5.3.3 Data entry - Validation and Verification

There are many different errors that can occur when inputting data into a system. Many are simple transcription errors (errors that occur when data is copied from a document into a computer).  To try and reduce the amount of input errors, a system designer can build in validation and verification checks into the software that the data is entered into.

It is important to understand that these checks cannot eliminate all mistakes, they just try and ensure that the data is as accurate as is reasonably possible for the purpose it is being collected for.

• Take a multiple choice quiz on validation and verification

VERIFICATION:

This type of data entry check is used to try and ensure that the data is ACTUALLY correct.  Since a computer cannot do this by itself it has to involve a human in some way.

Double entry - the data is entered twice, possibly by two different operators.  The two entries are then compared against each other and a warning given if they do not match.

Visual checks- this is basically proof reading, the data that has been entered into the computer is visually checked by a human, either on screen or from a printout, to be sure that it matches the data source.

VALIDATION:

This type of data entry check only checks that the data is VALID, in other words it is sensible. It does NOT mean that the data is actually correct!

Range checks - these are used with numerical fields to limit the range of numbers a user can enter.  For example:

• Imagine if you were making a database for a user to store details of yachts that are sold in a second-hand boatyard and you know that they never sell anything longer than 20 metres and less than 13 metres.
• It would make sense to make a range-check validation rule for the 'LENGTH' field so that if an entry was made that was outside this range then a warning message would pop up and the entry would not be accepted.
• A typical range-check validation rule might be: BETWEEN 13 AND 20
• This could also be entered as:  >=13 AND <=20

Format checks - these only allow valid text or numbers.  Many use input 'masks' such as 000LLL.  Such a mask would only allow 3 numbers followed by 3 letters.

Member lists - this method of validation is used to limit entries to those that are members of a list of allowed entries. The choices may appear in a drop-down list to reduce transcription errors or be in the form of a list of acceptable entries that the input is checked against.

• For example, a member list for a Vehicle Make field in a vehicle database might be Renault, Ford, Honda etc.
• A member list validation rule such as "Renault" OR "Ford" OR "Honda" would limit entries to members of this list.

Check digits - this type of check is used with numbers. An extra 'check digit' is calculated from the numbers to be entered and added to the end.  The numbers can then be checked at any stage by re-calculating the check digit from the other numbers and seeing if it matches the one entered. One example where a check digit is used is in the 10 digit ISBN number which uniquely identifies books. The last number of the ISBN is actually the check digit for the other numbers, for example - the ISBN 0192761501.

• The Modulus-11 system is one example of a check digit system and it can apparently detect 99% of input errors (there is still a small chance that more than one error occurs and the resulting number has the same check digit).

How the Modulus-11 system works:

• Each number input is 'weighted' by multiplying it by it's position +1
• The weightings are adding together (the checksum) and this is divided by the prime number 11
• The remainder is then subtracted from 11 to get the check digit
• The check digit is then added to the end of the number.
• To check if the number entered is correct the calculation is repeated but this time the check digit is included in the calculation.
• Because of this the remainder should now be 0 and it it is then there is a 99% chance the other numbers were entered correctly.
• Note: if the check digit is a 10 then this is printed as an x to keep it to a single character.

i.e. the number 019276150 would have a check digit of 1 as shown below.

• The weighted checksum without the check digit is 186.  186 divided by 11 = 16 remainder 10.  11 minus the remainder therefore gives a check digit of 1
• When the check digit is included in the calculation in position 1 then the checksum is 187.  187 divided by 11 = 17 remainder 0 so the check digit matches the rest of the numbers which means they were probably entered correctly.

Presence checks - these simply check that an entry has been made in a particular field. It it has not then the system will not allow the record to be saved or any entries to be made in later fields.