REPORT ON THE REVIEW OF ENUT-94’S SAMPLE AND METHODOLOGY FOR THE CALCULATION OF ITS ADJUSTMENT FACTORS

By

Alfredo Aliaga

Introduction

The present document presents a detailed revision of 1994 National Time-Use Survey (ENUT-94) sample’s design carried out by ONE and INSTRAW.

1. The suggested design is described as a multistage, probabilistic, by conglomerates, stratified and proportional sample.

2. A listing of census segments established by the last National Population and Housing Census (1993) constitutes the sampling framework.

3. This survey covers the totality of the national territory of .

4. Even when it has been stated that this sample is distributed proportionally to its population size, in fact we are dealing with a sample that approximates a proportional distribution. This can be observed in the presented Chart 2. See, in particular, the representation of the National District, which is larger in the sample compared to the census.

5. The sample’s size is about 1500 households. Even when the size of the sample enables a variety of analysis, greater levels or domains of desegregation in order to make an analysis of data with a higher level of confidence, are not clearly established. According to the number of segments selected, it can be stated that these greater levels of desegregation correspond to the total of the country, to the residence area and the nine established strata (with a minimum of 41 excepting stratum “rest”).

6. The segment framework was stratified in nine strata. National District constitutes a separated stratum and the rest of the areas of the country have been stratified in eight strata according to the following indicator: use of land. In total, there are 9 strata in the country.

7. On a first stage, sample’s selection consists of a selection of areas for every stratum. Those areas that are larger than the area average entered the sample with certainty, with a probability of 1. A sample of the rest of areas was done. On a second stage, the supervision areas were selected, and finally the number of segments was selected.

8. In the initial document of sample’s design it was stated that every selected segment should be updated in the field in order to obtain the current composition of households and finally, achieve a selection of two households per segment. We regret that this process was carried out only in a few segments, remaining unknown for which segments this was done, and which were the results.

9. It is evident the lack of information of many sample’s design parameters (such as the probability of area selection, etc.) and of implementation process’ records (apparently a correct classification of each case by rural/urban segment can not be obtained).

10. For every combination of strata and residence area, the ideal would have been to have the following parameters: a. Probability of area selection b. Probability of census area selection c. Probability of census segment selection d. Probability of household selection

11. The theoretic weight based on the sample’s design would have been the inverse of the product of these four probabilities detailed in point 10. When this information was asked in diverse opportunities to Gisela Quiteiro, in order to calculate these parameters, the author could not obtain this information. After waiting and due to the fact that these parameters were not available, it was necessary to adopt an approximate solution.

Apparently, the implementation of the sampling also lacks of necessary documentation, due to the fact that we do not count with explicit detail of the household updating process. This situation became much more complicated when a crossed table of complete (interviewed) cases per segment and residence area was asked in order to establish a better approximation of household selection probability for each segment.

12. The response to all this irregular situation in a study like ENUT-94 in Dominican Republic (supposing that census segments are quite alike in size inside every residence area) was the following: a. For every combination of strata and residence area (urban/rural) the final sample of census segments was considered as a random sample of simple segments of approximately same size. b. From the available documentation it was deduced that the average size of each urban segment was of 43 households and an average of 32 households for every rural segment.

13. Under these premises, for every combination of strata and residence area, the total probability of segment is expressed as the resultant value from the division of the number of segments selected in the sample by the total number of segments in the framework.

Inside the segment, the probability of household selection was 2 divided 43 for each urban segment and 2 divided 32 for every rural segment.

14. Finally, the approximate weight based on the sample’s design used in ENUT-94 was calculated as the inverse of the product of the two last values (12 and 13).

15. These values of approximate weight based on the sample’s design were adjusted by the household level of non-response for every stratum and residence area.

16. With the present document, two documents of Excel format are attached in order to follow the process that has been described in the precedent steps.

