Understand what test data is and describe the following types of test data:

The idea

Test data is the input chosen to run a program and check it behaves correctly. It is selected deliberately to exercise normal, extreme and invalid cases. Normal test data is sensible, typical input the program should accept, such as an age of 25 for an adult-age field. Boundary (extreme) test data sits at the edges of the accepted range, such as the minimum and maximum permitted ages, because errors often hide at limits. Erroneous (invalid) test data is input the program should reject, such as a negative age or letters in a numeric field, checking that validation and error handling work. Choosing all three types gives confidence the program is robust, not just correct for one easy case.

What earns the marks

• Defines test data as input values selected to check that a program produces the expected output and handles input correctly.
• Describes normal test data as typical, valid input the program should accept and process correctly.
• Describes boundary or extreme test data as values at the limits of the accepted range, where faults commonly occur.
• Describes erroneous or invalid test data as input the program should reject, testing validation and error handling.
• Explains why a range of test data types is needed to confirm a program is robust rather than correct for only one case.

Common mistakes and how to put them right

• Treating boundary data as the same as normal data.

  Put it right: Is to recognise boundary values sit at the very edge of the accepted range, not in the middle.

• Thinking erroneous data should be processed normally.

  Put it right: Is to recognise it is input the program should reject through validation.

• Assuming one valid input proves the program works.

  Put it right: Is to select normal, boundary and erroneous data to test different paths.

Exam tips

• Match each example input to its type and justify the choice by reference to the accepted range.
• For boundary questions, give the exact minimum and maximum values rather than a value near them.
• When asked why test data matters, link it to checking validation and confirming correct output.