PISA 2022 Results (Volume V)

Annex A1. Construction of indices

Explanation of the indices

This section explains the indices derived from the PISA 2022 student, well-being and Information and Communication Technology (ICT) familiarity questionnaires used in this volume. Several PISA measures reflect indices that summarise responses from students to a series of related questions. The questions were selected from a larger pool on the basis of theoretical considerations and previous research. The PISA 2022 Assessment and Analytical Framework (OECD, 2023[1]) provides an in-depth description of this conceptual framework. Item response theory (IRT) modelling and classical test theory were used to test the theoretically expected behaviour of the indices and to validate their comparability across countries. For a detailed description of the methods, see the section “Statistical criteria for reporting on scaled indices” in this Annex, and the PISA 2022 Technical Report (OECD, 2024[2]).

This volume uses four types of indices: simple indices, complex composite indices, new scale indices and trend scale indices. In addition to these indices, several single items of the questionnaires are used in this volume. The volume also uses data collected on students’ performance in mathematics, reading and science. These assessments are described in the PISA 2022 Assessment and Analytical Framework (OECD, 2023[1]), the PISA 2022 Technical Report (OECD, 2024[2]) and in Volume I of PISA 2022 Results (OECD, 2023[3]).

Simple indices are constructed through the arithmetic transformation or recoding of one or more items in the same way across assessments. Here, item responses are used to calculate meaningful indices, such as the recoding of the four-digit ISCO-08 codes into “Highest parents’ socio-economic index (HISEI)” or teacher-student ratio based on information from the school questionnaire.

Complex composite indices are based on a combination of two or more indices. The PISA index of economic, social and cultural status (ESCS) is a composite score derived from three indicators related to family background.

Scale indices are constructed by scaling multiple items. Unless otherwise indicated, the two-parameter logistic model (2PLM) (Birnbaum, 1968[4]) was used to scale items with only two response categories (i.e. dichotomous items), while the generalised partial credit model (GPCM) (Muraki, 1992[5]) was used to scale items with more than two response categories (i.e. polytomous items).1 Values of the index correspond to standardised Warm likelihood estimates (WLE) (Warm, 1989[6]).

For details on how each scale index was constructed, see the PISA 2022 Technical Report (OECD, 2024[2]). In general, the scaling was done in two stages:

1. The item parameters were estimated based on all students from approximately equally weighted countries and economies;2 only cases with a minimum number of three valid responses to items that are part of the index were included. For the trend scales, the scaling process began by fixing the item parameters of the trend items to the parameters that had been estimated for each group in the previous assessment, a procedure called fixed parameter linking. To compute trends, a scale needed to have at least three trend items, but some trend scales consisted of both trend items and new items. In this case, the item parameters for the trend items were fixed at the beginning of the scaling process, but the item parameters for the new items were estimated using the PISA 2022 data.

2. For new scale indices, the Warm likelihood estimates were then standardised so that the mean of the index value for the OECD student population was zero and the standard deviation was one (countries were given approximately equal weight in the standardisation process²). For the trend scales, to ensure the comparability of the scale scores from the current assessment to the scale scores from the previous assessment, the original WLEs of PISA 2022 were transformed using the same transformation constants of the original WLEs from the assessment to which the current assessment was linked.

Sequential codes were assigned to the different response categories of the questions in the sequence in which the latter appeared in the student, school, ICT or well-being questionnaire. For reversed items, these codes were inverted for the purpose of constructing indices or scales.

Negative values for an index do not necessarily imply that respondents answered negatively to the underlying questions (e.g. reporting no support from teachers or no school safety risks). A negative value merely indicates that a respondent answered more negatively than other respondents did on average across OECD countries. Likewise, a positive value on an index indicates that a respondent answered more favourably, or more positively, on average, than other respondents in OECD countries did (e.g. reporting more support from teachers or more school safety risks).

Some terms in the questionnaires were replaced in the national versions of the student, school, ICT or well-being questionnaire by the appropriate national equivalent (marked through brackets < > in the international versions of the questionnaires). For example, the term < qualification at ISCED level 5A > was adapted in the United States* to “Bachelor’s degree, post-graduate certificate program, Master’s degree program or first professional degree program”. All the context questionnaires, including information on nationally adapted terms, and the PISA international database, including all variables, are available through www.oecd.org/pisa.

