Beef cattle, except feed/ots (0212)— Production or feeding 

 of beef cattle, except feedlots. 



Dairy farms (024)— Production of cows' milk and other 

 dairy products and raising of dairy heifer replacements. 



Poultry and eggs (025)— Chickens, chicken eggs, turkeys, 

 duck, geese, pheasants, pigeons, and quail. 



Animal specialties ('027,/- Fur-bearing animals, rabbits, 

 horses, ponies, bees, fish in captivity except fish hatcheries, 

 worms, and laboratory animals. 



General farms, primarily livestock (029) — Lwjestock and 

 livestock products, including animal specialties, but less 

 than 50 percent of sales from any single 3digit industry 

 group. Includes farms with no agricultural products sold 

 reporting livestock or pasture. 



STATISTICAL ADJUSTMENTS 



The 1982 Census of Agriculture used two types of statis- 

 tical adjustments— nonresponse and sample. These adjustments 

 were necessary because, as noted previously, not all farms were 

 surveyed for all items and some farms failed to respond to the 

 numerous attempts to contact them. 



ISIonresponse 



Farms for which no response was obtained from the mailout/ 

 mailback self-enumeration procedure were classified as "large" 

 or "small" based on expected sales being above or below 

 $100,000. The small nonrespondents were classified even further 

 using other mail list characteristics. 



A short report form was then sent to a sample selected from 

 the various classifications of small farms to obtain information 

 needed to develop a State estimate of the number of nonre- 

 spondents that were in fact farms. If this second report form was 

 not returned, then a telephone follow-up was attempted. 



For all large farms, a telephone follow-up was conducted. 

 Some operators could not be contacted or they refused 

 to cooperate. In these cases, secondary sources such as 

 ASCS county offices, extension agents, and previous census 

 reports were utilized whenever possible to provide reports for 

 tabulation. 



State estimates of the percentage of nonrespondent farms 

 were made for several strata. These estimates were then used to 

 make estimates at the county level for each stratum. Respondent 

 farms were then randomly selected within county stratum to 

 represent the nonrespondent farms by having a nonresponse 

 weight of two. All other respondent farms had a nonresponse 

 weight of one. 



The nonresponse adjustment procedure makes the assump- 

 tion that the respondents and the nonrespondents have similar 

 statistical properties. Also, the use of secondary sources for 

 some large farms and the fact that a percentage of the sampled 

 farms were not contacted by either mail or telephone is assumed 

 not to affect the estimates for nonrespondents. 



Table A partially measures the effect of nonresponse on 

 selected major items at the State level. This estimate of the non- 

 response rate measures only the effect of small whole farm 

 nonresponse. The effect of the adjustment for nonresponse 

 due to complete nonresponse among large farms and item 



nonresponse among all farms is not included in table A. The 

 percentages in table A indicate the potential for bias in pub- 

 lished figures resulting from small farm nonresponse. However, 

 large farm nonresponse does not necessarily indicate bias. 

 Individual county nonresponse rates may differ substantially 

 from the State value in table A. 



Sample 



Report forms containing additional sample inquiries (sec- 

 tions 22 through 28) were mailed to all farms in Alaska and 

 Hawaii and to a sample of farms identified when the mail list 

 was constructed. Other farms were selected into the sample 

 with certainty because they were expected to be "large," to 

 have a complex organizational structure, or to have special 

 characteristics. The definition of large varied by State, from a 

 low of 1,000 acres or $100,000 sales to a high of 10,000 acres 

 or $500,000 sales. The median State level for the continental 

 United States was 1,800 for acres and $150,000 for sales. 

 Large farms identified during processing also were included with 

 certainty by sending the respondents a supplemental form con- 

 taining only sample questions. In addition to large farms, all 

 farms in counties containing less than 100 farms in 1978 and 

 all farms in SIC groups with a small number of farms at the 

 State level in 1978 were included in the sample with certainty. 

 Farms in counties containing 100 to 19Sr farms in 1978 were 

 sampled at a rate of 1 in 2. Farms in counties containing 200 or 

 more farms in 1978 were sampled at a rate of 1 in 6. 



The noncertainty samples were stratified into 128 strata 

 formed by the 8 sales groups, 8 SIC groups, and 2 acre groups. 



Sales 



SIC 



Acres 



$1 to $999 

 $1,000 to $2,499 

 $2,500 to $4,999 

 $5,000 to $9,999 

 $10,000 to $19,999 

 $20,000 to $39,999 

 $40,000 to $99,999 

 $100,000 or more 



Oil 0to69 



013 70 or more 



016,018 



017,019 



024, 021, except 0212 



0212,027 



025 



029 



This post -stratification was done to improve precision and to 

 reduce the bias in all estimates by weighting the sample farms 

 within stratum by the ratio of total farms to sample farms. If 

 the calculated weight was greater than 12 in counties sampled 

 at a rate of 1 in 6, or greater than 4 in counties sampled at a 

 rate of 1 in 2, then the post -stratum was collapsed into another 

 post-stratum. Collapsing of post-stratum also occurred if the 

 stratum contained fewer than 10 sample farms. 



Table A. Percent of State Total Represented by 



Adjustment of Whole Farm Nonresponse: 1982 



1982 CENSUS OF AGRICULTURE 



APPENDIX A A-9 



