This chapter introduces OECD work to standardise international measurement practices for key concepts that are instrumental in understanding what makes a good life. It provides a working definition of subjective well-being and its components – life evaluation, affect and eudaimonia – and establishes the importance and relevance of measuring these outcomes in official surveys. Specific updates to measurement recommendations in this edition of the Guidelines are highlighted, namely: (1) a shortened core module, (2) streamlined extended modules of each component of subjective well-being and (3) a new cross-cutting module of experimental measures and question banks for the experimental concepts. The chapter concludes with an overview of good measurement practice, touching on topics relating to survey design and methodology, question phrasing and placement, answer scale formulation, and data analysis and reporting.
OECD Guidelines on Measuring Subjective Well‑being (2025 Update)
Abstract
Subjective well-being encompasses the ways that people experience and think about their lives. It is a core component of people’s broader well-being, which the OECD has measured since 2011 using a multidimensional framework that encompasses a variety of economic, social and environmental outcomes. Subjective well-being outcomes not only are meaningfully associated with objective health, education and labour market outcomes (for example), but also shed light on trends in how people subjectively assess their experiences – which may diverge from trends in other measures of progress in meaningful ways. Incorporating subjective well-being data into countries’ processes of monitoring and benchmarking their well-being progress, and in designing and evaluating their policy programmes, provides an important complement to objective measures of economic, social and environmental progress (OECD, 2013[1]).
The first OECD Guidelines on Measuring Subjective Well-being were published in 2013 (OECD, 2013[1]), following the launch of the OECD Well-being Framework and the associated Better Life Initiative in 2011 (OECD, 2011[2]). This work was borne out of the Stiglitz-Sen-Fitoussi Commission, which centred on the necessity of moving beyond GDP when measuring and assessing societal progress (Stiglitz, Sen and Fitoussi, 2009[3]). The Commission report explicitly recommended collecting subjective well-being measures in official statistics, noting that these questions provide meaningful and valid data about key aspects of people’s quality of life. The follow-up report devoted a full chapter to subjective well-being measurement, with recommendations for next steps to improve uptake (Stiglitz, Fitoussi and Durand, 2018[4]). In 2013, few OECD countries were collecting data on subjective well-being in official statistics, and those that did were not necessarily doing so in standardised ways.
Thus, the original Guidelines sought to synthesise the existing body of evidence on subjective well-being measures in an easy-to-read format, with practical recommendations for national statistics offices and other interested data producers. The goals were four-fold: to improve the quality of subjective well-being data through recommendations on question wording and survey design, to improve the usefulness of the data by outlining methodological best practices, to increase the harmonisation of international statistics by converging practice around a core set of measures and to provide guidance to data users when analysing subjective well-being data.
Following the publication of the first iteration of the OECD Guidelines, the practice of collecting subjective well-being data in official surveys markedly increased in OECD countries. By 2023, close to 90% of OECD countries collected data on life satisfaction in nationally representative household surveys (with over 80% doing so annually), and over half included a measure of subjective well-being in national well-being initiatives (Mahoney, 2023[5]). Regular tracking of outcomes throughout the COVID-19 pandemic and the cost-of-living crisis yielded compelling insights into how aspects of subjective well-being were impacted by these shocks and how well these were able to rebound – or not – in the ensuing months and years (OECD, 2021[6]; What Works Wellbeing, 2021[7]; Perona, 2025[8]). Indeed, there has been growing attention to the importance of subjective well-being data both to understand how people in OECD countries are navigating the complex challenges facing citizens and policy makers alike, including the digital transition, increasing geopolitical instability and conflict, climate change and the transition to net zero, and to better understand the structural changes that have contributed to deteriorating outcomes for younger people.
This second edition of the OECD Guidelines on Measuring Subjective Well-being builds off the measurement recommendations and best practices published in the 2013 edition. It re-affirms the importance of measuring subjective well-being in official statistics and provides interested data producers with the information and tools they need to measure subjective well-being in a robust, well-validated and internationally comparable way. These updated Guidelines provide continuity for those data producers who took on board the recommendations outlined in the first edition, while clarifying recommendations in some areas to increase take-up, and expanding recommendations in other areas to account for developments in the evidence base in the intervening decade. More specifically, the second edition of the Guidelines makes three key changes from the first, by introducing:
A shortened core module of priority subjective well-being measures (Box 2.2).