Statistical criteria for reporting on scaled indices

The internal consistency of scaled indices and the invariance of item parameters are the two approaches that were used to decide on the reporting of indices. All indices reported in this volume met the criteria of both approaches. Indices were omitted for countries and economies where one or more of the criteria were not met. For countries/economies with more than one language version (e.g. Finland offered versions of the student questionnaire in Finnish and Swedish), the criteria were judged independently for each language version.3 Details about the scaling procedures and the construct validation of all context questionnaire data are provided in the PISA 2022 Technical Report (OECD, 2024[2]).

Internal consistency of scaled indices

The internal consistency was used in PISA 2022 to examine the reliability of scaled indices and as a criterion for reporting. Internal consistency refers to the extent to which the items that make up an index are inter-related. Cronbach’s Alpha was used to check the internal consistency of each scale within countries/economies and to compare it across countries/economies. The coefficient of Cronbach’s Alpha ranges from 0 to 1, with higher values indicating higher internal consistency. Similar and high values across countries/economies indicate reliable measures across countries/economies. Commonly accepted cut-off values are 0.9 for excellent, 0.8 for good, and 0.7 for acceptable internal consistency. Indices are not reported for countries and economies with values below 0.6.

Cross-country comparability of scaled indices

The invariance of item parameters was used in PISA 2022 to examine the cross-country comparability of scaled indices and as a criterion for reporting. It determined whether the item parameters of an index could be assumed to be the same or invariant across countries/economies and across language versions (international item parameter).

In a first step, item parameters were estimated using data from all individuals with available data from all countries/economies. In a second step, the fit of the international parameters for each item was evaluated for each country/economy and language version using the root mean square deviance (RMSD). Values close to zero signal a good item fit, indicating that the international model accurately describes student responses within countries/economies and across language versions. In 2022 PISA used an even more conservative approach than in previous assessments: any country/economy and language version that received a value above 0.25 was flagged. In 2018 and 2015, a cut-off of 0.3 was used. For any flagged item specific parameters were calculated. Steps were repeated until all items exhibited RMSD values below 0.25.

For each index, a country/economy needed to have at least three items with international parameters to be considered comparable to the results of other countries/economies and language versions. Indices are not reported for countries/economies in which one or more language version had fewer than three items with international parameters. For the reporting on trends for indices, a country/economy needed to have at least three trend items with international parameters in order to be considered comparable to the results of the previous assessment to which the current assessment was linked. Results for the trends of indices were not reported for countries/economies in which one or more language groups had fewer than three trend items with international parameters for the index.

The different indices used in this volume are described in the following sections. Those countries/economies and language versions that received specific item parameters are highlighted. The PISA 2022 Technical Report (OECD, 2024[2]) provides more details on the cross-country comparability of indices, including the items concerned and the specific item parameters for each country/economy and language version listed.

Complex composite indices

The PISA index of economic, social and cultural status (ESCS)

The PISA index of economic, social and cultural status (ESCS) is a composite score derived, as in previous assessments, from three indicators related to family background: parents’ highest education, in years (PAREDINT), parents’ highest occupational status (HISEI) and home possessions (HOMEPOS).

Parents’ highest level of education, in years (PAREDINT): The index of the highest education of parents, in years, was based on the median cumulative years of education associated with completion of the highest level of education attained by parents (HISCED). Parents’ highest level of education was derived from students’ responses to questions about their parents’ education (ST005 and ST006 for mother’s level of education, and ST007 and ST008 for father’s level of education). Responses were classified according to ISCED-11 (UNESCO, 2012[7]) using the following categories: (1) Less than ISCED Level 1, (2) ISCED level 1 (primary education), (3) ISCED level 2 (lower secondary), (4) ISCED level 3.3 (upper secondary education with no direct access to tertiary education), (5) ISCED level 3.4 (upper secondary education with direct access to tertiary education), (6) ISCED level 4 (post-secondary non-tertiary), (7) ISCED level 5 (short-cycle tertiary education [at least two years]), (8) ISCED level 6 (Bachelor’s or equivalent first or long first-degree programme [three to more than four years]), (9) ISCED level 7 (Master’s or equivalent long first-degree programme [at least five years]) and (10) ISCED level 8 (Doctoral or equivalent level). In the event that students’ responses to the two questions about their mothers’ and fathers’ level of education conflicted (e.g. if a student indicated in ST006 that their mother has a postsecondary qualification but indicated in ST005 that their mother had not completed lower secondary education), the higher education level provided by the student was used. This differs from the PISA 2018 procedure where the lower level was used. Indices with these categories were provided for a student’s mother (MISCED) and father (FISCED). In addition, the index of parents’ highest level of education (HISCED) corresponded to the higher ISCED level of either parent.

