Developing the Sampling Framework

• This was an extensive process in order to ensure footprint sampling accuracy and was a joint excercise between Kuper Research, Spacial Technologies and African Response.
• Step 1 was to get details of the distribution areas - provided by the publications and Ads24.
• Step 2 was to use geo-coding of the distribution areas/points as provided for each of the titles to establish the number of households and race breakdown of populations in the areas.
• This was a joint exercise between Kuper Research and Spatial Technologies in order to provide African Response with the relevant areas in which to draw the survey samples.
  

• There were different considerations to take into account for the home delivered, sold and bulk distribution titles.
• For free home delivered titles we focused on the distribution footprint.
• Sampling was concentrated in areas of distribution.
• We also took into consideration the major arterial roads and the geography of the area.

Sampling Scrutiny

• In order to ensure that the area sampling was as accurate as possible bearing census constraints in mind, we checked a number of key components to ensure that the household and population numbers stood scrutiny.
• For example, if there were 100 households in an area and the publication was a free distribution newspaper, then the number of copies could not exceed 100, unless some of the distribution was handouts or bulk.
• To check the latter, we accessed ABC VFD certificates and compared the numbers from the sampling frame with the distribution numbers (where VFDs are available).
• We also compared the households and profiles from Compass24 2005, though in many cases the distribution areas had changed or the publication itself was new and had no precedent in the previous survey.

• Previous Compass data was population based and we were not supplied with household data.
• This year we have looked at both household and population data.
• Multiple “census” data sources were used to assess the composition and size of the publication distribution areas.
• All data sources, however, start with the 2001 census and then make adjustments.
• It should be noted that the Cape in particular is revealing rapid demographic changes in many areas.
• For many publications the sampling areas have been broadened to include the full distribution footprint rather than just the reader heartland.
  

Limitations

• The area weighting for each newspaper footprint was limited in terms of the availability of official population statistics.
• Each footprint was broken down into SP codes to first define the footprint area and then further broken down to SAL codes to design the sample frame.
• There were instances where SP codes could not be found for areas detailed in the distribution data provided to us. These areas have subsequently been left out of the household and population totals and contribute to the 100%+ coverage in some areas.
  

The Maps and the Website

• The maps are provided by Spatial Technologies and illustrate the boundaries of the footprints and the households within the footprint.
• As far as is possible they show main roads and major retailers.
• The website is again pen to all Ads24 clients.