Streamlined extended modules for each component of subjective well-being (Box 2.3 – Box 2.8).
A new, cross-cutting experimental module and experimental question banks (Box 3.1 and Table 3.1 – Table 3.3); this provides a useful resource to data producers interested in expanding their coverage of subjective well-being elements that have thus far been under-measured.
Forthcoming OECD work will provide additional recommendations for measuring subjective well-being in children and young people.
This introductory chapter defines what subjective well-being is and why it should be measured. It provides more detailed information on which aspects of these guidelines have been updated, as well as important methodological and measurement considerations that apply to the collection of all subjective well-being data, regardless of which modules or measures survey designers choose to integrate.
Chapters 2 and 3 provide specific modules of subjective well-being measures that can be integrated into existing household and time use surveys. Chapter 2 introduces a series of recommended survey modules: a core module of subjective well-being, followed by extended modules that narrow in on specific component areas (life evaluation, domain evaluation, affect, eudaimonia and population mental health); three modules for time use surveys are also included. Chapter 3 provides a cross-cutting module of experimental subjective well-being measures, followed by three question banks for important topics for which the statistical evidence base is still emerging. In both chapters, each module is presented alongside detailed implementation instructions so as to provide enumerators and survey designers with the contextual information needed to field these measures and interpret the resulting data in a robust way. Additional information supporting measure selection can be found in Annex A.
Subjective well-being refers to the ways in which people experience and evaluate their own lives. More specifically, the OECD defines subjective well-being as:
Good mental states, including all of the various evaluations, positive and negative, that people make of their lives and the affective reactions of people to their experiences.
This broad definition, initially put forth in the first edition of the Guidelines, includes different concepts housed under the general umbrella of subjective well-being. It provides space for cognitive evaluations that people make of their lives, for hedonic dimensions of human experience, and for concepts relating to living well, having good psychological functioning and making the most of one’s talents and capacities. This definition is distinct from well-being more broadly defined (Box 1.1). Subjective well-being is also a distinct concept in its own right; its definition does not extend to any and all perception-based indicators (for example, a household’s assessment of facing financial difficulties, or trust in government); nor does it encompass all self-reported indicators, some of which can still refer to objectively observable conditions (for example, self-reported income).
Well-being, as set out in the OECD’s Well-being Framework, is a multidimensional construct that encompasses the key outcomes that matter to people’s lives, spanning material conditions, quality of life, and relational and environmental aspects (Figure 1.1). Subjective well-being, then, is a component of broader well-being, to be assessed alongside other measures including income, health, knowledge and skills, safety, environmental quality and social connections.
In addition to serving as a tool to benchmark country progress and shape policy analysis, the OECD’s Well-being Framework also highlights gaps in the evidence base, thereby guiding OECD efforts to improve the availability, frequency and cross-country comparability of key concepts relevant to well-being. The 2013 subjective well-being publication was the first of the OECD’s measurement guidelines, but in the years since, the organisation has published recommendations for official data producers on many other topics, including: micro-statistics on household wealth (OECD, 2013[9]); the distribution of household income, consumption and wealth (OECD, 2013[10]); trust (OECD, 2017[11]); the quality of the working environment (OECD, 2017[12]); and population mental health (OECD, 2023[13]) (refer to Box A A.1 for a discussion on how mental health outcomes relate to subjective well-being). Forthcoming work will focus on social connections (see Box A A.2 for more details on how this work relates to subjective well-being measurement).
Source: OECD (2024[14]), How's Life? 2024: Well-being and Resilience in Times of Crisis, OECD Publishing, Paris, https://doi.org/10.1787/90ba854a-en.
To help clarify how the different components of subjective well-being relate to one another, and the underlying factors that shape these outcomes, the original guidelines proposed a conceptual framework of subjective well-being (Figure 1.2). Subjective well-being contains three distinct components: life evaluation, affect and eudaimonia. Each component has intrinsic value, and approaches to subjective well-being measurement should ideally capture all three.
Life evaluation is a reflective assessment on a person’s life or some specific aspect of it.
Affect refers to a person’s feelings or emotional states, typically measured with reference to a particular point in time.
