records were not subject to nonresponse 
weighting because they were subjected to 
complete nonresponse follow-up. If the data 
were not captured from the nonrespondents, the 
records were then manually estimated using data 
from respondents. 
Nonresponse weighting was monitored in an 
effort to keep nonresponse weights from 
exceeding 2.0; if an RHG’s nonresponse weight 
exceeded this upper limit, a collapsing algorithm 
was applied to combine neighboring RHGs 
within a municipio, and recomputed the weights 
for the combined groups. For published 
tabulations of complete count items, the 
noninteger nonresponse weight was rounded to 
an integer weight of 1, 2, or 3. A random 
rounding algorithm based on cumulated 
fractional parts of the noninteger nonresponse 
weight values was used to produce an unbiased 
result at island level. 
COVERAGE ADJUSTMENT 
Due to the dynamic nature of mail lists, some 
farm operators may not be included in the CML. 
To account for this undercoverage, an area frame 
consisting of the entire island of Puerto Rico was 
sampled. For sampling, NASS stratified the 
Puerto Rico area frame on the basis of 
agricultural intensity, with strata consisting of: 1) 
land area with dense agriculture; 2) land area 
with sparse agriculture and few houses; 3) land 
area with sparse agriculture and many houses; 4) 
cities with no apparent agricultural activity; and 
5) ostensibly nonagricultural land such as parks 
and military reservations. Primary Sampling 
Units (PSUs) were created based on specific size 
requirements and permanent boundaries. An 
additional sampling enhancement involved the 
grouping of municipios with similar agriculture 
into nine clusters. Within each stratum and 
cluster, a random sample of PSUs was selected 
and then further subdivided into target sampling 
units called segments. Of approximately 7,500 
segments available for sampling, 300 segments 
were selected. Aerial photography and maps for 
the 300 segments were provided to support field 
data collection. All NML farms discovered 
within the 300 sampled segments were included 
obtener los datos del agricultor, la informacion para la 
finca fue estimada por el enumerador basada en 
observacion y datos administrativos. 
La ponderacion aplicada fue monitoreada para el fin de 
no excediera un valor mayor de 2.0; si una ponderacion 
de RHG excedia el limite superior, se le aplico un 
algoritmo para combinar RHGs en la vecindad dentro 
del municipio, y se computo la ponderacion para el 
grupo combinado. Para propositos de la tabulacion, en 
cada record, la ponderacion de no respuesta para 
fracciones de finca se redondeo a una ponderacion de 1 , 
2 6 3. Para garantizar un resultado sin sesgo, a nivel de 
Isla, utilizamos un algoritmo de redondeo aleatorio para 
calcular los remanentes acumulativos. 
AJUSTE DE COBERTURA 
Se presume que una lista de nombres y direcciones 
nunca esta completa. Por lo tanto; y para ampliar la 
cobertura, se utilize un marco muestral de area 
representativo de toda la Isla de Puerto Rico. El marco 
muestral se estratifico por el NASS de acuerdo con la 
actividad agricola. Los estratos se clasificaron como: 1) 
areas con mucha agricultura, 2) agricultura dispersa y 
pocas viviendas, 3) agricultura dispersa y muchas 
viviendas, 4) ciudades con cero agricultura, y 5) areas 
con cero agricultura tales como parques e instalaciones 
militares. Las unidades muestrales primarias, se crearon 
utilizando criterios de tamano y limites permanentes. 
Ademas, se crearon nueve grupos de municipios con 
caracteristicas agricolas similares. Dentro de cada 
estrato y cada grupo, se selecciono una muestra aleatoria 
de unidades primarias que luego se subdividio en 
unidades de muestreo de enfoque llamadas segmentos. 
Se seleccionaron 300 segmentos de alrededor de 7,500 
segmentos disponibles. A los enumeradores se les 
proveyo una serie de fotografias aereas y mapas de los 
300 segmentos para el trabajo de campo. Toda operacion 
NML que se encontro dentro de los 300 segmentos y que 
no estaba en la lista fue designada como operacion de la 
muestra. 
A - 2 APPENDIX A/APENDICE A 2012 Census of Agriculture 
USDA, National Agricultural Statistics Service 
