It must be pointed out that non-resident coding of chemical -biological test Information by the 

 CBCC Biology Code can not be strictly compared to the usual non-resident abstracting and indexing 

 which does not involve coding. This is not to say that a simple coding program involving only 

 organisms' identities, or only specific effects, or only specific chemicals might not be comparable to 

 ordinary indexing. However, the extraction and coding of all aspects of a chemical-biological test is 

 another and more complex matter. 



Communication between coder, checker, arbitrator, and the Center was without a doubt, one 

 of the more trying aspects of the Center's procedures and if any one factor would influence a decision 

 about the residence of coders, this would probably be the most convincing. Errors in interpretation of 

 literature data, in coding, and in interpretation of the Key to the Code could only be avoided by advis- 

 ing the coder of errors he had already made; an error could only be discovered by the checker and 

 communicated to the Center. Repeated efforts to establish a system of "comment sheets" for this 

 communication was never very successful, partly because of the unavoidable gaps of time between the 

 material leaving the coder's hands and his receipt of comments about his errors. Furthermore, once 

 the coder or checker received the notification of his errors, he never had the coded Sheets on which 

 the errors occurred so that frequently the comments, unless they were extraordinarily thorough explan- 

 ations, conveyed little. As a result, when it was finally discovered that a coder was consistently 

 making a coding error, a special letter had most frequently to be prepared to point out and discuss the 

 matter. 



Remuneration for coders on a regular basis, considering its problems, was organized to a 

 satisfactory level of efficiency, but never was free of flaws, partly because of the complex system 

 which demanded that payment not be made for code lines which the checker or Center had deleted as 

 unjustifiable or erroneous. 



A problem to be considered with resident coders is that of efficiency in an unbroken routine of 

 coding. The staff members' own experience has been that it is not possible to restrict professional 

 activity to coding without eventually arriving at a point of ennui and consequent inefficiency. 



Thus, although the CBCC never arrived at a totally satisfactory answer to the matter of coding 

 personnel, it would be strongly recommended that, for thorough and complex coding comparable to 

 that done by the CBCC, an arrangement using resident coders whose routine can be varied might be 

 far preferable to attempting establishment of a system for a staff of non-resident coders. 



Checking of Abstracting and Coding : 



It seems impossible to overemphasize the importance of coding information correctly and 

 according to a standardized procedure. Errors in coding and punching can only mean information that 

 is lost to retrieval or retrieval of unwanted information. It must be recognized that this process of 

 information "in-put" is dependent on human intelligence and subject to human error; the success of 

 the result of retrieval and correlation depends on holding that error to a minimum. To assure that 

 information entering the files was coded (indexed) accurately , the CBCC found no alternative to having 

 the coding checked and arbitrated. (See the descriptions of these procedures in Appendix A. ) 



In establishing any new project in which information is coded, the question of checking the 

 accuracy of coders' "translations" into code is apt to be weighed carefully. Checking has been 

 viewed variously by persons contemplating such a program; frequently, the attitude is expressed that 

 checking is desirable but impractical, or even that it is superfluous. Others concede checking to be 

 essential. It would be unreasonable to attempt to formulate, for all prospective coding projects, a 

 fixed recommendation about the extent of checking, since so many variables are involved. Neverthe- 

 less, the CBCC experience would discourage any idea that checking in some form and for some period 

 of time can be avoided, except perhaps in building coded information collections in which the material 

 coded has very little variation and the coding is extremely brief and all done by the same coder or a 

 continuous coding staff. 



In spite of the care exercised in coding and checking, the Center not infrequently discovered 

 major coding errors on Code Sheets, even after the coding had been punched and filed. Nevertheless, 

 the CBCC felt it had arrived at the point of diminishing returns in the process of checking the coded 

 data entering its files. A certain small number of coding errors had to be accepted as unavoidable 

 within its coding system, just as some factor of error must be established for any suchsystem. 



219 - 



