In a study, produced by the University of Melbourne and University of NSW, researchers found students who received low grades in NAPLAN tests were ‘much more likely’ to be pulled out of exams in subsequent years, mainly at the request of their parents.

The response was twice as strong in the independent private sector and researchers said the data is consistent with some independent schools attempting to ‘game’ the system.

“Our evidence is consistent with the idea a number of relatively poorly-performing public and independent private schools are gaming the system, to strategically adjust their NAPLAN testing pool,” Professor Michael Coelli from the faculty of Business and Economics, said.

“Manipulating the system like this artificially boosts their average scores, arguably to protect their academic reputation.”

The My School website went live in 2010, providing full transparency around all NAPLAN results.

The research team analysed non-participation rates of 6981 Australian schools (in Grades 3, 5, 7 and 9), between 2008 and 2015.

The findings, published in Economics of Education Review, show the overall rate of student withdrawal due to a formal parental request rose from just 0.33 percent in 2008 (before My School was introduced) to 2.29 percent in 2015.

Comparatively, overall non-participation rates due to absence or exemption barely changed over the same period.

The researchers said these exemptions applied to children with diagnosed learning difficulties as well as children who have been residing in Australia for less than one year with a language background other than English.

The results also reveal that since My School launched, poorly performing independent private schools were withdrawing low-achieving students at twice the rate of public schools.

Researchers have put forward several policy remedies they said could prevent schools from manipulating the system.

These include a requirement for a minimum percentage of students in each school participating in testing each year. Other recommendations include making it mandatory for schools to explain any significant year-on-year changes in non-participation rates and automatically auditing schools which report large drops in test participation.

“Skewing NAPLAN results threatens the efficacy of the public accountability mechanism in Australia and raises doubt over the accuracy of the My School data, which can influence parents’ enrolment decisions,” Coelli, whose research focuses mainly on labour economics, gender earnings gaps and the economics of education, said.

“A few changes could help close clear deficiencies in the system.”