Socio-Economic Levels:
Pan-Regional vs. National Definitions
Elsewhere on Zona Latina, we have presented a paper on the definition of Socio-Economic Level used in the Los Medios y Mercados de Latinoamérica study. The fact of the matter is that different countries have different definitions of socio-economic level, as is only proper, which reflect their own historical trajectories and circumstances. For our pan-regional study, we came up with a common definition that we apply to every country in order to have comparable results.
The question arises as to how close is our pan-regional definition to the national ones. This is an important question, because if the definitions diverge significantly, then this may have serious implications for media planning.
There are a number of ways to address this issue. First of all, we can argue on the basis of face validity. When one examines our pan-regional definition as well as the national ones, it should be self-evident that these measures should all reflect some latent (that is, unobservable) socio-economic continuum. Of course, face validity is not scientific proof per se.
A better way is to compare the outcomes of different definitions for the same group of people. In the recently completed 1996 edition of Los Medios y Mercados de Latinoamérica, our field agency GAP (Grupo de Assessoria e Pesquisa) coded the Brazilian ANEP definition of socio-economic level.
Here is the cross-classification of the two socio-economic definitions for the 1700 Brazilian respondents:
% of Respondents by ANEP and LMML Socio-Economic Levels
LMML Socio-Economic Level
| ANEP Socio-economic Level | Level A |
Level B |
Level C |
Level D |
TOTAL |
| Level A | 5% |
1% |
0% |
0% |
6% |
| Level B | 5% |
11% |
3% |
0% |
19% |
| Level C | 0% |
8% |
21% |
6% |
36% |
| Level D | 0% |
0% |
7% |
27% |
35% |
| Level E | 0% |
0% |
0% |
5% |
5% |
| TOTAL | 10% |
20% |
31% |
39% |
100% |
(Source: LOS MEDIOS Y MERCADOS DE LATINOAMERICA 1996)
If we collapse Levels D and E in Brazil, then we find that 5% + 11% + 21% + 27% + 5% = 69% of all cases fall along the diagonal. Here are some standard statistics: Pearson’s R is 0.810, Spearman correlation 0.809, and Gamma 0.930. We should not that the circumstances will not permit perfect agreement, as the marginal distributions are different by choice. That is, the ANEP definition has 6% A’s while the LMML definition has 10% A’s; this means that at least 4% of the cases will have disagree on socio-economic level.
In all, we have very strong evidence that these two definitions are correlated with each other. This is hardly surprising, as we would expect reasonable definitions for the same thing should be reasonably consistent with each other. But it is nice to have good empirical evidence.
REFERENCE
(posted by Roland Soong on 3/17/97)
(Return to Zona Latina's Home Page)