This edition of the EF EPI is based on test data from more than 2,200,000 test takers around the world who took the EF Standard English Test (EF SET) or one of our English placement tests in 2019.
EF EPI 2020 scores have been found to correlate strongly with TOEFL iBT 2018 scores (r=0.79) and IELTS Academic Test 2018 scores (r=0.68). These correlations show that, while these tests have different designs and test taker profiles, they reveal similar trends in national English proficiency.
Although the sample of test takers for the EF EPI is biased toward respondents who are interested in pursuing language study and younger adults, the sample is balanced between male and female respondents and represents adult language learners from a broad range of ages.
Only cities, regions, and countries with a minimum of 400 test takers were included in the Index, but in most cases the number of test takers was far greater. The Maldives and Libya were included in the previous edition of the EF EPI but did not have enough test takers to be included in this edition.
The test-taking population represented in this Index is self-selected and not guaranteed to be representative. Only those who want to learn English or are curious about their English skills will participate in one of these tests. This could skew scores lower or higher than those of the general population. However, there is no incentive for test takers to inflate their scores artificially on these low-stakes tests by cheating, as the results are purely for personal use.
The EF SET is free and online, so anyone with an Internet connection can participate. Almost all of our test takers are working adults or young adults finishing their studies. People without Internet access would be automatically excluded. The EF SET site is fully adaptive and 30% of test takers complete the exam from a mobile device.
In parts of the world where Internet usage is low, we would expect the impact of an online format to be strong. This sampling bias would tend to pull scores upward by excluding poorer and less educated people. Nevertheless, open access online tests have proven effective in gathering very large amounts of data about a range of indicators, and we believe they provide valuable information about global English proficiency levels.
To calculate an EF EPI score, we used weighted components which include English tests and the EF EPI from 2019. Inclusion of the previous year’s Index helps to stabilize scores year over year, but test takers from the previous year are not counted in the total test taker count for the current year. Regional averages are weighted by population.
For the first time this year we have moved to an 800 point scale aligned strictly to the CEFR. The aim of this new scale is to eliminate confusion between the EF EPI and the EF SET. The two have always been distinct but were both scored out of 100 until this year. In addition, the EF EPI was often misinterpreted as a percentage. The new scale clears up these ambiguities.
Based on score thresholds, we assign countries, regions, and cities to proficiency bands. This allows recognition of clusters with similar English skill levels and comparisons within and between regions.
The EF EPI does not aim to compete with or contradict national test results, language polling data, or any other data set. Instead, these data sets complement each other. Some are granular but limited in scope to a single age group, country, region, or test taker profile. The EF EPI is broad, examining working-aged adults around the world using a common assessment method. There is no other data set of comparable size and scope, and, despite its limitations, we, along with many policymakers, scholars, and analysts, believe it to be a valuable reference point in the global conversation about English language education.
The EF EPI is created through a different process from the one used by public opinion research organizations such as Euromonitor and Gallup, or by the OECD in skills surveys such as PISA and PIAAC. Those studies select survey participants using age, gender, level of education, income, and other factors. Their survey panels tend to be small, with at most a few thousand participants. Because they have been composed using complex sampling methods, they are considered representative of the entire population. Unfortunately, no such survey of English skills has ever been performed at an international level.
Another source of data about English proficiency comes from national education systems. Many schools test the English skills of every high school student or university applicant using a standardized national assessment. The results may or may not be made public, but educators and government officials use the data to assess the efficacy of education reform and pinpoint areas for improvement. Unfortunately, those national assessments are not comparable to each other, and they are not administered to adults, so while they give a good indication of English proficiency among high school students in one part of the world, they cannot be used for international comparison, nor can they tell us much about adult English proficiency levels.
The EF EPI research series has two separate reports: this main EF EPI report, which is published annually and looks at adult English proficiency; and the EF EPI for Schools (EF EPI-s), which is published biennially and looks at English proficiency among secondary school and university students. This year, we are publishing the EF EPI 2020 edition. All EF EPI reports are available for download at www.ef.com/epi.
EF Education First (www.ef.com) is an international education company that focuses on language, academics, cultural exchange, and educational travel programs. Founded in 1965, EF’s mission is “opening the world through education.” EF is the Official Language Training Partner for the Tokyo 2020 Olympic and Paralympic Games. The EF English Proficiency Index is published by Signum International AG.