY 43 2 4 1
Note that, in the above table, column 1 and 2 are married to each other.
I want to create a variable that is one if the person is in an interracial marriage.
As a first step, I used the following code
by household_id year: gen inter=0 if race==race[partner_id]
replace inter=1 if inter==.
This code worked well but gave the wrong result in a few cases. As an alternative, I created a string variable identifying each user and its partner, using
gen id_user=string(household_id)+"."+string(individual_id)+string(year)
gen id_partner=string(household_id)+"."+string(partner_id)+string(year)
What I want to do now is to create something like what vlookup does in Excel: for each column, save locally the id_partner, find it in the id_user and find their race, and compare it with the race of the original user.
I guess it should be something like this?
gen inter2==1 if (find race[idpartner]) == (race[iduser])
The expected output should be like this
year state household_id individual_id partner_id race inter2
1980 CA 23 2 1 3 1
1980 CA 23 1 2 1 1
1990 NY 43 4 2 1 0
1990 NY 43 2 4 1 0
I don't think you need anything so general. As you realise, the information on identifiers suffices to find couples, and that in turn allows comparison of race for the people in each couple.
In the code below _N == 2 is meant to catch data errors, such as one partner but not the other being an observation in the dataset or repetitions of one partner or both.
clear
input year str2 state household_id individual_id partner_id race
1980 CA 23 2 1 3
1980 CA 23 1 2 1
1990 NY 43 4 2 1
1990 NY 43 2 4 1
end
generate couple_id = cond(individual_id < partner_id, string(individual_id) + ///
" " + string(partner_id), string(partner_id) + ///
" " + string(individual_id))
bysort state year household_id couple_id : generate mixed = race[1] != race[2] if _N == 2
list, sepby(household_id) abbreviate(15)
+-------------------------------------------------------------------------------------+
| year state household_id individual_id partner_id race couple_id mixed |
|-------------------------------------------------------------------------------------|
1. | 1980 CA 23 2 1 3 1 2 1 |
2. | 1980 CA 23 1 2 1 1 2 1 |
|-------------------------------------------------------------------------------------|
3. | 1990 NY 43 4 2 1 2 4 0 |
4. | 1990 NY 43 2 4 1 2 4 0 |
+-------------------------------------------------------------------------------------+
This idea is documented in this article. The link gives free access to a pdf file.

Populating new variable using vlookup with multiple criteria in another variable

1) A new variable should be created for each unique observation listed in variable sku, which contains repeated values.
2) These newly created variables should be assigned the value of own product's price at the store/week level, as long as observations' sku value is in the same subcategory (subc) as the variable itself. For example, in eta2,3, observations in line 3, 4, and 5 have the same value because they all belong to the same subcategory as sku #3. [eta2,3 indicates sku 3, subc 2.]
3) x indicates that this is the original value for the product/subcategory that is currently being replicated.
4) If an observation doesn't belong to the same subcategory, it should reflect "0".
Orange is the given data. In green are the values from the steps 1, 2, and 3. White cells are step 4.
I am unable to offer a solution of my own, as searching for a
way to generate a variable using existing observations hasn't given me results.
I also understand that it must be a combination of forvalues, foreach, and levelsof commands?
clear
input units price sku week store subc
3 4.3 1 1 1 1
2 3 2 1 1 1
1 2.5 3 1 1 2
4 12 5 1 1 2
5 12 6 1 1 3
35 4.3 1 1 2 1
23 3 2 1 2 1
12 2.5 3 1 2 2
35 12 5 1 2 2
35 12 6 1 2 3
3 20 1 2 1 1
2 30 2 2 1 1
4 40 3 2 2 2
1 50 4 2 2 2
9 10 5 2 2 2
2 90 6 2 2 3
end
UPDATE
Based on Nick Cox' feedback, this is the final code that gives the result I have been looking for:
clear
input units price sku week store subc
35 4.3 1 1 1 1
23 3 2 1 1 1
12 2.5 3 1 1 2
10 1 4 1 1 2
35 12 5 1 1 2
35 12 6 1 1 3
35 5.3 1 2 1 1
23 4 2 2 1 1
12 3.5 3 2 1 2
10 2 4 2 1 2
35 13 5 2 1 2
35 13 6 2 1 3
end
egen joint = group(subc sku), label
bysort store week : gen freq = _N
su freq, meanonly
local jmax = r(max)
drop freq
tostring subc sku, replace
gen new = subc + "_"+sku
su joint, meanonly
forval j = 1/`r(max)'{
local J = new[`j']
gen eta`J' = .