The index of parents’ highest level of education was recoded into the estimated number of years of education (PAREDINT). This international conversion was determined by using the PISA 2018 measure of cumulative years of education associated with parents’ completion of the highest level of education across countries/economies for each ISCED level. The correspondence is available in the PISA 2022 Technical Report (OECD, 2024[2]).

Parents’ highest occupational status (HISEI): Occupational data for both the student’s father and the student’s mother were obtained from responses to open-ended questions (ST014 and ST015). The responses were coded to four-digit ISCO codes (ILO, 2007) and then mapped to the international socio-economic index of occupational status (ISEI) using the 2008 version of both (Ganzeboom and Treiman, 2003[8]). Three indices were calculated based on this information: father’s occupational status (BFMJ2); mother’s occupational status (BMMJ1); and the highest occupational status of parents (HISEI), which corresponds to the higher ISEI score of either parent or to the only available parent’s ISEI score. For all three indices, higher ISEI scores indicate higher levels of occupational status.

Home possessions (HOMEPOS): Home possessions were used as a proxy measure for family wealth. In PISA 2022, students reported the availability of household items at home, including books at home and country-specific household items that were seen as appropriate measures of family wealth in the country’s context. HOMEPOS is a summary index of all household and possession items (ST250, ST251, ST253, ST254, ST255, ST256). Some HOMEPOS items used in PISA 2018 were removed in PISA 2022 while new ones were added (e.g. new items developed specifically with low-income countries in mind). Furthermore, some HOMEPOS that were previously dichotomous (yes/no) items were revised to polytomous items (1, 2, 3, etc.) making it possible to capture a greater variation in responses. Note that all countries/economies and language versions received unique item parameters for the country/economy-specific items (i.e. no international parameters were estimated for these items) and that for some items, the response categories were collapsed to align with the response categories used in previous assessments (see Tables 19.15 and 19.16 of the PISA 2022 Technical Report (OECD, 2024[2]) for details).

For the purpose of computing the PISA index of economic, social and cultural status (ESCS), values for students with missing data on one of the three components (PAREDIND, HISEI or HOMEPOS) were imputed (see (OECD, 2020[9]; Avvisati, 2020[10]; OECD, 2024[2]) for details). If students had missing data for more than one component, the ESCS was not computed; a missing value was assigned instead. In PISA 2022, ESCS was computed by attributing equal weight to the three components. The final ESCS variable is standardised, so that 0 is the score of an average OECD student and 1 is the standard deviation across approximately equally weighted OECD countries.²

ESCS scores for PISA 2012, PISA 2015 and PISA 2018 were recomputed to be comparable to the respective scores for PISA 2022. More details are provided in the PISA 2022 Technical Report (OECD, 2024[2]).

Simple indices

Education level

PISA collects data on study programmes available to 15-year-old students in each country/economy. This information is obtained through the student tracking form and the Student Questionnaire (ST002). All study programmes were classified using the International Standard Classification of Education (ISCED 1997). From this information, a study programme level and orientation index (ISCEDP) was derived: a three-digit index that describes whether students were at the lower or upper secondary level (ISCED 2 or ISCED 3) and the type of programme in which they were enrolled. This index was used to classify students into those attending upper vs. lower secondary education programmes.

Immigrant background

Information on the country of birth of the students and their parents was collected from students (ST019). Three binary country-specific indices indicate whether the student (COBN_S), mother (COBN_M) and father (COBN_F) were born in the country of assessment or elsewhere. The index on immigrant background (IMMIG) is calculated from these indices, and has the following categories: (1) native students (those students who had at least one parent born in the country of assessment); (2) second-generation students (those born in the country of assessment but whose parent[s] were born in another country); and (3) first-generation students (those students born outside the country of assessment and whose parents were also born in another country). Students with missing responses for either the student or for both parents were given missing values for this variable.

