Education and Labor
Sampling Plan
The Monthly Employment Survey - PME - is conducted with the use of independent samples for each Metropolitan Area covered by means of probability selection design in two stages: census tracts and housing units (private housing units and collective housing units).
In the selection of census tracts for PME samples, we adopt the sector mesh of the latest Population Census. The units were selected, from each municipality of the metropolitan area covered, with proportional probability to the number of housing units counted in the 1991 Population Census.
At a second stage, we selected, with equal probability, from every census tract in the sample, private and collective housing units in order to investigate characteristics of the persons living in them.
Nowadays, with the aim of keeping the basic list of housing units updated and, as a consequence, preserving the pre-fixed fractions, all the sectors in the sample go through an operation of information (or list) update. It is an update of the list of residential and non-residential units existing in the selected areas in order to form the samples.
Besides this update, the demographic increase of the municipalities covered is monitored according to a list formed by sets of 30 or more adjacent housing units, all of which came into existence after the end of the last Population Census.
The supplementary survey Education and Labor used the PME sample of April 1996, and was directed to the residents who were 20 years of age and over, at the time of the interview.
The end of this chapter presents, for each Metropolitan Area covered, the sampling fraction, the number of census tracts selected, the number of housing units surveyed, and the number of persons interviewed in the PME of April and in the supplementary survey as well.
The expansion of the sample uses ratio estimators whose independent variable is the projection of the resident population of the metropolitan area. These projections take into consideration population's evolution between the Population Censuses of 1980 and 1981, with growth hypothetically associated with fertility, mortality and migration rates. Below are some considerations which make it possible to evaluate the level of confidentiality of estimates, and may provide more elements to aid in the interpretation of results in this volume.
Considering the expansion process adopted by PME, it is necessary to highlight that precision is strongly related to the hypotheses made for fertility, mortality and migration rates. The calculation of the sampling error should, therefore, take into consideration two sources of variation:
• the sampling error which comes from the selection of housing units for the sample; and
• the error which comes from the mathematical model employed to project the population.
The results presented refer only to sampling errors. The difficulty that may exist in the calculation of sampling errors, expressed by means of variation coefficients for all the tabulation cells in this volume, calls attention to the need of an alternative way to present these coefficients.
In order to provide an approximation for the variation coefficients associated to the estimates and quantify the sampling error as a function of the size of the estimate, regression models were adjusted for each of the metropolitan areas covered and also for the group of six metropolitan areas.
The model used was the type Y = Axb, where x is the value of estimate and Y is the respective coefficient of variation. The regression coefficients A and B, found in each adjustment, are presented in table 1.
To evaluate, approximately, the variation coefficient associated to an estimate, one must apply convenient parameters (A and B) to the expression Axb.
The coefficients of variation, calculated by means of proper parameters to certain sizes of estimates, are presented in Table 2.
Sampling Fraction and Sample Composition by Metropolitan Areas Covered
| Metropolitan Area | Sampling Fraction | Sample Composition | ||||
|---|---|---|---|---|---|---|
| Persons | ||||||