}
sort subc week store sku
egen joint1 = group(subc week store), label
gen long id = _n
su joint1, meanonly
quietly forval i = 1/`r(max)' {
su id if joint1 == `i', meanonly
local jmin = r(min)
local jmax = r(max)
forval j = `jmin'/`jmax' {
local subc = subc[`j']
local sku = sku[`j']
replace eta`subc'_`sku' = price[`j'] in `jmin'/`jmax'
replace eta`subc'_`sku' = 0 in `j'/`j'
}
}
I worry on your behalf that in a dataset of any size what you ask for would mean many, many extra variables. I wonder on your behalf whether you need all of them any way for whatever you want to do with them.
That aside, this seems to be what you want. Naturally your column headers in your spreadsheet view aren't legal variable names. Disclosure: despite being the original author of levelsof I wouldn't prefer its use here.
clear
input units price sku week store subc
35 4.3 1 1 1 1
23 3 2 1 1 1
12 2.5 3 1 1 2
10 1 4 1 1 2
35 12 5 1 1 2
35 12 6 1 1 3
end
sort subc sku
* subc identifiers guaranteed to be integers 1 up
egen subc_id = group(subc), label
* observation numbers in a variable
gen long id = _n
* how many subc? loop over the range
su subc_id, meanonly
forval i = 1/`r(max)' {
* which subc is this one? look it up using -summarize-
* assuming that subc is numeric!
su subc if subc_id == `i', meanonly
local I = r(min)
* which observation numbers for this subc?
* given the prior sort, they are all contiguous
su id if subc_id == `i', meanonly
* for each observation in the subc, find out the sku and copy its price
* to all observations in that subc
forval j = `r(min)'/`r(max)' {
local J = sku[`j']
gen eta_`I'_`J' = cond(subc_id == `i', price[`j'], 0)
}
}
list subc eta*, sepby(subc)
+------------------------------------------------------------------+
| subc eta_1_1 eta_1_2 eta_2_3 eta_2_4 eta_2_5 eta_3_6 |
|------------------------------------------------------------------|
1. | 1 4.3 3 0 0 0 0 |
2. | 1 4.3 3 0 0 0 0 |
|------------------------------------------------------------------|
3. | 2 0 0 2.5 1 12 0 |
4. | 2 0 0 2.5 1 12 0 |
5. | 2 0 0 2.5 1 12 0 |
|------------------------------------------------------------------|
6. | 3 0 0 0 0 0 12 |
+------------------------------------------------------------------+
Notes:
N1. In your example, subc is numbered 1, 2, etc. My extra variable subc_id ensures that to be true even if in your real data the identifiers are not so clean.
N2. The expression
cond(subc_id == `i', price[`j'], 0)
could also be
(subc_id == `i') * price[`j']
N3. It seems possible that a different data structure would be much more efficient.
EDIT: Here is code and results for another data structure.
clear
input units price sku week store subc
35 4.3 1 1 1 1
23 3 2 1 1 1
12 2.5 3 1 1 2
10 1 4 1 1 2
35 12 5 1 1 2
35 12 6 1 1 3
end
sort subc sku
egen subc_id = group(subc), label
bysort subc : gen freq = _N
su freq, meanonly
local jmax = r(max)
drop freq
forval j = 1/`jmax' {
gen eta`j' = .
gen which`j' = .
