The prevalence of cheating in National Standardised Assessments in South African schools : applying the Jacob and Levitt (2003) method to the Annual National Assessments of 2013

This paper builds on the underlying framework of Jacob and Levitt (2003) and further work by Gustafsson (2014) to determine the prevalence of cheating in South African schools using a nationally representative sample (Verification ANA) of the Annual National Assessments of 2013. This research forms part of a broader research project on 'binding constraints in education' promoted by the Programme to Support Pro-poor Policy Development (PSPPD), housed at Stellenbosch University. The adapted methodology is verified in order to ensure that the suspicious string indicators as discovered by Jacob and Levitt are indeed indicative of suspicious behaviour when applied to the South African ANA. At a national level, the data suggests that cheating or suspicious behaviour is likely to be prevalent in up to 10 percent of schools with respect to Mathematics and Language in Grades 3 and 6. The manner in which schools behave suspiciously varies significantly by province, subject and measure. As many as 37% of primary schools in the Eastern Cape, 26% of primary schools in KwaZulu-Natal and 24% of primary schools in Limpopo show some evidence of cheating in Grade 3 mathematics, compared to just 0% to 2% of primary schools in the Western Cape and Gauteng. Similar extremes are noted in the Grade 6 results. These results suggest that the mere act of assessment and measurement induces behavioural distortions such as gaming behaviour even in the absence of high-stakes.

Grantor dc:publisher.institution

School of Economics

Year dc:date.issued

2016

Author dc:creator

• Ross, Lauren

Advisors dc:contributor.advisor

• Woolard, Ingrid
• Spaull, Nicholas

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/22939

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/22939