Before you can correct your errors, you need to know what kind of errors you have. You have two methods of finding these errors: manually or automatically. Manually checking large quantities of data is an extremely time-consuming and error-prone task and not recommended. By contrast, automated searches for your errors is quick and, if done properly, a reliable method to use. If you wish to use CSPro to automate the search for errors, you must create a Batch Edit application.
Using a Batch Edit application to identify errors is a relatively easy task, though care must be taken to do so correctly. Improperly identifying errors can waste precious personnel resources, so a precise set of rules should be developed with input from subject-matter specialists.
Just as you have two methods available to you when searching for errors, you have two methods available to you for correcting errors: manually or automatically.
Manual correction of a census could take years, and the possibility of human error is great. When large volumes of data are collected in censuses and surveys, it is not always practical to refer to the original document in order to correct an error. Often the data recorded on the original questionnaires are wrong or inconsistent.
With computer editing, both time and the possibility of human error is reduced significantly (just how much depends on how well your logic is written!). The high degree of accuracy and uniformity in computer editing cannot be obtained in manual editing. In computer editing, range checks and within-record consistency checks can easily be made, between-record edits can be done, and unknown information can be allocated (imputed) automatically. If an allocation method is used, you should strive to retain as much of the original information as possible. With a computer editing and imputation system like CSBatch, erroneous data can be corrected immediately and reports can be generated of all errors found and all changes made.
The programmer should plan and design computer edits to inspect the data and have the computer correct them according to specifications supplied by a subject-matter specialist. It would most likely be an extension of the original program written to find the errors—when you reach the point where there is an error, instead of (or in addition to) printing out an error message, you should now correct it with an appropriate value.
Actual methods of making corrections vary depending upon the item. In most instances, data items can be assigned valid codes with reasonable assurance that they are correct by using responses for other data items within the record, or in other records in the questionnaire. When recorded responses are missing, impossible, inconsistent, or unreasonable and cannot be determined from other responses in the same questionnaire, the hot deck technique can be used to assign entries.