}
gen long id = _n
su subc_id, meanonly
quietly forval i = 1/`r(max)' {
su id if subc_id == `i', meanonly
local jmin = r(min)
local jmax = r(max)
local k = 1
forval j = `jmin'/`jmax' {
replace which`k' = sku[`j'] in `jmin'/`jmax'
replace eta`k' = price[`j'] in `jmin'/`jmax'
local ++k
}
}
list subc sku *1 *2 *3 , sepby(subc)
+------------------------------------------------------------+
| subc sku eta1 which1 eta2 which2 eta3 which3 |
|------------------------------------------------------------|
1. | 1 1 4.3 1 3 2 . . |
2. | 1 2 4.3 1 3 2 . . |
|------------------------------------------------------------|
3. | 2 3 2.5 3 1 4 12 5 |
4. | 2 4 2.5 3 1 4 12 5 |
5. | 2 5 2.5 3 1 4 12 5 |
|------------------------------------------------------------|
6. | 3 6 12 6 . . . . |
+------------------------------------------------------------+
I am adding another answer that tackles combinations of subc and week. Previous discussion establishes that what you are trying to do would add an extra variable for every observation. This can't be a good idea! At best, you might just have many new variables, mostly zeros. At worst, you will run into Stata's limits.
Hence I won't support your endeavour to go further down the same road, but show how the second data structure I discuss in my previous answer can be produced. Indeed, you haven't indicated (a) why you want all these variables, which are just the existing data redistributed; (b) what your strategy is for dealing with them; (c) why rangestat (SSC) or some other program could not remove the need to create them in the first place.
clear
input units price sku week store subc
35 4.3 1 1 1 1
23 3 2 1 1 1
12 2.5 3 1 1 2
10 1 4 1 1 2
35 12 5 1 1 2
35 12 6 1 1 3
35 5.3 1 2 1 1
23 4 2 2 1 1
12 3.5 3 2 1 2
10 2 4 2 1 2
35 13 5 2 1 2
35 13 6 2 1 3
end
sort subc week sku
egen joint = group(subc week), label
bysort joint : gen freq = _N
su freq, meanonly
local jmax = r(max)
drop freq
forval j = 1/`jmax' {
gen eta`j' = .
gen which`j' = .
}
gen long id = _n
su joint, meanonly
quietly forval i = 1/`r(max)' {
su id if joint == `i', meanonly
local jmin = r(min)
local jmax = r(max)
local k = 1
forval j = `jmin'/`jmax' {
replace which`k' = sku[`j'] in `jmin'/`jmax'
replace eta`k' = price[`j'] in `jmin'/`jmax'
local ++k
}
}
list subc week sku *1 *2 *3 , sepby(subc week)
+-------------------------------------------------------------------+
| subc week sku eta1 which1 eta2 which2 eta3 which3 |
|-------------------------------------------------------------------|
1. | 1 1 1 4.3 1 3 2 . . |
2. | 1 1 2 4.3 1 3 2 . . |
|-------------------------------------------------------------------|
3. | 1 2 1 5.3 1 4 2 . . |
4. | 1 2 2 5.3 1 4 2 . . |
|-------------------------------------------------------------------|
5. | 2 1 3 2.5 3 1 4 12 5 |
6. | 2 1 4 2.5 3 1 4 12 5 |
7. | 2 1 5 2.5 3 1 4 12 5 |
|-------------------------------------------------------------------|
8. | 2 2 3 3.5 3 2 4 13 5 |
9. | 2 2 4 3.5 3 2 4 13 5 |
10. | 2 2 5 3.5 3 2 4 13 5 |
|-------------------------------------------------------------------|
11. | 3 1 6 12 6 . . . . |
|-------------------------------------------------------------------|
12. | 3 2 6 13 6 . . . . |
+-------------------------------------------------------------------+
clear
input units price sku week store subc
35 4.3 1 1 1 1
23 3 2 1 1 1
12 2.5 3 1 1 2
10 1 4 1 1 2
35 12 5 1 1 2
35 12 6 1 1 3
35 5.3 1 2 1 1
23 4 2 2 1 1
12 3.5 3 2 1 2
10 2 4 2 1 2
35 13 5 2 1 2
35 13 6 2 1 3
end
egen joint = group(subc sku), label
bysort store week : gen freq = _N
su freq, meanonly
local jmax = r(max)
drop freq
tostring subc sku, replace
gen new = subc + "_"+sku
su joint, meanonly
forval j = 1/`r(max)'{
local J = new[`j']
gen eta`J' = .
