|

 
|
| | Name : | Doug Jones | Organization : | N/A | Post Date : | 9/27/2005 |
| Glossary Term : | Data Accuracy | Definition : | | Comment : | In response to your request for comments on the draft glossary of voting system terminology, I offer the following, picking up with the letter D. As before, in each case, I am quoting the original and then offering an alternative definition.
Data Accuracy: (1) Data accuracy is defined in terms of ballot position error rate. This rate applies to the voting functions and supporting equipment that capture, record, store, consolidate and report the specific selections, and absence of selections, made by the voter for each ballot position. (2) The system's ability to process voting data absent internal errors generated by the system. It is distinguished from data integrity, which encompasses errors introduced by an outside source.
-- Comment: There is a serious problem here. The general concept of data accuracy applies to all cases where data is read by the voting system from some external source. Some of this data is ballot data, but accuracy questions arise generally. For example, because the voting system standards have traditionally focused tightly on ballot data accuracy, many voting systems have reasonable interfaces for soliciting this data from voters, but very bad interfaces for interacting with the pollworkers who do such things as selecting the ballot style for the next voter. Therefore, I propose this general definition:
Data Accuracy: The accuracy with which data is processed or represented within a system and in transfers between systems. Accuracy can be degraded at system interfaces, for example, by the introduction of input-output errors, it can be degraded during storage, by transient or permanent failure of storage devices, and it can be degraded during processing, by transient errors or the use of inappropriate algorithms. For purposes where imprecise data representations such as floating-point numbers are not used, data accuracy can be numerically measured as the inverse of the error rate.
| |
|
|