Socio-economic profile of the school

The average PISA index of economic, social and cultural status (ESCS) of a school was used as an indicator of the socio-economic profile of a school. To define advantaged and disadvantaged schools, all schools in each PISA-participating education system are ranked according to their average PISA index of economic, social and cultural status (ESCS) and then divided into four groups with approximately an equal number of students (quarters). Schools in the bottom quarter are referred to as “socio-economically disadvantaged schools”; and schools in the top quarter are referred to as “socio-economically advantaged schools”.

Working for pay before or after school

Students’ answers on how many days during a typical school week they worked for pay before going to school and/or after leaving school in questions ST294 and ST295 were scaled into the index of “Work for pay before or after school”. Each item included six response options (“0 days”, “1 day”, “2 days”, “3 days”, “4 days”, “5 or more days”). Values on this index range from 0 (no work for pay) to 10 (10 or more times of working for pay per week).

Trend scale indices

Mathematics anxiety

The index of mathematics anxiety (ANXMAT) was constructed using the six student responses to question ST345. This question asked students how much they agree (“strongly agreed”, “agreed”, “disagreed” or “strongly disagreed”) with six statements about their feelings when studying mathematics (e.g. “I often worry that it will be difficult for me in mathematics classes”; “I get very tense when I have to do mathematics homework”). Positive values in this index mean that students reported greater anxiety towards mathematics than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Baku (Azerbaijan) (Azeri, Russian), Brazil (Portuguese), Cambodia (Khmer), the Czech Republic (Czech), Georgia (Georgian, Azerbaijani, Russian), Kazakhstan (Kazakh, Russian), Malaysia (Malay), Moldova (Russian), Mongolia (Mongolian, Kazakh), the Slovak Republic (Slovak, Hungarian), the Ukrainian regions (18 of 27) (Ukrainian, Russian) and Uzbekistan (Uzbek, Karakalpak).

Teacher support in mathematics

Students were asked how often (“never or hardly ever”, “some lessons”, “most lessons”, “every lesson”) certain things happen in their mathematics classes (e.g. “The teacher shows an interest in every student’s learning”; “The teacher gives extra help when students need it”). The four statements of question ST270 were combined to create an index of teacher support (TEACHSUP) with an average of zero and a standard deviation of one across OECD countries. Positive values on the indices mean that the student reported more frequent teacher support in mathematics lessons than did students on average across OECD countries.

In 2012 students answered similar statements about teacher support and disciplinary climate in mathematics lessons. One item from the scale received specific item parameters for Hong Kong (China)* (Chinese).

New scale indices

Confidence in self-directed learning

Students were asked how confident (“not at all confident”, “not very confident”, “confident”, “very confident”) they are about different aspects related to self-directed learning (e.g. “Finding learning resources on line on my own”; “Planning when to do schoolwork on my own”) if their school building closed again in the future. Students’ responses to the eight statements (ST355) were combined into an index of “Confidence in self-directed learning” (SDLEFF) whose average is zero and standard deviation is one across OECD countries.4 Positive values in the index indicate that the student felt more confident than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Cambodia (Khmer), Indonesia (Indonesian), Kazakhstan (Kazakh), Mongolia (Mongolian, Kazakh), Montenegro (Montenegrin, Albanian), the Philippines (English) and Thailand (Thai).

Co-operation

Question ST343 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”). with statements about a range of behaviours indicative of cooperation (e.g., “I work well with other people.”, “I get annoyed when I have to compromise with others.”). Answers to the 10 questions were used to build the index of “Co-operation” (COOPAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Each of the 10 items included in this scale had five response options. Positive values in the index indicate that the student reported co-operating more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for France (French), Korea (Korean), the Netherlands* (Dutch) and Norway (Bokmål, Nynorsk).

Cognitive activation in mathematics: Foster reasoning

Students answered a question (ST285) on how often (“Never or almost never”, “Less than half of the lessons”, “About half of the lessons”, “More than half of the lessons”, “Every lesson or almost every lesson”) during the ongoing school year their mathematics teacher showed a range of behaviours indicative of fostering mathematics reasoning (e.g., “The teacher asked us to explain our reasoning when solving a mathematics problem.”, “The teacher asked us to defend our answer to a mathematics problem.”). Answers to the questions were used to build the index of “Cognitive activation in mathematics: Foster reasoning” (COGACRCO) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student was frequently exposed to a range of behaviours indicative of the promotion of mathematical thinking by their mathematics teachers, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Cambodia (Khmer), Guatemala (Spanish), Thailand (Thai) and Viet Nam (Vietnamese).

