RELIABILITY AND VALIDITY OF THE ASER TESTING TOOLS

The ASER reading and math tools are universally thought to be easy to use, simple to understand, and quick to administer. But two questions are often asked about these tools.

First, are they reliable? Reliability in this context refers to consistency in the judgment we make about the level of mastery that a child has achieved (for example, whether she is at letter, word, para or story level in reading). A reliable test would be one where the decision about a child’s level of mastery is the same across multiple test administrations and across different test administrators. In other words, if I test a child today and I test the same child in a few days’ time, will I make the same decision about the child’s ability to read on both occasions? Similarly, if I test a child today and you test the same child tomorrow, will we agree on how much the child can read? If the assessment tool is reliable, the answer to both questions should be yes.

Second, are the testing tools valid? Validity refers to whether the test actually measures the constructs it is intended to measure, in this case children’s ability to read and to do basic arithmetic. One way to judge whether a tool is valid is to compare children’s results on the test with their results on other tests that are intended to measure the same construct. If the assessment tool is valid, children’s results using the ASER tests will be highly correlated with results using other tests that measure mastery of basic reading and arithmetic.

A series of empirical studies were conducted to evaluate the reliability and validity of the ASER testing tools. In a paper discussing the results of these studies, psychometrician Dr Shaher Banu Vagh concludes: “The findings… provide favorable empirical evidence for the reliability and validity of these tests. Specifically, the findings indicate substantial reliability of decisions across repeat measurements, satisfactory inter-rater reliability and favorable evidence for concurrent and convergent-discriminant validity.”

Read the full text of the paper here.
.