811_ftp

2
 STATI STICS IN MEDICINE Statist. Med.  2001;  20:655 CORRECTION The following are cor rec tions to Kim, H-J, Fay MP, Feu er EJ, Midthune DN. Per mut ati on tests for joinpoint regression with applications to cancer rates.  Statis tics in Medi cine  2000; 19:335–351: 1. In Ta bl e I, for eac h set of ( 1 ; 2 ); ( 1  = 80; 90, and   2 = 0:0001; 0:0002; 0:000 5 and 0.001 0), Mean estimated  jp ± SE  for (apc 1 ; apc 2 )=(3; 2:4) sho uld be exchan ged with mean estimated  jp ± SE  for (apc 1 ; apc 2 )=(3; 1:5). 2. In Section 2.3, den otes the covar iance mat rix of the error . For ij  of autocorrelated data with lag 1, the correct form of the  ijth element of is ij  = 2 |i j| = (1 2 ). 3. In Section 2.4, the probabi lity of concludi ng exactly k  joinpoints when there is no change is at most (  N J  P k  + 1)=N J  P  instead of  k=N J  P . See the twelfth line from the bottom of the page. In addition, the program to t joinpoint regression can be downloaded from the web site http:== www-dccps.ims.nci.nih.gov = SRAB=  joinpoint= index.html. HYUNE-JU  K IM Department of Mathematics Syracuse University  215 Carnegie Hall Syracuse, NY 13244-1150 U.S.A. Copyright  ?  2001 John Wiley & Sons, Ltd.

Upload: eugenio-martinez

Post on 03-Nov-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

  • STATISTICS IN MEDICINEStatist. Med. 2001; 20:655

    CORRECTION

    The following are corrections to Kim, H-J, Fay MP, Feuer EJ, Midthune DN. Permutationtests for joinpoint regression with applications to cancer rates. Statistics in Medicine 2000;19:335{351:

    1. In Table I, for each set of (1; 2); (1 = 80; 90, and 2 = 0:0001; 0:0002; 0:0005 and0.0010), Mean estimated jp SE for (apc1; apc2)= (3; 2:4) should be exchanged withmean estimated jp SE for (apc1; apc2)= (3; 1:5).

    2. In Section 2.3, denotes the covariance matrix of the error. For ij of autocorrelateddata with lag 1, the correct form of the ijth element of is ij=2jijj=(1 2).

    3. In Section 2.4, the probability of concluding exactly k joinpoints when there is no changeis at most (NJP k +1)=NJP instead of k=NJP. See the twelfth line from the bottom ofthe page.

    In addition, the program to t joinpoint regression can be downloaded from the web sitehttp:==www-dccps.ims.nci.nih.gov=SRAB=joinpoint=index.html.

    HYUNE-JU KIMDepartment of Mathematics

    Syracuse University215 Carnegie Hall

    Syracuse, NY 13244-1150U.S.A.

    Copyright ? 2001 John Wiley & Sons, Ltd.