The MOST Awards utilises a random stratified sampling methodology.
A web-based, self-completion method is used to invite people working in media agencies, media departments, selected corporates and media owners to participate in the annual survey.
Since its inception in 2009, the MOST Awards has achieved a sample size of more than 200 respondents every year for the Media Owner and Media Agency survey respectively. Against a population with low variance (like practitioners in the SA Media Industry), the margin of error for a sample of 200 respondents yields a margin of error of less than 4 percentage points at the 95% level of confidence.
In terms of the sampling & weighting method, media agencies are segmented according to their published billing (in AdReview). According to Adex, the annual media billing in SA is approximately R19 billion. Media agencies are segmented into quartiles according to billing. So, for example the Mediashop and OMD fall into the 1st quartile (their combined billing is just short of R7 billion).
Then a sample base is drawn in proportion with the % billing of the total SA billing per quartile, as follows:
% of total billing No of companies Sample %
Quartile 1 36% 2 35%
Quartile 2 46% 5 45%
Quartile 3 13% 4 15%
Quartile 4 5% +-30 5%
As far as the Media Agency leaders goes, an equal weighting is applied. So, each respondent receives the same weighting irrespective of size of Agency or Agency billing, the reason being that each respondent has the similar task of running a media agency or a media department.
With the media decision-makers in Corporates, the people invited to participate in the survey need to have frequent and meaningful contact with media owner in order to qualify.
The sample is monitored and managed on a real-time basis to ensure that the correct proportion of respondents from each media agency is collected. Specifically, if FGI sees that the sample is skewing disproportionately to the sample design, it engages with Media Agency leaders directly to encourage participation in order to normalize the “representativeness” of the sample.
As a final quality and validity control measure, every respondent to the MOST Awards survey is reviewed and validated.