Market Research Sampling Solutions                                                                      T. +44 (0)1306 621062
                                                                                                                                       E.  sales@marsc.com

Sampling Models

Interlocking cells

The most common model used for deriving sample quota is the ‘interlocking cells’ method.  Take the following example: In a sample requiring 50 male, 50 female and 60 single, 40 married people, MARSC will calculate the following quota cells:

This is achieved by apportioning sample across the quota cells based upon the marginal (or high-level) totals specified by the user.

MARSC supports this model of sample definition as standard.

Non-interlocking cells

Several difficulties with complex interlocked models, such as unachieved targets or empty cells, or even knowing what legitimate targets to set (e.g. Males, Divorced, Aged 25-34, Educated to degree level, in the North-west region) can be overcome by using a non-interlocking model. In this model, only the high-level targets are specified. 

MARSC offers two different methodologies depending on the circumstances:

1. Shortage of sample/unachieved targets - the targets will be reworked repeatedly by making minor adjustments to them so that the shortfall is absorbed into neighbouring cells, but without unduly disrupting the overall targets set.
   

2. An abundance of sample in all the cells in the model - an optimisation routine, which follows the Deming Stephan principle commonly used in many rim- or target-weighting schemes, will applied to the low-level estimated cell targets. This adjusts the cell targets within the matrix so as to ensure that the cells with the most abundant supply of sample are favoured over those with a more meagre supply of available sample.

Universe fit

If you maintain a panel with balanced demographics, or are feeding your sampling database with the totality of your customers, you may take the view that your project database is, in effect, the population universe for your research. The ‘universe fit’ method, will replicate the proportions of the demographics in your overall database in the sample you draw.

Single dimension with sub-totals

As an alternative to the other models, targets may be grouped together, to allow for quotas in neighbouring cells within the same group to be redistributed to top-up individual cells where there is a shortage of available sample.

    Back to top...

MARSC Web Portal Features...

MARSC Web Portal Benefits...

Home...

MARSC software 
is produced by
Centurion Marketing
Systems Ltd

email us

T. +44 (0)1306 621062
F. +44 (0)1306 621597