}
sort subc week store sku
egen joint1 = group(subc week store), label
gen long id = _n
su joint1, meanonly
quietly forval i = 1/`r(max)' {
su id if joint1 == `i', meanonly
local jmin = r(min)
local jmax = r(max)
forval j = `jmin'/`jmax' {
local subc = subc[`j']
local sku = sku[`j']
replace eta`subc'_`sku' = price[`j'] in `jmin'/`jmax'
replace eta`subc'_`sku' = 0 in `j'/`j'
}
}
list subc sku store week eta*, sepby(subc)
+---------------------------------------------------------------------------------+
| store week subc sku eta1_1 eta1_2 eta2_3 eta2_4 eta2_5 eta3_6 |
|---------------------------------------------------------------------------------|
1. | 1 1 1 2 4.3 0 . . . . |
2. | 1 1 1 1 0 3 . . . . |
|---------------------------------------------------------------------------------|
3. | 1 1 2 4 . . 2.5 0 12 . |
4. | 1 1 2 3 . . 0 1 12 . |
5. | 1 1 2 5 . . 2.5 1 0 . |
|---------------------------------------------------------------------------------|
6. | 1 1 3 6 . . . . . 0 |
|---------------------------------------------------------------------------------|
7. | 1 2 1 2 5.3 0 . . . . |
8. | 1 2 1 1 0 4 . . . . |
|---------------------------------------------------------------------------------|
9. | 1 2 2 3 . . 0 2 13 . |
10. | 1 2 2 5 . . 3.5 2 0 . |
11. | 1 2 2 4 . . 3.5 0 13 . |
|---------------------------------------------------------------------------------|
12. | 1 2 3 6 . . . . . 0 |
+---------------------------------------------------------------------------------+

Looking up data within a file versus merging

I have a file that look at ratings that teacher X gives to teacher Y and the date it occurs
clear
rating_id RatingTeacher RatedTeacher Rating Date
1 15 12 1 "1/1/2010"
2 12 11 2 "1/2/2010"
3 14 11 3 "1/2/2010"
4 14 13 2 "1/5/2010"
5 19 11 4 "1/6/2010"
5 11 13 1 "1/7/2010"
end
I want to look in the history to see how many times the RatingTeacher had been rated at the time they make the rating and the cumulative score. The result would look like this.
rating_id RatingTeacher RatedTeacher Rating Date TimesRated CumulativeRating
1 15 12 1 "1/1/2010" 0 0
2 12 11 2 "1/2/2010" 1 1
3 14 11 3 "1/2/2010" 0 0
4 14 13 2 "1/5/2010" 0 0
5 19 11 4 "1/6/2010" 0 0
5 11 13 1 "1/7/2010" 3 9
end
I have been merging the dataset with itself to get this to work, and it is fine. I was wondering if there was a more efficient way to do this within the file
In your input data, I guess that the last rating_id should be 6 and that dates are MDY. Statalist members are asked to use dataex (SSC) to set up data examples. This isn't Statalist but there is no reason for lower standards to apply. See the Statalist FAQ
I rarely see even programmers be precise about what they mean by "efficient", whether it means fewer lines of code, less use of memory, more speed, something else or is just some all-purpose term of praise. This code loops over observations, which can certainly be slow for large datasets. More in this paper
We can't compare with your merge solution because you don't give the code.