Creative school and class environment

Question ST335 asked students if they agree (“Strongly disagree”, “Disagree”, “Agree”, “Strongly agree”) with statements about the degree to which creative thinking is fostered and supported in their school and class environment (e.g., “My teachers value students’ creativity.”, “At school, I am given a chance to express my ideas.”). Answers to the questions were scaled into the index of “Creative school and class environment” (CREATSCH) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported exposure to creative thinking in their school and class environment, more than did students on average across OECD countries.

Confidence in 21st-century mathematics skills

Students were asked how confident (“Not at all confident”, “Not very confident”, “Confident”, “Very confident”) they felt about having to do a range of mathematical reasoning and 21^st century mathematics tasks (e.g., “Extracting mathematical information from diagrams, graphs, or simulations”, “Using the concept of statistical variation to make a decision”). Students’ responses to the 10 statements (ST291) were scaled into the index of “Confidence in 21st-century mathematics skills” (MATHEF21) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student felt more confident than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Baku (Azerbaijan) (Azeri), El Salvador (Spanish), Germany (German), Indonesia (Indonesian), Kazakhstan (Kazakh, Russian), Malaysia (Malay), North Macedonia (Macedonian, Albanian), Qatar (Arabic), Saudi Arabia (Arabic, English), Thailand (Thai), the United Arab Emirates (Arabic) and Uzbekistan (Uzbek, Karakalpak).

Confidence in digital skills

Question IC183 asked students asked how well (“I cannot do this”, “I struggle to do this on my own”, “I can do with a bit of effort”, “I can easily do this” and an additional response option “I don’t know what this is” which was recoded as missing prior to scaling) they can do various tasks using digital resources (e.g., “Search for and find relevant information online”, “Write or edit text for a school assignment”). Answers to the 14 items were scaled into the index of “Confidence in digital skills” (ICTEFFIC) with an average value of zero and a standard deviation of one across OECD countries. Positive values in the index indicate that the student reported confidence in their digital competencies, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Bulgaria (Bulgarian), Finland (Finnish, Swedish), Hong Kong (China)* (Chinese), Israel (Arabic), Jordan (Arabic), Kazakhstan (Kazakh), Korea (Korean), Saudi Arabia (Arabic, English), Chinese Taipei (Chinese) and Thailand (Thai).

Curiosity

Question ST301 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”) with statements about a range of behaviours indicative of curiosity (e.g., “I like to know how things work.”, “I am more curious than most people I know.”). Answers to the 10 questions were scaled into the index of ‘Curiosity” (CURIOAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported curiosity more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Baku (Azerbaijan) (Russian), Colombia (Spanish), the Dominican Republic (Spanish), El Salvador (Spanish), Georgia (Georgian, Azerbaijani, Russian), Guatemala (Spanish), Kazakhstan (Kazakh), Moldova (Russian), Mongolia (Mongolian, Kazakh), Norway (Nynorsk), Paraguay (Spanish), Peru (Spanish), the Philippines (English) and Uzbekistan (Uzbek, Karakalpak).

Emotional control

Question ST313 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”) with statements about a range of behaviours indicative of emotional control (e.g., “I keep my emotions under control.”, “I get mad easily.”). Answers to the 10 questions were scaled into the index of “Emotional control” (EMOCOAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported emotional control, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Finland (Finish, Swedish), Greece (Greek), Indonesia (Indonesian), Kazakhstan (Kazakh), Malaysia (Malay, English), Thailand (Thai) and Viet Nam (Vietnamese).

Empathy

Question ST311 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”) with statements about a range of behaviours indicative of empathy (e.g., “I predict the needs of others.”, “It is difficult for me to sense what others think.”). Answers to the 10 questions were scaled into the index of “Empathy” (EMPATAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported empathy, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Hong Kong (China)* (Chinese), Hungary (Hungarian), Korea (Korean), Macao (China) (Chinese) and the Netherlands* (Dutch).

