Page 38 - 2010 (3) Agricultural Census 2010, Punjab Report
P. 38

Volume 3: Special Reportscomprise  two reports i.e. one for Gilgit Baltistan and
                       other for Azad Jammu & Kashmir. Each report contains tabulations at the district
                       and overall levels in acres for the local users.


                       QUALITY  OF CENSUS DATA.

                       A sample-based census gives rise to sampling error as well as the non-sampling
                       error.ln spite of the best efforts made by the ACO during sample selection
                       procedures, data collection  and data processing methods,  some margin of non-
                       sampling errors is inevitable. Therefore, the most important consideration was to
                       control the extent of non-sampling errors during the census.  In addition, the quality of
                       the census data was considered  on the criterion of:

                           i)  Internal consistencies  and


                           ii) Comparison  with other corresponding series

                       The quality of data had been assured by the following  methods:

                        SAMPLING  ERROR:

                       To estimate sampling error the relative variance  (R.V  and then the coefficient of
                                                                                )
                        variation (C.V.)were  calculated  for a number of important  data items and presented
                        in Appendix-lll.  The formula used for estimating  relative varianceand  the coefficient
                        of vaiiationat district, province and country  levels is described  in the succeeding
                        paragraphs

                        At the district level, R. V. was estimated simply by first raising the sample
                        observations  of each Patwar Circle to stratum level using the overall raising factor for
                        each holding but excluding all National Certainty  Holdings  (NCHs), and then
                        summing  up to obtain an estimate for each half stratum as describedbelow:
                                     r, I  o, 1lvu-Ni-t'ti'1  ,,   Qt Ni  ,,   I
                               i,        \-    \-    ')'h'tjkl*  r-)\1'/h'tjkr   I
                                   -\-l
                               'n'r-LlL        L       p     'L/-t   p     I
                                     i  I  k    l'    r  h'tir  k  I'   'n'ti  -)


                                     ,, f   qi (N*  _Ni_N i\  y0,,,,-
                               Orr:411         s             *gf     zr*r]
                                               /J      Prrio 77       Prr,  )


                        Then, the R V for the district estimate of aggregates  of the attribute is obtained  by
                        taking the square of differences  between estimates for successive  pairs of half strata
                        divided by the district estimates  as follows:
                                                                         r  ^  ^    .t2

                                                                               y
                                                    \ / L  \ lJ
                                                                      TL
                                                                                    l
                                                vo,(f\=l cv (t\1' =fl"'="'  I
                        si mi rar'v' the R' *                                            bute is
                                              rii'ffi  l= ;i              ;:[jTri
                                               -\
                                                      lL  \  /t
                                                                      Tl  X
                                                                                      I
   33   34   35   36   37   38   39   40   41   42   43