Eudaimonia is broadly defined as a feeling that one is living well, and pertains to whether individuals perceive that the things they do in life are worthwhile and have meaning, and whether they feel competent and autonomous and have a sense of personal growth and self-acceptance.
Note: The constructs listed within each component of subjective well-being are illustrative examples that capture different aspects of the overall concept; it is not assumed that the constructs are additive (i.e. that all constructs under life evaluation are summed to equal overall life evaluation, for example).
Source: Adapted from OECD (2013[1]), OECD Guidelines on Measuring Subjective Well-being, OECD Publishing, Paris, https://doi.org/10.1787/9789264191655-en.
Subjective well-being data can shed light on human motivations, behaviours and broader well-being outcomes that policy makers care about. For this reason, a growing number of governments, community organisations and businesses collect these data, using them to monitor trends and inform decision-making processes. Three common approaches are outlined below, but are discussed in greater detail in the original guidelines (OECD, 2013[1]), with updated examples provided in (Mahoney, 2023[5]).
To monitor trends and benchmark progress. Subjective well-being data can be used to monitor societal trends, identifying developing points of tension that more objective measures of economic and social progress – including GDP, interest rates, consumer spending, changes in greenhouse gas emissions, or trends in life expectancy – may not capture (OECD, 2024[14]). Subjective well-being measures are particularly well-suited to pick up on the combined impact of events across multiple domains of life (Delhey and Kroll, 2012[15]), and they can help identify other relevant drivers – such as interpersonal relationships, availability of leisure time and community life – that may otherwise be neglected in discussions of societal progress. Trends in life evaluation, affect and eudaimonia may reveal important societal developments, such as dissatisfaction with the status quo, that objective measures of economic performance do not pick up, and policy makers who do not measure the former may miss important signs of growing dissatisfaction or unrest (Ianchovichina, 2018[16]; Hadzi-Vaskov and Ricci, 2021[17]). Subjective well-being data can also help to shed light on the full extent of the impacts of socio-economic hardship (Ryff, 2024[18]; Morozink et al., 2025[19]).
To serve as an early warning sign for other well-being outcomes. The evidence base has established that subjective well-being outcomes today – such as life satisfaction, positive affect, hope, and meaning and purpose in life – are heavily correlated with certain objective outcomes in the future (Kaiser and Oswald, 2022[20]), in particular those relating to physical and mental health (Chida and Steptoe, 2008[21]; Cohen et al., 2006[22]; Graham and Pinto, 2019[23]; Cohen, Bavishi and Rozanski, 2016[24]; Alimujiang et al., 2019[25]), including deaths of despair (Case and Deaton, 2021[26]), but also educational and labour market achievement (Clark, 2001[27]; Clark, Georgellis and Sanfey, 1999[28]; Graham and Pinto, 2021[29]; Martikainen et al., 2022[30]; O’Connor, 2020[31]) and voting behaviours (Algan, Blanc and Senik, 2025[32]; Ward et al., 2021[33]; Ward, 2019[34]). For policy makers concerned about providing cost-effective services to their constituents, and to be able to do so in a sustainable and long-term way, subjective well-being data provide an insight into risks and opportunities for the future health and financial well-being of their population (Blanchflower and Oswald, 2020[35]; Kim, Strecher and Ryff, 2014[36]).
To support policy design and evaluation. Over half of OECD member states have integrated subjective well-being measures into their national well-being initiatives, which in turn have been used to inform strategic planning and performance frameworks, shape new institutional structures and serve as an input to budgeting processes (including as a way of assessing budget proposals on a broader range of outcomes) (OECD, 2023[37]; Durand and Exton, 2019[38]; Exton and Shinwell, 2018[39]). Subjective well-being data have also been used to supplement cost-benefit (CBA) and cost-effectiveness (CEA) analyses, when both designing and evaluating policies (HM Treasury, 2021[40]; Wright, Peasgood and MacLennan, 2017[41]; Clark et al., 2019[42]; Murtin et al., 2017[43]). Guidelines on the use of subjective well-being data have also been issued for businesses, charities and social enterprises interested in measuring the broader impacts of their programmes (Hey, 2018[44]; Measure Wellbeing, 2018[45]; Siegerink and Murtin, 2024[46]).