| Sectors | Housing Units | Total | 18 years and over | 20 years and over | ||
| Recife | 1 / 170 | 196 | 4755 | 14468 | 8867 | 8213 |
| Salvador | 1 / 170 | 169 | 4805 | 15367 | 8952 | 8252 |
| Belo Horizonte | 1 / 170 | 244 | 6563 | 20880 | 13185 | 12334 |
| Rio de Janeiro | 1 / 430 | 315 | 7723 | 19957 | 13548 | 12851 |
| São Paulo | 1 / 600 | 332 | 8177 | 23453 | 14771 | 13926 |
| Porto Alegre | 1 / 170 | 254 | 6513 | 16371 | 10447 | 9913 |
| Total | - | 1510 | 38536 | 110496 | 69770 | 65489 |
Table 1
| Regression Coefficients | |
|---|---|
| A | B |
| 1741.8005 | -0.4678 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 68.8 |
| 2,000 | 49.7 |
| 3,000 | 41.1 |
| 4,000 | 36.0 |
| 5,000 | 32.4 |
| 10,000 | 23.4 |
| 20,000 | 16.9 |
| 30,000 | 14.0 |
| 40,000 | 12.2 |
| 50,000 | 11.0 |
| 100,000 | 8.0 |
| 200,000 | 5.8 |
| 300,000 | 4.8 |
| 400,000 | 4.2 |
| 500,000 | 3.8 |
| 1,000,000 | 2.7 |
| 2,000,000 | 2.0 |
| 3,000,000 | 1.6 |
| 4,000,000 | 1.4 |
| 5,000,000 | 1.3 |
| 10,000,000 | 0.9 |
| 20,000,000 | 0.7 |
| 30,000,000 | 0.6 |
Table 1 - Metropolitan Area of Recife
| Regression Coefficients | |
|---|---|
| A | B |
| 1287.9141 | -0.4725 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 49.2 |
| 2,000 | 35.5 |
| 3,000 | 29.3 |
| 4,000 | 25.6 |
| 5,000 | 23.0 |
| 10,000 | 16.6 |
| 20,000 | 12.0 |
| 30,000 | 9.9 |
| 40,000 | 8.6 |
| 50,000 | 7.8 |
| 100,000 | 5.6 |
| 200,000 | 4.0 |
| 300,000 | 3.3 |
| 400,000 | 2.9 |
| 500,000 | 2.6 |
| 1,000,000 | 1.9 |
| 2,000,000 | 1.4 |
Table 1 - Metropolitan Area of Salvador
| Regression Coefficients | |
|---|---|
| A | B |
| 1126.3287 | -0.4619 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 46.4 |
| 2,000 | 33.7 |
| 3,000 | 27.9 |
| 4,000 | 24.4 |
| 5,000 | 22.0 |
| 10,000 | 16.0 |
| 20,000 | 11.6 |
| 30,000 | 9.6 |
| 40,000 | 8.4 |
| 50,000 | 7.6 |
| 100,000 | 5.5 |
| 200,000 | 4.0 |
| 300,000 | 3.3 |
| 400,000 | 2.9 |
| 500,000 | 2.6 |
| 1,000,000 | 1.9 |
| 2,000,000 | 1.4 |
Table 1 - Metropolitan Area of Belo Horizonte
| Regression Coefficients | |
|---|---|
| A | B |
| 1185.6871 | -0.4717 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 45.6 |
| 2,000 | 32.9 |
| 3,000 | 27.1 |
| 4,000 | 23.7 |
| 5,000 | 21.3 |
| 10,000 | 15.4 |
| 20,000 | 11.1 |
| 30,000 | 9.2 |
| 40,000 | 8.0 |
| 50,000 | 7.2 |
| 100,000 | 5.2 |
| 200,000 | 3.7 |
| 300,000 | 3.1 |
| 400,000 | 2.7 |
| 500,000 | 2.4 |
| 1.000,000 | 1.8 |
| 2000,000 | 1.3 |
| 3,000,000 | 1.0 |
Table 1 - Metropolitan Area of Rio de Janeiro
| Regression Coefficients | |
|---|---|
| A | B |
| 1960.7375 | -0.4735 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 74.5 |
| 2,000 | 53.6 |
| 3,000 | 44.3 |
| 4,000 | 38.6 |
| 5,000 | 34.7 |
| 10,000 | 25.0 |
| 20,000 | 18.0 |
| 30,000 | 14.9 |
| 40,000 | 13.0 |
| 50,000 | 11.7 |
| 100,000 | 8.4 |
| 200,000 | 6.1 |
| 300,000 | 5.0 |
| 400,000 | 4.4 |
| 500,000 | 3.9 |
| 1,000,000 | 2.8 |
| 2,000,000 | 2.0 |
| 3,000,000 | 1.7 |
| 4,000,000 | 1.5 |
| 5,000,000 | 1.3 |
| 10,000,000 | 1.0 |
Table 1 - Metropolitan Area of São Paulo
| Regression Coefficients | |
|---|---|
| A | B |
| 2217.9205 | -0.4685 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 87.2 |
| 2,000 | 63.0 |
| 3,000 | 52.1 |
| 4,000 | 45.5 |
| 5,000 | 41.0 |
| 10,000 | 29.6 |
| 20,000 | 21.4 |
| 30,000 | 17.7 |
| 40,000 | 15.5 |
| 50,000 | 13.9 |
| 100,000 | 10.1 |
| 200,000 | 7.3 |
| 300,000 | 6.0 |
| 400,000 | 5.3 |
| 500,000 | 4.7 |
| 1,000,000 | 3.4 |
| 2,000,000 | 2.5 |
| 3,000,000 | 2.0 |
| 4,000,000 | 1.8 |
| 5,000,000 | 1.6 |
| 10,000,000 | 1.2 |
| 20,000,000 | 0.8 |
Table 1 - Metropolitan Area of Porto Alegre
| Regression Coefficients | |
|---|---|
| A | B |
1380.5797 |
-0.4848 |
Table 2 - Coefficients of variation, by size of the estimate
| Size of the Estimate | Coefficients of Variation (%) |
|---|---|
| 1,000 | 48.5 |
| 2,000 | 34.6 |
| 3,000 | 28.5 |
| 4,000 | 24.8 |
| 5,000 | 22.2 |
| 10,000 | 15.9 |
| 20,000 | 11.3 |
| 30,000 | 9.3 |
| 40,000 | 8.1 |
| 50,000 | 7.3 |
| 100,000 | 5.2 |
| 200,000 | 3.7 |
| 300,000 | 3.1 |
| 400,000 | 2.7 |
| 500,000 | 2.4 |
| 1,000,000 | 1.7 |
| 2,000,000 | 1.2 |
| 3,000,000 | 1.0 |