HIV                 package:Rcapture                 R Documentation

_C_l_o_s_e_d _P_o_p_u_l_a_t_i_o_n _E_p_i_d_e_m_i_o_l_o_g_i_c_a_l _D_a_t_a

_D_e_s_c_r_i_p_t_i_o_n:

     Epidemiological capture-recapture data on HIV from four reporting
     centers in Rome, Italy.

_U_s_a_g_e:

     data(HIV)

_F_o_r_m_a_t:

     '_c_1', '_c_2', '_c_3', '_c_4' Capture histories for the four capture
          occasions

     _f_r_e_q Observed frequencies for each capture history

_D_e_t_a_i_l_s:

     The capture histories are obtained by linking the records of the
     four reporting centers. 

     This data set's format is the alternative one, i.e. each row
     represents an observed capture history followed by its frequency.

_S_o_u_r_c_e:

     Abeni, D.A., Brancato, G. and Perucci, C. A. (1994)
     Capture-recapture to estimate the size of the population with
     human immunodeficiency virus type 1 infection.  _Epidemiology_, 5,
     410-414

_R_e_f_e_r_e_n_c_e_s:

     Baillargeon, S. and Rivest, L.P. (2007). The Rcapture package:
     Loglinear models for capture-recapture in R. _Journal of
     Statistical Software_, to appear (available online at <URL:
     http://www.jstatsoft.org/>).

_E_x_a_m_p_l_e_s:

     data(HIV)
     desc<-descriptive(HIV,dfreq=TRUE)
     desc
       # 1774 out of 1896 individuals (94%) appear on one list only.

     plot(desc)
       # The fi plot is linear showing that heterogeneity is not a problem.  
       # Models with a time (or list) effect and possible pairwise dependencies 
       # between lists will be considered.

     mat<-histpos.t(4)
     mX1<-cbind(mat,mat[,1]*mat[,2],mat[,1]*mat[,3],mat[,1]*mat[,4],
                mat[,2]*mat[,3],mat[,2]*mat[,4],mat[,3]*mat[,4])
     cp.m1<-closedp.mX(HIV,dfreq=TRUE,mX=mX1,mname="Mt double interaction")
     cp.m1
       # The model fits well. Let's find out the interactions that are important.
     summary(cp.m1$glm)$coefficients
       # Eliminating the non significant interactions stepwise shows that only 
       # the [1,2] interaction is important.

     mX2<-cbind(mat,mat[,1]*mat[,2])
     closedp.mX(HIV,dfreq=TRUE,mX=mX2,mname="Mt interaction 1,2")
     profileCI(HIV,dfreq=TRUE,mX=mX2,mname="Mt interaction 1,2")