Exposure to mathematical reasoning and 21st century mathematics tasks

Students answered a question (ST276) on how often (“Frequently”, “Sometimes”, “Rarely”, “Never”) they had encountered a range of different types of mathematics tasks related to mathematical reasoning and 21st century mathematics tasks during their time at school (e.g., “Extracting mathematical information from diagrams, graphs, or simulations”, “Using the concept of statistical variation to make a decision”). Answers were scaled into the index “Exposure to mathematical reasoning and 21st century mathematics tasks” (EXPO21ST) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported frequent exposure to a range of different types of mathematics tasks related to mathematical reasoning and 21st century mathematics tasks, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Denmark* (Danish), Iceland (Icelandic), Norway (Bokmål), North Macedonia (Albanian), Qatar (Arabic), Thailand (Thai) and the United Arab Emirates (Arabic).

Family support

Family support (FAMSUP) was measured by asking students, in question ST300, how often (“never or almost never”, “about once or twice a year”, “about once or twice a month”, “about once or twice a week”, “every day or almost every day”) their parents or someone in their family do different things with them indicative of family support (e.g. “Discuss how well you are doing at school”; “Eat the main meal with you”; or “Spend time just talking with you”). An index of family support with an average of zero and a standard deviation one across OECD countries is formed by combining students’ responses to ten scenarios. Students with positive values on this index perceived their family as more supportive than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Denmark* (Danish), Estonia (Russian), Guatemala (Spanish), Hong Kong (China)* (Chinese), Japan (Japanese), Macao (China) (Chinese, Portuguese), the Netherlands* (Dutch), North Macedonia (Albanian), Poland (Polish), Qatar (Arabic), the Slovak Republic (Slovak, Hungarian) and Thailand (Thai).

Growth mindset

Question ST263 asked students if they agree (“Strongly disagree”, “Disagree”, “Agree”, “Strongly agree”) with a range of statements indicative of their mindset (e.g., “Your intelligence is something about you that you cannot change very much.”, “Some people are just not good at mathematics, no matter how hard they study.”). An index of “Growth mindset” (GROSAGR) with an average of zero and a standard deviation one across OECD countries is formed by combining students’ responses to these questions. Students with positive values on this reported more of a growth mindset than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Austria (German), El Salvador (Spanish), Germany (German), Indonesia (Indonesian), North Macedonia (Albanian) and the United States* (English).

Highest expected educational level

Highest expected educational level (EXPECEDU) was measured by asking students, in question ST327, which of a list of possible educational levels they expect to complete. Answers were transformed into the index which was newly created for 2022. Values on the index can range from 1 “Less than ISCED level 2” to 9 “ISCED level 8”.

ICT use in enquiry-based learning activities

Students’ frequency ratings of how often they use digital resources for various school-related activities (e.g., “Create a multi-media presentation with pictures, sound or video”, “Track the progress of your own work or projects”) in question IC174 were scaled into the index of “ICT use in enquiry-based learning activities” (ICTENQ) with an average value of zero and a standard deviation of one across OECD countries. Each of the 10 items included in this scale had five response options (“Never or almost never”, “About once or twice a year”, “About once or twice a month”, “About once or twice a week”, “Every day or almost every day”). Positive values in the index indicate that the student reported more frequently using digital resources for these practices, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Denmark* (Danish), Kazakhstan (Kazakh), Saudi Arabia (Arabic, English), Sweden (Swedish, English) and the United States* (English).

Information-seeking regarding future career

Students’ ratings of whether they had undertaken a range of possible activities to find out about future study or types of work (e.g., “I did an internship.”, “I researched the internet for information about careers.”) in question ST330 were scaled into the Index of “information-seeking regarding future career” (INFOSEEK) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Each of the 11 items included in this scale had three response options (“Yes, once”, “Yes, two or more times”, “No”). The index has an average of zero and a standard deviation one across OECD countries. Students with positive values reported had undertaken activities to find out about future study or types of work, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Argentina (Spanish), Austria (German), Brunei Darussalam (English), Bulgaria (Bulgarian), Canada* (English, French), Colombia (Spanish), Croatia (Croatian), Cyprus (Greek, English), the Czech Republic (Czech), Denmark* (Danish), the Dominican Republic (Spanish), Georgia (Georgian, Azerbaijani, Russian), Hungary (Hungarian), Jamaica* (English), Kazakhstan (Kazakh, Russian), Kosovo (Albanian, Serbian), Lithuania (Lithuanian, Russian, Polish), Macao (China) (Chinese, Portuguese, English), Mexico (Spanish), Montenegro (Montenegrin, Albanian), Morocco (Arabic), the Netherlands (Dutch), New Zealand* (English), Peru (Spanish), the Philippines (English), Poland (Polish), Saudi Arabia (Arabic, English), Switzerland (German, Italian), Chinese Taipei (Chinese), the United Arab Emirates (English), the United States* (English) and Uzbekistan (Uzbek, Karakalpak).