The process of updating the OECD Guidelines on Measuring Subjective Well-being began in 2023, ten years following the publication of the first edition, with a scoping exercise designed to understand the extent to which those recommendations had been adopted by official data producers in OECD countries. The goal was to understand where measurement recommendations had been accepted and implemented, and where additional guidance from the OECD could be helpful in improving the take-up, usefulness and policy relevance of selected measures. Furthermore, the decade since the publication of the first edition of the Guidelines saw a large increase in the number of academic publications centred on subjective well-being data – covering the methods for measuring subjective well-being, its drivers, and its impact on a variety of other well-being outcomes, with implications for policy. The resulting working paper found that country practices had converged around a standardised measure of life satisfaction, but differences persisted in the measures used to capture affect and in the constructs measured to assess eudaimonia (Mahoney, 2023[5]). In reviewing advances in the literature, the same paper also sought both to ensure that the methodological guidance in the first edition remained robust and to understand whether new research had identified conceptual gaps or areas where international guidance could be strengthened.
The working paper set out a streamlined set of research objectives for the updated guidelines:
1. Refine measurement of affect and clarify its relationship to population mental health measures.
2. Seek a clearer definition of, and more meaningful measures for, eudaimonia.
3. Explore more globally inclusive approaches to measurement.
4. Draft measurement recommendations specific to children.
Objectives one through three are covered in this publication and therefore are focused on lessons relevant to surveying adult populations; objective four will be covered in a separate, stand-alone future OECD publication.
To inform the recommendations included in this update, three technical working papers were commissioned from external subject matter experts, to provide an in-depth focus on each of the selected research questions. Each paper conducts a detailed investigation into theoretical frameworks, definitions, and statistical properties and the policy relevance of the target construct. Interested readers can refer to the following supporting documents for more details on each:
Kudrna, L. et al. (2024[47]), “Measuring affective components of subjective well-being: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 31, https://doi.org/10.1787/6c72da70-en.
Abdallah, S. and J. Mahoney (2024[48]), “Measuring eudaimonic components of subjective well-being: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 30, https://doi.org/10.1787/667fbe08-en.
Smith, C. et al. (2025[49]), “Globally inclusive measures of subjective wellbeing: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 35, https://doi.org/10.1787/bd72752a-en.
The findings of these independent research papers provided a starting point for reflecting on the topics addressed by this update, which has also been guided by an informal advisory group (see the Foreword) and further research by the OECD Secretariat, and ultimately by the views expressed by the OECD’s Committee on Statistics and Statistical Policy. Annex A provides additional details on procedural matters relating to the update, as well as explanatory information describing how and why measures in Chapters 2 and 3 were selected for inclusion.
The first edition of the Guidelines synthesised evidence from a wide array of sources to summarise the validity and statistical quality of subjective well-being measures, as well as information on how survey design can affect subjective well-being measures – thereby outlining good practice in survey design and methodology. A short summary of the main findings of this analysis are listed below; interested readers can refer to the 2013 Guidelines for more detailed discussions and specific references, which remain valid. In particular, Chapter 1 of the 2013 Guidelines reviews the reliability and (face, convergent and construct) validity of subjective well-being measures. Chapter 2 overviews question construction; response formats; question context, placement and order effects; mode effects and survey context; and response styles, including cultural contexts. Chapter 3 covers questionnaire design, including suggestions for additional covariates to collect; target population; duration of enumeration; sample size; survey implementation and recommendations for enumerators; and data processing. Finally, Chapter 4 concludes with a discussion on how to report and analyse subjective well-being data (OECD, 2013[1]). While the below sections provide relevant information for subjective well-being data collection in general, Chapters 2 and 3 of this report provide the exact questions and answer scales to use, alongside detailed implementation instructions. Survey designers should field the questions as they appear, without making changes to the phrasing or the response scale.
Survey vehicles and covariates
To enable the monitoring and benchmarking of societal progress, subjective well-being modules are particularly relevant for household surveys. For example, general social surveys are particularly useful vehicles (Fleischer, Smith and Viac, 2016[50]), since they typically already include a range of other well-being covariates (see the bulleted list below), but subjective well-being measures can also be valuable additions to thematic surveys, while varying the specific modules and measures included depending on the ultimate goal of the corresponding survey: labour force surveys, health surveys, social inclusion surveys, financial inclusion surveys, local area surveys, etc. Time use surveys are well suited to collect data on affect, in particular, since the most accurate affect measures are those that ask respondents to report their experiences over a short (e.g. 24 hour) recall period, and it can be useful to build a picture of how people’s affective experiences interact with their time use.
