The Effectiveness of Turnitin

Plagiarism Prevention and Online Grading in US Higher Education

Turnitin analyzed more than 55 million student papers from 1003 higher education institutions in the United States to understand the change in levels of unoriginal content in student papers and the impact of digital grading over time.

To view the results for institutions similar to yours, select the type of institution (2-year or 4-year) and enter your student population in the calculator below.

4-Year    2-Year
SEE RESULTS >

Methodology

This section explains the methodology behind the study, including customers included in the study, how effectiveness rates were calculated and the determination of the first year of usage.

Criteria for Inclusion

This report analyzes the full population of Turnitin customers at non-profit, higher education institutions in the United States that began using the service before January 1, 2011 and remained a customer through January 1, 2013.

Time-Aligning Accounts Based on Usage

To provide comparable measurement between schools and to produce accurate results, this study aligned customers based on years of use rather than the year in which they began using the service. For example, if Institution A began using Turnitin in 2010, it would have two years of usage (2011, 2012) and if Institution B began using Turnitin in 2008, it would have four years of usage. Customers were aligned based on their years of usage (First Year, Second Year, Third Year, etc.) to study the levels of unoriginal content after identical years of usage.

Calculating the First True Usage Start Year

Institutions that subscribe to Turnitin in a given year do not necessarily generate meaningful usage information in the first year under license. In particular, Turnitin deployments at larger institutions are often part of larger initiatives to change academic integrity policies, adopt a new learning management system or support professional development and training.

For example, an institution of 4,000 students might only submit 40 papers to Turnitin as they test the service in the first year, then increase submission volume to 3,000 in the next year. Using the 40-submission year as a baseline measurement would skew the analysis because it does not reflect a sufficient sample size.

In order to control for this occurrence, this study calculates the “true usage start year” based on the following guidelines:

  • A Customer must have use the product for a full calendar year.
  • The first year is defined as the first year of usage that reflects a minimum of 10 percent of total lifetime submissions by the institution.

"Unoriginal" Content and Writing

The Turnitin service identifies matches in copy between papers submitted to the service and content in the Turnitin databases. Each paper that is submitted receives an Overall Similarity Index (OSI) that shows the percentage of content in the submitted paper that matches to content in the Turnitin databases. This study only considered papers with OSI scores between 50-100 percent of matched content. While 50-100 percent matched content represents a significant amount of unoriginal writing, it does not necessarily equate to plagiarism.

Calculating Effectiveness

After accounts were aligned based on years of true usage, a sum of originality submissions for all institutions was generated, as well as a sum of all papers with levels of unoriginal content above 50 percent for the first year of usage. By dividing the sum of 50-100 percent unoriginal papers into the total submissions, we arrive at a percentage of unoriginal papers. We used this same formula to calculate the level of unoriginal writing for each year of usage and compared it to the first year of usage to measure the change in highly unoriginal papers over time.

Comparisons Between Groups

This study examines the changing levels of unoriginal content by type of institution (two-year or four-year) and student population. It does not provide a basis of comparison for levels of unoriginal writing between groups. Such a comparison would lead to improper conclusions, because each group is measured against a different baseline of unoriginal papers. Therefore, it is not appropriate to conclude that two-year colleges with populations of 3,000-5,000 students (with a percent reduction in unoriginal papers of 78 percent) have a lower rate of unoriginal content than four-year schools with populations of 5,000-10,000 (26 percent reduction) because the study (intentionally) does not provide the baseline percentage of unoriginal content per group.

Back to the top