Which of the following is used for test data generation ?

Concept: Test-data generation strategies in software testing are classified by the information used to derive the inputs. White-box (structural) testing derives test data from the program's internal code, branches, and conditions; black-box (functional) testing derives test data from the specification/behaviour without looking at the code; and boundary-value analysis (BVA) is a specific, systematic technique — usable within black-box analysis of input domains, and also applicable to white-box branch/loop conditions — that targets values at and near the edges of an input range, since defects cluster at boundaries.

Application: The stem asks generically which of the listed items is "used for test data generation." Each one independently qualifies: white-box testing generates data via path, branch, and condition-coverage criteria; black-box testing generates data via specification-based rules such as equivalence classes; and boundary-value analysis generates data specifically at boundary limits (for example, minimum, minimum-1, maximum, maximum+1). None of the three is exclusive of the others — they are all standard, recognised means of generating test data, just derived from different information sources.

Cross-check: If only one of white-box, black-box, or boundary-value analysis were selected, the other two recognised sources of test data (code-structure-driven data and specification-driven data) would be left unaccounted for. Since the stem does not restrict "test data generation" to a single source, the only option that is not under-inclusive is the one covering all three together.