Mathematics self-efficacy: Formal and applied mathematics

Students were asked how confident (“Not at all confident”, “Not very confident”, “Confident”, “Very confident”) they are about having to do a range of formal and applied mathematics tasks (e.g., “Calculating how much more expensive a computer would be after adding tax”, “Solving an equation like 2(x+3) = (x+3)(x-3)”). Students’ responses to the nine statements (ST290) were scaled into the index of “Mathematics self-efficacy: Formal and applied mathematics” (MATHEFF) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale was linked to the MATHEFF scale in PISA 2012 and was scaled using the PCM, in line with the model used in PISA 2012. Also, it used a within- construct matrix sampling design. Positive values in the index indicate that the student felt more confident than did students on average across OECD countries. Note that in PISA 2012, the response options were presented to the students ordered from “Very confident” to “Not at all confident” possibly eliciting different response patterns related to the format of the question, and not necessarily related the construct. Because of this, caution should be exercised when comparing scale scores across these two cycles.

One or more items from the scale received specific item parameters for Albania (Albanian), Colombia (Spanish), Israel (Hebrew), Malaysia (Malay), Mexico (Spanish), North Macedonia (Macedonian, Albanian), Paraguay (Spanish), Peru (Spanish), the Philippines (English), Uzbekistan (Uzbek, Karakalpak) and Viet Nam (Vietnamese).

Persistence

Question ST307 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”) with statements about a range of behaviours indicative of persistence (e.g. “I keep working on a task until it is finished.”, “I give up after making mistakes.”). Answers to the 10 questions were scaled into the index of “Persistence” (PERSEVAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported persistence, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Finland (Finnish, Swedish).

Proactive mathematics study behaviour

Question ST293 asked students how often (“Never or almost never”, “Less than half of the time”, “About half of the time”, “More than half of the time”, “All or almost all of the time”) they engaged in behaviours indicative of effort and persistence in mathematics (e.g. “I actively participated in group discussions during mathematics class.”, “I put effort into my assignments for mathematics class.”). Answers to the 8 questions were scaled into the index of “Proactive mathematics study behaviour” (MATHPERS) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported proactive mathematics study behaviour, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Cambodia (Khmer), Costa Rica (Spanish), Israel (Arabic), Paraguay (Spanish), Qatar (Arabic), Thailand (Thai), Uzbekistan (Uzbek, Karakalpak) and Viet Nam (Vietnamese).

Quality of access to ICT

Question IC172 asked students asked students if they agree (“Strongly disagree”, “Disagree”, “Agree”, “Strongly agree”) with various statements about ICT resources at their school (e.g. “There are enough digital devices with access to the Internet at my school.”, “The school’s Internet speed is sufficient.”). Answers to the nine questions were scaled into the index of “Quality of access to ICT” (ICTQUAL) with an average value of zero and a standard deviation of one across OECD countries. Positive values in the index indicate that the student reported quality access to ICT resources at their school, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Kazakhstan (Kazakh).

Quality of student-teacher relationships

Students’ ratings of their agreement with the eight statements (e.g. “The teachers at my school are respectful towards me.”, “When my teachers ask how I am doing, they are really interested in my answer.”) in question ST267 were scaled into the index of “Quality of student-teacher relationships” (RELATST) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Each of the eight items included in this scale had four response options (“Strongly disagree”, “Disagree”, “Agree”, “Strongly agree”). Positive values in the index indicate that the student reported quality student-teacher relationships, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Albania (Albanian), Denmark* (Danish), Finland (Finnish, Swedish), Georgia (Georgian, Azerbaijani, Russian), Hong Kong (China)* (Chinese), Japan (Japanese), Qatar (Arabic), Singapore (English), Sweden (Swedish, English), Thailand (Thai), the United Arab Emirates (English) and Viet Nam (Vietnamese).

