Validity
Any psychological evaluation instrument must fully satisfy the requirements of validity,
defined as: “the extent to which a test measures what it intends to measure” (Anastasi,
1976). Validity is the relationship between the score obtained in the test and a certain
external criterion or performance. This criterion may belong to any field: staff
recruitment, success in school, or medical classification. The literature indicates various
types of psychological validity. They must be understood as techniques by which
optimisation of a psychological measurement instrument is attempted. The modality most
frequently encountered in the literature (Anastasi, 1976; Cohen and colab., 2000) is based
on the taxonomy below:
- construct validity;
- criterion validity;
- content validity.
Construct validity represents the extent to which it can be held that the test measures a
specific feature or variable. The construct validity of BTPAC tests has been achieved
through various means: convergent and divergent validity, factorial analysis, etc.
Criterion validity indicates the extent to which the test predicts accurately for a sample
of future behaviours. In this case, performance in a test must be reported to another
performance called criterion. Criterion is defined as the standard against which is
compared a test performance.
Content validity involves the systematic examination of the test content as to verify
whether it covers a representative sample of the field under evaluation (Anastasi, 1976).
In order to have high content validity, the items making up the test must be representative
for what the test intends to measure.
Statistical procedures employed in item analysis are, more often than not, extremely
complex, this is why we will only present the main indicators employed in the analysis of
BTPAC items:
- difficulty index;
- discrimination index.
The theoretical value of the difficulty index is between 0 (no subject solved the item
correctly) and 1 (all subjects answered an item correctly). In case of BTPAC only those
items were included whose difficulty indices ranged between 0,30 and 0,70.
The discrimination index of an item measures how well an item manages to separate or
discriminate between subjects with high scores and subjects with low scores. For BTPAC
the discrimination indices of all items was calculated and only those whose minimum
value was 0.30 were retained.
In conclusion, the validation process for BTPAC was carried out as outlined above. A
pilot study analysed the correlation between evaluators as regards items content. Were