Operational Problems and Errors

As most of the data are gathered from Census, we should expect the data to be relatively accurately but not error-free.  For if we look closer at the raw metadata from Census, it is obvious that the sums of many sub-categories does not correspond to the grouped total in the same table (which the two should have been identical).  This happens with all derived data, that we see increments of 5s in each field.  However, such minor difference should have minimal or no effect on general accuracy of data.

Another source of error is that the Census data are from 1996, this has two shortcomings: First, the data itself is considered outdated simply because it was taken from five years ago.  People might have moved from one place to another, many newcomers might have moved in during this five-year period.  Second, because the lag between the last Census and now is 5 years, as a result the 60 to 64 age group in 1996, is now 65 to 69 and is officially being classified as senior citizens (by Census data).  Consequently, the analysis could not have taken this group of senior citizens into account.

Finally, as with most other complex derived data set, the sample size is only a fifth of raw data set (i.e.: 20% of sample data). On the other hand, it is impossible to derive raw data into usable derived data.  For example, we do have the exact number of persons aged 65 and above, in each five-year age group.  Unfortunately with age alone, it is impossible to come up with a sensible set of results reflecting the actual need for each community.  Nevertheless, the data should be representative enough for us to make decisions.

< BACK                                                   back to Index                                           NEXT>


RESEARCH   DATA MANIPULATION    METHODOLOGY  SPATIAL ANALYSIS 1   SPATIAL ANALYSIS 2    PROBLEMS AND ERRORS   REFERENCE