Too much sample or just not enough: 
Sharper sample selection features to fulfil quota levels

A new version of the MARSC sampling tool is scheduled for release shortly representing a big step up in the sophistication of the sampling capability it currently offers to users.  Some of the new features available are outlined below.

Hit rates at the quota cell level:

Current users of MARSC will be familiar with its ability to calculate expected interview hit rates based on the history of work with a particular agency (or client).  These hit rates are then used to calculate how much sample should be issued to achieve the required number of interviews.  This facility has been expanded in version 4.0 to calculate interview hit rates at the target matrix cell level.  This means that the user can boost sample in cells where hit rates are normally low (e.g. 18-24 year olds), and save sample in cells where interviews are easier to achieve (e.g. retired women). 

In addition you can now apply filters such as “Type of Project” or “Preferred Contact Mode” to add an extra level of sophistication to the hit rate calculation.  In situations where sample is often in short supply, this type of hit rate calculation will help the researcher make much better use of the sample available.

Non-Interlocking cells

The new version also has the ability to define sample using non-interlocking quota cells.  This is often needed in the situation where for example, sample is required over a range of age groups and regions, but there is much more sample available in one region than in any others.  MARSC can adjust the sample to use more in the region where plenty is available while maintaining the marginal totals for each age range.  It can either do this if sample actually runs out in some regions, in which case it will top up from a plentiful region; or, if plenty is available across all sample cells, it can use the Deming Stephan algorithm to do a “least squares” fit to balance the sample out in proportion to its availability in each cell.  This enables the user to use sample wisely across regions and maintain a good balance of sample for future projects.

Single dimension with sub-totals

In this case a dimension might be a region and there may be areas or postal/zip codes within each region.  The new MARSC feature will allow a regional total to be specified and MARSC can balance out the sample within that region according to what is available in each postal/zip code.  Again this is a feature that allows the sample to be balanced against availability thereby leaving sample for future project needs.

For further information about MARSC please contact Nicky Forsyth on +44 (0) 1306 621062 or email nicky.forsyth@marsc.com.