enumerator introduce error into the census data. Sucli 

 incorrect information can lead, in some cases, to incorrect 

 enumeration of farms. Tliis type of reporting error is 

 measured by the Classification Error Study discussed later 

 in this section. To reduce all types of reporting error, 

 questions were phrased as clearly as possible based on 

 tests of the census report form, and detailed instructions 

 for completing the report form were provided to each 

 addressee. In addition, each respondent's answers were 

 checked for completeness and consistency. 



Item Nonresponse 



Nonresponse to particular questions on the census 

 report that we would logically or statistically expect to be 

 present may create a type of nonsampling error in both 

 complete count and sample data. When information reported 

 for another farm with similar characteristics is used to edit 

 or impute for item nonresponse, the data may be biased 

 because the characteristics of the nonrespondents have 

 not been observed and may differ from those reported by 

 respondents. Any attempt to correct the data for nonre- 

 sponse may not completely reflect this difference either at 

 the element level (individual farm operation) or on the 

 average. 



Processing Error 



The many steps of processing of each census report 

 form are sources for the introduction of nonsampling error. 

 The processing of the census report forms includes cleri- 

 cal screening for farm activity, computerized check-in of 

 report forms and followup of nonrespondents, keying and 

 transmittal of completed report forms, computerized edit- 

 ing of inconsistent and missing data, review and correction 

 of individual records referred from the computer edit, 

 review and correction of tabulated data, and electronic 

 data processing. These operations undergo a number of 

 quality control checks to ensure as accurate an application 

 as possible, yet some errors are not detected and cor- 

 rected. 



Classification Error 



An evaluation study of classification errors was con- 

 ducted in the 1987 Census of Agriculture as part of the 

 census coverage evaluation program. A sample of mail list 

 respondents was selected, and these addresses reenu- 

 merated to determine whether they were a farm or non- 

 farm. A farm status determination was made based on the 

 evaluation questionnaire and compared with the status 

 based on the data reported on the census form. Differ- 

 ences in status were reconciled. 



In past censuses, the proportion of farms undercounted 

 due to classification errors was higher for farms with small 

 values of sales. The classification error rate was higher for 

 (1) livestock farms than crop farms, (2) farms with a small 



number of acres than larger farms, or (3) tenant farms than 

 full or part-owner farms. Results from the 1987 classifica- 

 tion error study will be published in the Coverage Evalua- 

 tion report. 



EDITING DATA AND IMPUTATION FOR ITEM 

 NONRESPONSE 



For the 1987 Census of Agriculture, as in previous 

 censuses, all reported data were keyed and then edited by 

 computer. The edits were used to determine whether the 

 reports met the minimum criteria to be counted as farms in 

 the census. Computer edits also performed a series of 

 complex, logical checks of consistency and completeness 

 of item responses. They provided the basis for deciding to 

 accept, impute (supply), delete, or alter the reported value 

 for each data record item. 



Whenever possible, edit imputations, deletions, and 

 changes were based on component or related data on the 

 respondent's report form. For some items, such as oper- 

 ator characteristics, data from the previous census were 

 used when available. Values for other missing or unaccept- 

 able reported data items were calculated based on reported 

 quantities and known price parameters. 



When these and similar methods were not available and 

 values had to be supplied, the imputation process used 

 information reported for another farm operation in a geo- 

 graphically adjacent area with characteristics similar to 

 those of the farm operation with incomplete data. For 

 example, a farm operation that reported acres of corn 

 harvested, but did not report quantity of corn harvested, 

 was assigned the same bushels of corn per acre harvested 

 as that of the last nearby farm with similar characteristics 

 that reported acceptable yields during that particular exe- 

 cution of the computer edit. The imputation for missing 

 items in each section of the report form was conducted 

 separately; thus, assigned values for one operation could 

 come from more than one respondent. 



Prior to the imputation operation, a set of default values 

 and relationships were assigned to the possible imputation 

 variables. The relationships and values varied depending 

 on the item being imputed. For example, different default 

 values were assigned for several standard industrial clas- 

 sification and total value of sales categories when imputing 

 hired farm labor expenses. These values and item relation- 

 ships for the possible imputation variables were stored in 

 the computer in a series of matrices. The computer 

 records were sorted by reported State and county, where 

 the county sequence was based on similar types of farms 

 and agricultural practices. 



Each execution of the computer edit consisted of records 

 from only one State. For a given execution of the edit, the 

 stored entries in the various matrices were retained in the 

 computer only until a succeeding record having acceptable 

 characteristics for some sections of the report form was 

 processed by the computer. Then the acceptable responses 



C-6 APPENDIX C 



1987 CENSUS OF AGRICULTURE 