CHART 1. 1993 SURVEY'S SEGMENTS FRAMEWORK

Number of census Average of Number of urban Average of Number of Average of STRATA segments segments per segments urban rural rural stratum segments segments segments NATIONAL DISTRICT 11425 26.7 8115 71 3310 29 VEGETABLES AND GROCERIES 2366 5.5 890 37.6 1476 62.4 SUGAR CANE 3336 7.8 2024 60.7 1312 39.3 AND CACAO 4820 11.3 1197 24.8 3623 75.2 LIVESTOCK 3838 9 1237 32.2 2601 67.8 BANANAS AND ALIKE 4324 10.1 1732 40.1 2592 59.9 GRAINS 7163 16.8 2349 32.8 4814 67.2 TOBACCO 3528 8.3 2206 62.5 1322 37.5 REST 1928 4.5 611 31.7 1317 68.3 TOTAL 42728 100 20361 43.7 22367 56.3 CHART2. SAMPLE'S DISTRIBUTION

Number of Average of Average of STRATA segments segments census in the sample in the sample segments NATIONAL DISTRICT 223 30.1 26.7 VEGETABLES AND GROCERIES 41 5.5 5.5 SUGAR CANE 56 7.6 7.8 COFFEE AND CACAO 70 9.5 11.3 LIVESTOCK 70 9.5 9 BANANAS AND ALIKE 80 10.8 10.1 GRAINS 103 13.9 16.8 TOBACCO 64 8.6 8.3 REST 33 4.5 4.5 TOTAL 740 100 100 Chart 3. VALUE OF ADJUSTEMENT FACTORS FOR THE ENUT-94 SAMPLE

Area Weight per Selected Complete Gross weight Expanded Weight STRATA approximate households housholds expansion households design (interviewed) standard NATIONAL DISTRICT Urban 1170.9564 292 207 1651.7839 341919.3 1.379610 Rural 913.1034 111 85 1192.4057 101354.5 0.995926 VEGETABLES AND GROCERIES Urban 1063.0556 34 32 1129.4965 36143.9 0.943383 Rural 944.6400 49 47 984.8374 46287.4 0.822560 SUGAR CANE Urban 1036.0952 83 62 1387.0307 85995.9 1.158482 Rural 749.7143 54 50 809.6914 40484.6 0.676274 COFFEE AND CACAO Urban 887.4310 57 55 919.7013 50583.6 0.768157 Rural 1093.7358 103 98 1149.5387 112654.8 0.960123 LIVESTOCK Urban 1108.1458 45 42 1187.2991 49866.6 0.991661 Rural 1095.1579 74 67 1209.5774 81041.7 1.010268 BANANAS AND ALIKE Urban 1095.2353 63 62 1112.9004 68999.8 0.929521 Rural 1152.0000 69 66 1204.3636 79488.0 1.005914 GRAINS Urban 1442.9571 68 62 1582.5982 98121.1 1.321824 Rural 875.2727 163 158 902.9712 142669.5 0.754183 TOBACCO Urban 2062.1304 45 38 2441.9966 92795.9 2.039615 Rural 528.8000 78 75 549.9520 41246.4 0.459333 REST Urban 827.7500 6 6 827.7500 4966.5 0.691357 Rural 709.3333 58 54 761.8765 41141.3 0.636338 TOTAL 1452 1266 1515760.8 ANNEX 1: STRATA

ESTRATA/AREAS URBAN ZONE RURAL ZONE URBAN ZONE RURAL ZONE ALFREDO SEGMENTS ALFREDO SEGMENTS SEGMENTS SEGMENTS

TOTAL 20361 100 22367 100 378 100 362 100

NATIONAL DISTRICT 8115 39.9 3310 14.8 149 39.4 58 16.0

VEGETABLES AND GROCERIES 890 4.4 1476 6.6 18 4.8 25 6.9 Constanza 78 220 2 4 105 210 2 4 Azua 229 187 5 4 Estebanía 18 35 17 30 Bani 243 342 5 7 Ocoa 102 218 2 4 25 80 Palenque 28 54 1 1 8 36 Fundación (Peravia) 21 24 Matanzas 16 40

SUGAR CANE 2024 9.9 1312 5.9 42 11.1 28 7.7 Sabana Grande de Boya 83 206 2 4 San Pedro de Macorís 696 175 14 4 Ramon Santana 10 77 La Romana 666 95 14 2 Puerto Plata 371 309 8 6 Ing. Consuelo 87 67 2 1 Ingenio Quisqueya 62 19 13 78 18 160 18 126

