Farm Sample and Sampling Procedure 



A random sampling process was used in each of the geographic 

 areas identified in the previous section. However, the sampling pro- 

 cedure used varied throughout the region. In the New England States, 

 the dairy farm was the sampling unit. This procedure was justified hy 

 the fact that nondairy farms (with potential to shift into dairying) were 

 almost negligible. In the other States the sampling unit was an area of 

 land containing a number of dairy farms. The land segments were 

 drawn randomly, and both dairy and nondairy farms within each of the 

 segments were surveyed. This made it possible to identify "potential" 

 dairy farms. 



The specific sampling procedures used in each of the areas are 

 described in detail in the following subsections. 



New England {Areas 1-8). In New England, a dairy farm was de- 

 fined as "a bundle of farm resources containing ten or more milk cows 

 under the single management of one or more operators." 



Each State made available a list of dairy farms which represented 

 the universe (i.e., all dairy farmers). The farms were ordered by area 

 and each farm in the area was assigned a number. A random sample was 

 drawn in such a manner as to insure a 2-percent sampling rate in each 

 of the areas (Table 3) . 



New York. New Jersey, Northern Pennsylvania (Areas 9, 10, 11, 12, 

 13, 14, 15, and 18). Information required was obtained from a sample 

 drawn by G. J. Conneman for his study of long-run changes in milk 

 production in the New York milk shed.*^ The sampling procedure des- 

 cribed here was developed for Conneman's study and was used for the 

 Northeast regional study. 



The Conneman study defined a farm or producing unit as a "bundle 

 of farm resources — - land, buildings, cattle, and machinery — under the 

 single management or control of one or more operators." The farm 

 universe was defined as all the farms or producing units delivering milk 

 under the New York-New Jersey, Buffalo, or Rochester orders. 



Area segments consisting of groups of producers were the sampling 

 units. Each of these segments contained approximately 10 farms. To 

 achieve a sampling rate of 21/2 percent of the farms, one segment of ten 

 producing units was drawn for every 400 farms. Thus, the number of 

 segments drawn in an area was approximately equal to the number of 

 producing units (established from records of the milk market adminis- 

 trator) divided by 400. All farmers in the segment were interviewed 

 regardless of whether or not they were shipping milk. 



In summarizing the procedure, Conneman comments as follows: 



"This sampling procedure is the equivalent of dividing the entire 

 milk shed into area segments of approximately ten producing units, and 



8 George J. Conneman. An Economic Analysis of Changes in Milk Production in 

 the Neiv York Milk Shed, Progress Report 1, Cornell University, Department of 

 Agricultural Economics, A. E. Res. 135, December 1963. 



13 