clear
input rating_id RatingTeacher RatedTeacher Rating str8 SDate
1 15 12 1 "1/1/2010"
2 12 11 2 "1/2/2010"
3 14 11 3 "1/2/2010"
4 14 13 2 "1/5/2010"
5 19 11 4 "1/6/2010"
6 11 13 1 "1/7/2010"
end
gen Date = daily(SDate, "MDY")
sort Date
gen Wanted = .
quietly forval i = 1/`=_N' {
count if Date < Date[`i'] & RatedT == RatingT[`i']
replace Wanted = r(N) in `i'
}
list, sep(0)
+---------------------------------------------------------------------+
| rating~d Rating~r RatedT~r Rating SDate Date Wanted |
|---------------------------------------------------------------------|
1. | 1 15 12 1 1/1/2010 18263 0 |
2. | 2 12 11 2 1/2/2010 18264 1 |
3. | 3 14 11 3 1/2/2010 18264 0 |
4. | 4 14 13 2 1/5/2010 18267 0 |
5. | 5 19 11 4 1/6/2010 18268 0 |
6. | 6 11 13 1 1/7/2010 18269 3 |
+---------------------------------------------------------------------+
The building block is that the rater and ratee are a pair. You can use egen's group() to give a unique ID to each rater ratee pair.
egen pair = group(rater ratee)
bysort pair (date): timesRated = _n

subset of dataset using first and last in sas

Hi I am trying to subset a dataset which has following
ID sal count
1 10 1
1 10 2
1 10 3
1 10 4
2 20 1
2 20 2
2 20 3
3 30 1
3 30 2
3 30 3
3 30 4
I want to take out only those IDs who are recorded 4 times.
I wrote like
data AN; set BU
if last.count gt 4 and last.count lt 4 then delete;
run;
But there is something wrong.
EDIT - Thanks for clarifying. Based on your needs, PROC SQL will be more direct:
proc sql;
CREATE TABLE AN as
SELECT * FROM BU
GROUP BY ID
HAVING MAX(COUNT) = 4
;quit;
For posterity, here is how you could do it with only a data step:
In order to use first. and last., you need to use a by clause, which requires sorting:
proc sort data=BU;
by ID DESCENDING count;
run;
When using a SET statement BY ID, first.ID will be equal to 1 (TRUE) on the first instance of a given ID, 0 (FALSE) for all other records.
data AN;
set BU;
by ID;
retain keepMe;
If first.ID THEN DO;
IF count = 4 THEN keepMe=1;
ELSE keepMe=0;
END;
if keepMe=0 THEN DELETE;
run;
During the datastep BY ID, your data will look like:
ID sal count keepMe first.ID
1 10 4 1 1
1 10 3 1 0
1 10 2 1 0
1 10 1 1 0
2 20 3 0 1
2 20 2 0 0
2 20 1 0 0
3 30 4 1 1
3 30 3 1 0
3 30 2 1 0
3 30 1 1 0
If I understand correct, you are trying to extract all observations are are repeated 4 time or more. if so, your use of last.count and first.count is wrong. last.var is a boolean and it will indicate which observation is last in the group. Have a look at Tim's suggestion.
In order to extract all observations that are repeated four times or more, I would suggest to use the following PROC SQL:
PROC SQL;
CREATE TABLE WORK.WANT AS
SELECT /* COUNT_of_ID */
(COUNT(t1.ID)) AS COUNT_of_ID,
t1.ID,
t1.SAL,
t1.count
FROM WORK.HAVE t1
GROUP BY t1.ID
HAVING (CALCULATED COUNT_of_ID) ge 4
ORDER BY t1.ID,
t1.SAL,
t1.count;
QUIT;
Result:
1 10 1
1 10 2
1 10 3
1 10 4
3 30 1
3 30 2
3 30 3
3 30 4
Slight variation on Tims - assuming you don't necessarily have the count variable.
proc sql;
CREATE TABLE AN as
SELECT * FROM BU
GROUP BY ID
HAVING Count(ID) >= 4;
quit;