[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
"Nick Cox" <n.j.cox@durham.ac.uk> |

To |
<statalist@hsphsun2.harvard.edu> |

Subject |
st: -tabcond- available from SSC |

Date |
Thu, 7 Nov 2002 15:30:35 -0000 |

On Saturday, Kazi Ali Toufique posted a problem about tabulating all codes, specifically a desire to see all zero counts made explicit. In response, a -tabcount- hack was posted. Kit Baum reminded me privately that one way of seeing this problem was as a special case of a problem we call the -tabcond- problem, namely You want a table showing how many observations satisfy one or more conditions. That is, you have a series of true or false conditions which, in general, need not be mutually exclusive and need not refer to any single variable. A couple of programs we produced a while back, although producing more or less the tables desired, did not seem to be worth publicising for various reasons, including an ugly syntax. But Kazi's question was the trigger to have another go, and as the headline indicates, a -tabcond- based on a quite different approach is now available from SSC, thanks to Kit. What's much more noteworthy than the program itself is the approach it employs: user-programmers may write a wrapper for -tabdisp-, generating the contents of a table on the fly. Examples will indicate what -tabcond- does. -tabcond groups- gives, for the distinct groups of a variable, the numbers of observations satisfying specified conditions. I sometimes like to see how tail frequencies go down as you go outwards: . tabcond g rep78 , cond(mpg>15 mpg>20 mpg>25 mpg>30 mpg>35) ---------------------------------------------------------------------- Repair | Record | 1978 | mpg>15 mpg>20 mpg>25 mpg>30 mpg>35 ----------+----------------------------------------------------------- 1 | 2 1 0 0 0 2 | 7 3 0 0 0 3 | 25 10 3 0 0 4 | 15 12 3 0 0 5 | 11 7 6 5 1 ---------------------------------------------------------------------- Here the conditions can all be written tightly without spaces. If you want spaces to be echoed to column headings, or the condition cannot be written without embedded spaces, you must bind a condition in " ": . tabcond g rep78 , cond(mpg>25 price<5000 "mpg>25 & price<5000") ---------------------------------------------------------------------- --- Repair | Record | 1978 | mpg>25 price<5000 mpg>25 & price<5000 ----------+----------------------------------------------------------- --- 1 | 0 2 0 2 | 0 4 0 3 | 3 17 3 4 | 3 6 3 5 | 6 5 5 ---------------------------------------------------------------------- --- -tabcond variables- gives for each of a specified varlist how many observations satisfy a generic condition, expressed using a wildcard @: . tabcond var price-foreign, cond("@ < 0" "@ == 0" "@ > 0 & @ < ." "@ == .") listwise ---------------------------------------------------------------------- ------------- Variable | @ < 0 @ == 0 @ > 0 & @ < . @ == . -----------------------+---------------------------------------------- ------------- Price | 0 0 74 0 Mileage (mpg) | 0 0 74 0 Repair Record 1978 | 0 0 69 5 Headroom (in.) | 0 0 74 0 Trunk space (cu. ft.) | 0 0 74 0 Weight (lbs.) | 0 0 74 0 Length (in.) | 0 0 74 0 Turn Circle (ft.) | 0 0 74 0 Displacement (cu. in.) | 0 0 74 0 Gear Ratio | 0 0 74 0 Car type | 0 52 22 0 ---------------------------------------------------------------------- ------------- That may suggest an application to multiple response data. Given say five variables which are the first, second, ... choices from a list of five, then the cross-tabulation can be done by something resembling . tabcond var r1-r5, cond(@==1 @==2 @==3 @==4 @==5) However, -tabm- from the -tab_chi- package on SSC is better for that purpose. A limitation on -tabcond- is that no more than 5 conditions can be specified. This is inherited from a documented limit in -tabdisp-. To download -tabcond- in an up-to-date Stata, type . ssc inst tabcond 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/

- Prev by Date:
**st: Re: hausman test** - Next by Date:
**st: RE: bug in xpose.ado?** - Previous by thread:
**st: RE: bug in xpose.ado?** - Next by thread:
**st: Stratified Wilcoxon Test** - Index(es):

© Copyright 1996–2015 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |