We combine data from the Programme for the International Assessment of Adult Competencies (PIAAC) and the Adult Literacy and Lifeskills Survey (ALL), both conducted by the OECD. These surveys were designed to measure adult skills and competencies across different countries. In this section, we describe the ALL and PIAAC dataset, and provide descriptive statistics.
The ALL Data
The ALL measured literacy and numeracy skills of nationally representative samples of 16–65 year olds in participating countries in two rounds. The first round was conducted in 2003 and the second round was conducted between 2006 and 2008. The six countries that took part in the first round were Canada, Italy, Norway, Switzerland, the United States, and Bermuda. In the second round, Australia, Hungary, the Netherlands, and New Zealand participated. The ALL study is the successor of the International Adult Literacy Survey (IALS), which was the first international comparative survey of adult skills undertaken between 1994 and 1998.Footnote 2 Measured skills in the ALL survey include prose literacy, document literacy, numeracy, and problem solving. Literacy was defined as “the knowledge and skills needed to understand and use information from texts and other written formats.” Numeracy was defined as “the knowledge and skills required to manage mathematical demands of diverse situations.”
The PIAAC Data
PIAAC measures skills in three domains: literacy, numeracy and problem solving in technology-rich environments. PIAAC has two cycles of assessment: the first cycle is conducted in two rounds, while the second cycle is expected to take place from 2018 to 2023. The first round of the first cycle took place from January 2008 to October 2013. We use data from the first round of the first cycle. The countries that took part in the first round of the first cycle were Australia, Austria, Belgium, Canada, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Norway, Poland, Russia, the Slovak Republic, Spain, Sweden, the United Kingdom and the United States.
Merging the Datasets and Descriptive Statistics
We identify teachers based on their 4 digit ISCO-88 occupation code.Footnote 3 The occupation codes allow us to distinguish between primary and secondary school teachers. For some countries, no teachers are present in the dataset or the occupation code is not detailed enough to correctly identify them. We exclude these countries from our analyses. The countries that are dropped are Australia, Austria, Canada, Cyprus, Estonia, Finland, Germany, Ireland, Sweden, and the United States.
Table 1 shows the summary statistics by country. Italy, Norway and the Netherlands have most respondents; these are also the countries which were sampled in both PIAAC and ALL. In most countries there is approximately an equal amount of men and women in the samples. The Russian data are not representative for the entire population as inhabitants of the Moscow municipal area were not included in the PIAAC survey. For this reason, we exclude Russia from our analyses.
The number of tertiary educated people differs substantially across countries. This implies that in our analyses, it is better to relate the skills of teachers to all other respondents in a country than to the tertiary educated sub-sample only. In a robustness check, we nonetheless make the latter comparison.
Table 2 shows the number of teachers and the number of primary and secondary school teachers per country sample. The Netherlands, Denmark, Norway, New Zealand, and the United Kingdom have relatively many teachers in their sample, while in other countries like Bermuda, Switzerland and Japan few teachers were sampled. For these latter countries our results are less generalizable to the entire teacher population. In our analyses, we will therefore restrict our sample to those countries for which more than 50 primary or secondary school teachers can be identified.Footnote 4 Practically, this means that when analyzing primary school teachers Japan, Korea, Bermuda and Switzerland will be excluded. When analyzing secondary school teachers we exclude Japan, Korea, Bermuda, Denmark, Slovakia and the Czech Republic.
Tables 1 and 2 reveal that the average teacher is somewhat older than the average respondent. They are also more highly educated on average and the proportion of women is higher for teachers. In robustness analyses, we control for these differences.
Both the ALL and PIAAC surveys measure the skill domains on a 0–500-point scale. While ALL and PIAAC both aim to measure the same skills and use the same measurement scale, they do not employ the same tests. In order to be able to pool the two datasets, we standardize the test results based on the full sample, so that they have a mean of 0 and a standard deviation of 1.Footnote 5
Table 3 shows the mean and standard deviation for the skills in all countries that have teachers in the sample. The results are depicted separately for all respondents and for teachers only. For each country, the mean test score of teachers is higher than the average score of all respondents.
This table reveals the differences in the absolute skill level for teachers between countries. In this article, we study the relative differences of skill levels between teachers and others within countries in order to indicate which segment of the population has become a teacher. We choose to focus on relative differences within a country because we are interested in studying the scope for improvement of the skill level of the teacher population. I.e., if the teacher population is relatively low skilled, then it will be more feasible to improve its skill level because there is a large pool of other people in society which may be induced to become a teacher. If this pool is smaller, improving the teachers’ skill level is less feasible.
However, the absolute differences of skill levels for teachers between countries also provide an important source of information. Policy to improve teacher skills will partly depend on a minimal required skills level. Hence, the interplay between absolute and relative skill levels is policy relevant. Using the table, we can see that literacy and numeracy levels of teachers are lowest in Italy, while Japanese teachers score highest. Teachers in the Netherlands score very highly both on the literacy and numeracy tests relative to teachers in other countries.