86 MONTANA AGRICULTURAL STATISTICS 



STATISTICAL METHODS 



"Montana Agricultural Statistics" represents an accumulation of agricultural information from many 

 sources. Each source is evaluated for validit\- and in all cases, the data must pass the test of statistical reliability. 



The most frequently used source is the agricultural producer himself. In order to build a sound foundation 

 for data, it is necessary to find the primary element. Once the primary element has been discovered and defined, 

 then a major decision has to be made. That decision is this: can we count each element in the universe to discover 

 the total or the average or must we devise a sample to represent the universe? In most areas of interest, the Crop 

 Reporting Service finds the cost of 100 percent coverage far beyond the available resources. Therefore, except in 

 a very small universe, for example chern.- packers, sampling is used. 



In the early days of American agriculture, each farm was likely to have many types of agricultural 

 commodities — cattle, hogs, sheep, chickens, small grains, corn, fruit, etc. One could send a single question- 

 naire to a cross-section of farms and obtain a good picture of agricultural production and prices for all items. But 

 that has all changed in this day of specialization. It is now necessary to build lists for each type of agricultural 

 commodity. Each questionnaire must be designed to obtain information for one commodity or group of 

 commodities. 



.Another development that must be considered in sample design is the influence of extremely large 

 operations. When a sample is selected we must be sure that the inclusion or e.xclusion of one or a few large 

 operations will not affect the sample results. The best way to prevent this adverse influence is to select the large 

 operations with 100 percent certainty. This is standard procedure for all major surveys. An extension of this 

 sampling technique ( stratified sampling t requires giving the largest operations a 100 percent chance of selection 

 and each group of smaller operations a lesser chance, so that the smallest strata may have only a. 5 or 10 percent 

 chance for selection. 



Sampling from a list is one method of drawing a sample. Another method used in the Montana Crop and 

 Livestock Reporting Service is area frame sampling. This makes use of the geographic land area as the universe. 

 Aerial photos and maps are used to identify and delineate small areas of land. These segments of land are 

 surveyed by a local enumerator contacting the land operator and obtaining information for all the land area 

 inside the segment. 



Crop acreage estimates are determined through the useof t)oth list sampling and area sampling. Crop yield 

 estimates are also generated by two sampling methods — list and area frame. Within the area frame we use two 

 types of data collection. One is by asking growers how many acres were harvested and how many bushels were 

 obtained. The other method is called objective yield measurements. In the latter method, plant, head and kernel 

 counts are taken from designated plots. Moisture and weight measures are taken and a formula is used to 

 compute yield. 



For cattle and calf inventory we are now using a combination of list sampling and area frame sampling. 

 This is a highly efficient probability sample design and provides extremely accurate estimates (it should be 

 noted that this survey design is one of the most recent statistical developments in use at this time i . All of these 

 sampling methods are aimed at obtaining accurate estimates at the State level. 



County level estimates require some redirection of our samples. For example, a larger total sample is 

 needed in order to provide sufficient reports in each county. Also, the distribution must be based more on 

 geography than on size of operation. Therefore, in order to have a well rounded statistical procedure for both 

 State and County estimates, text book sample designs must be revised and compromises obtained. This results in 

 many types of survey approaches; each commodity- or group of commodities must have a "tailor made" survey 

 design. 



In addition to sample surveys, we make extensive use of other reliable data sources — especially for 

 allocating State estimates by counties. We rely heavily on the .Agricultural Census County data available e%'ery 5 

 years. We also use published county data from ASCS. assessors. BLM. State Apiary. State Horticulture, and 

 others. Personal contacts with county extension agents and other informed individuals are an integral part of the 

 Montana Crop and Livestock Reporting Ser\-ice data collection operations. 



Professional statisticians in the Helena office use all available statistical tools and equipment including 

 the latest electronic data processing and communication devices to assure accurate data handling. 



Survey design is under their continuous sur\'eillance for proper application and execution. The Helena 

 office has full access to a statistical methods staff and a research and development staff located in the 

 Washington. D.C. office of the Statistical Reporting Service. State estimates for all commodities are reviewed 

 and approved by the Crop Reporting Board in Washington, D.C. before release at a specified time and date. 



The Montana Crop and Livestock Reporting Service is under the general direction of the USD.A Statistical 

 Reporting Ser\-ice and the .Montana Department of .Agriculture. The office enjoys the full support and confidence 

 of both agencies, whose cooperation goes back to 1945. 



The Service has two goals — one is to provide accurate, reliable agricultural information and the other is to 

 move that information into the hands of the user as fast as possible. 