COFFEE AND CACAO 1197 5.9 3623 16.2 29 7.7 53 14.6 39 183 1 3 Peralta 23 47 25 138 Paraiso 28 52 1 1 Enriquillo 36 65 Polo 11 41 0 1 23 23 44 36 San Jose de las Matas 47 432 1 8 17 83 La Ciénaga 0 64 Hostos 10 35 Poster Río 13 32 Esperalvillo 18 104 José Contreras 6 41 0 1 Los Ríos 19 30 Janico 10 221 0 4 Pedro García 7 55 0 1 Gaspar Hernández 22 199 0 4 18 38 Altamira 24 172 0 3 Restauración 12 50 San Cristobal 409 456 8 9 166 266 3 5 Cambita 37 166 1 3 12 66 0 1 Yamasa 31 305 1 6 49 103 Imbert 41 120 1 2

LIVESTOCK 1237 6.1 2601 11.6 24 6.3 38 10.5 67 260 2 7 Higuey 294 332 8 9 Hato Mayor 126 214 4 6 El Seibo 93 391 3 11 Dajabón 68 87 Partido 11 24 0 1 Montecristi 68 80 9 45 Sosúa 35 143 1 4 22 7 Pedernales 46 37 Oviedo 15 22 0 1 Maimón 46 33 35 61 1 2 Sabana Iglesisa 0 78 El Pino 7 40 0 1 36 26 Otra Banda 15 37 0 1 Don Juan 15 48 0 1 Duvergé 50 35 Mella 13 7 0 0 Las Matas de Farfán 90 214 Pedro Santana 6 75 Bánica 13 67 Río San Juan 38 57 Cabrera 19 181 1 5

BANANAS AND ALIKE 1732 8.5 2592 11.6 34 9.0 36 9.9 17 126 0 3 Barahona 296 52 7 1 Cabral 53 17 Cristóbal 14 21 Galvan 27 62 Tamayo 31 27 1 1 51 69 Tamboril 94 126 2 3 24 61 Moca 240 376 6 9 Cayetano Germosen 12 28 Mao 210 137 5 3 Monción 29 54 Pepillo Salcedo 18 38 0 1 75 242 2 6 El Valle 38 32 La Vega 383 774 9 18 El Peñon 19 57 Fundación (Barahona) 0 48 0 1 Ubilla 11 87 52 10 Pueblo Viejo 0 84 San Victor 38 64 1 2

GRAINS 2349 11.5 4814 21.5 35 9.3 88 24.3 Esperanza 134 133 3 3 San Ignacio de Sabaneta 74 301 1 6 Los Almacigos 19 14 38 88 39 37 1 1 Villa Vasquez 61 38 Guayubin 12 208 0 4 Castañuela 28 76 Luperon 21 122 0 2 San Francisco de Macoris 506 545 10 10 Pimentel 52 65 1 1 Guayabal 16 33 Castillo 35 86 19 86 Cotui 167 282 3 5 40 69 Las Guaranas 52 16 3 6 San Juan 292 425 6 8 Juan de Herrera 25 81 Comendador (Elias Piña) 41 96 El Llano 14 30 0 1 San José de los Llanos 23 230 0 4 26 148 Nisibón 22 68 71 63 1 1 20 63 Bohechío 14 38 0 1 Padre Las Casas 42 168 1 3 27 111 1 2 281 269 5 5 35 32 28 22 32 122 Juan Santiago 9 32 La Mata 34 171 1 3

TOBACCO 2206 10.8 1322 5.9 23 6.1 40 11.0 Tabara Arriba 21 62 Santiago 1984 834 41 17 Villa Bisono (Navarrete) 109 93 21 117 0 2 36 82 Villa Gonzalez 35 134 1 3

REST 611 3.0 1317 5.9 4 1.1 31 8.6 Samana 42 243 1 5 Neyba 77 133 2 3 Sanchez 57 110 23 206 0 4 Jimaní 22 26 288 135 6 3 Nigua 8 132 0 3 20 44 Salcedo 74 236 2 5 La Cueva 20 20 Sabana Buey 10 2