Stress resistance

Question ST345 asked students if they agree (“Strongly disagree”, “Disagree”, “Neither agree nor disagree”, “Agree”, “Strongly agree”) with statements about a range of behaviours indicative of stress resistance (e.g. “I remain calm under stress.”, “I get nervous easily.”). Answers to the 10 questions were scaled into the index of “Stress resistance” (STRESAGR) with an average value of zero and a standard deviation of one across OECD countries. Note that this scale used a within-construct matrix sampling design. Positive values in the index indicate that the student reported stress resistance, more than did students on average across OECD countries.

Social connection to parents

Social connection to parents (SOCONPA) was measured by asking students, in question WB163, how often (“Almost never”, “Sometimes”, “Almost always”) their parents engage in various activities (e.g. “Show that they care”, “Encourage me to make my own decisions”). An index of index of “Social connection to parents” with an average of zero and a standard deviation one across OECD countries is formed by combining students’ responses. Students with positive values on this index perceived their social connection to their parents better than did students on average across OECD countries.

Students’ practices regarding online information

Question IC180 asked students their agreement (“Strongly disagree”, “Disagree”, “Agree”, “Strongly agree”) with various statements about their practices regarding online information (e.g., “When searching for information online I compare different sources.”, “I discuss the accuracy of online information with friends or other students.”). Answers to the six items were scaled into the index of “Students’ practices regarding online information” (ICTINFO) with an average value of zero and a standard deviation of one across OECD countries. Positive values in the index indicate that the student reported agreeing to these practices, more than did students on average across OECD countries.

One or more items from the scale received specific item parameters for Japan (Japanese) and Thailand (Thai).

References

[10] Avvisati, F. (2020), “The measure of socio-economic status in PISA: A review and some suggested improvements”, Large-Scale Assessments in Education, Vol. 8/1, pp. 1-37, https://doi.org/10.1186/s40536-020-00086-x.

[8] Ganzeboom, H. and D. Treiman (2003), “Three Internationally Standardised Measures for Comparative Research on Occupational Status”, in Advances in Cross-National Comparison, Springer US, Boston, MA, https://doi.org/10.1007/978-1-4419-9186-7_9.

[5] Muraki, E. (1992), “A generalized partial credit model: Application of an EM algorithm”, Applied Psychological Measurement, Vol. 16/2, pp. 159-177, https://doi.org/10.1002/j.2333-8504.1992.tb01436.x.

[4] Novick, F. (ed.) (1968), Some latent trait models and their use in inferring an examinee’s ability, Addison-Wesley, Menlo Park.

[2] OECD (2024), PISA 2022 Technical Report, PISA, OECD Publishing, Paris, https://doi.org/10.1787/01820d6d-en.

[1] OECD (2023), PISA 2022 Assessment and Analytical Framework, PISA, OECD Publishing, Paris, https://doi.org/10.1787/dfe0bf9c-en.

[3] OECD (2023), PISA 2022 Results (Volume I): The State of Learning and Equity in Education, PISA, OECD Publishing, Paris, https://doi.org/10.1787/53f23881-en.

[9] OECD (2020), PISA 2018 Technical Report, OECD publishing, Paris, https://www.oecd.org/pisa/data/pisa2018technicalreport/.

[13] OECD (n.d.), PISA 2022 Technical Report, PISA, OECD Publishing, Paris.

[12] OECD (forthcoming), The PISA 2022 Results: The State of Learning and Equity in Education, PISA, OECD Publishing, Paris.

[11] Rasch, G. (1960), Probabilistic models for some intelligence and attainment tests, Nielsen and Lydiche.

[7] UNESCO (2012), International Standard Classification of Education ISCED 2011.

[6] Warm, T. (1989), “Weighted likelihood estimation of ability in item response theory”, Psychometrika, Vol. 54/3, pp. 427-450, https://doi.org/10.1007/BF02294627.

Notes

← 1.  To keep the 2022 trend scales linked to PISA 2012 comparable, the Rasch model (Rasch, 1960[11]) was used to scale the dichotomous items, while the partial credit model (PCM) was used to scale the polytomous items, in line with the models used in PISA 2012.

← 2.  Due to missing data from the countries/economies, countries/economies were only approximately equally weighted.

← 3.  Different language versions were only analysed independently, if the version was distributed to a sample of over 150 and the sum of the weights was over 300. The sum of weights for all cases within a country/economy add up to a constant of 5 000 but varied on a scale-by-scale basis because missing responses varied across scales.

← 4.  Denmark*, Norway and Singapore did not collect data for any of the questions related to students’ responses and experiences during COVID-19 school closures.