The exact list of covariates that should be measured alongside the subjective well-being module will vary according to the specific needs of the data producers, survey space and the anticipated research question. However, the following is recommended for collection with subjective well-being data in most household surveys:
Demographics: Age, gender, relationship status (legal or social marital status), family composition, number of children, household size, geographic information, migrant status, disability status and – for those data producers already collecting this type of information – data on sexual orientation, race, ethnicity and/or inclusion in a minority group
Material conditions: Household income, consumption, material deprivation, housing quality, employment status
Quality of life: Physical and mental health status (refer to Box 2.7 and Box A A.1 for details on the latter), outcomes related to respondent education and skills, environmental quality and personal security
Communal relationships: Respondent work-life balance and civic engagement and governance, as well as social connections (refer to Box A A.2)
Time use surveys enable data producers to capture fine-grained detail about how people spend their time and with whom. When affect measure(s) are incorporated into time use diaries, it advances an understanding of how people felt while engaging in their activities. In addition to measuring activities (both primary and simultaneous / secondary activities for a given period of time), the following covariates (sometimes referred to as contextual information in time use surveys) should be measured:
Time frame: Time of day and day of the week.
Location: The location of the respondent during each activity (e.g. home, work, school); if the respondent is in transit, the mode of transportation can be specified (e.g. walking, car, bus).
With whom: The other people with the respondent while the activity was ongoing, which advances understanding of the amount of time spent with others vs. spent alone. This question is sometimes divided into two, with the first asking who was with the respondent and the second clarifying who was present, but not participating in the activity (UNSD, 2025[51]).
The OECD’s Well-being Framework (Box 1.1) provides a useful structure for identifying both some of the key living conditions that shape subjective well-being, as well as some of the life outcomes that evidence increasingly shows may in turn be shaped by subjective well-being. Concurrently, the OECD Well-being Database collates internationally comparable data from member states spanning each of the topics listed above; detailed information about the specific measures used to capture information on material conditions, quality of life and communal relationships can be found in the database’s metadata document (OECD, 2020[52]).
Target population
The recommendations in this report are designed for the late adolescent and adult population. Many social surveys begin surveying respondents from age 15 or 16 onwards. Specific measurement recommendations for children will be outlined in a forthcoming, companion publication. Regardless of which adult is surveyed in the household, the sampling frame of the survey must ensure a representative sample of individuals can be reported on. Proxy responses – i.e. an individual answering about the subjective well-being of another person in the household – are not appropriate. Subjective well-being data should only be collected directly from individuals themselves.
Frequency and duration of enumeration
The core module of subjective well-being measures (see Box 2.2) should be collected on an annual basis at minimum. If possible, more frequent data collection is recommended. Some OECD national statistical offices have conducted quarterly subjective well-being data collection, which yielded telling insights during the COVID-19 pandemic, the cost of living crisis and in time of geopolitical tensions (OECD, 2021[6]; What Works Wellbeing, 2021[7]; Perona, 2025[8]). Beyond tracking the impact of large shocks, quarterly collection allows subjective well-being data to be reported alongside other quarterly indicators of growth, such as GDP or employment rates (Ametepe et al., 2024[53]).
In an ideal scenario, enumeration of a given survey (i.e. the duration of time the survey is out in the field) would take place over a full year, including all days of the week including holidays. This would remove the biasing effects of seasonality and of certain days of the week, which can influence affect measures in particular due to their shorter recall period (“yesterday” framing, refer to Box 2.5 for more details). However, this is often not feasible for a number of practical reasons. At minimum, enumeration should take place during the same time period each year (to hold seasonal and other effects constant) and be proportionately spread over all days of the week, to the extent possible.
Sample size
Large sample sizes of at least 2 000 respondents (and ideally 5 000 or more) are needed when fielding subjective well-being questions in order to reduce the standard errors of estimates and allow for precise estimates that enable cross-tabulations to analyse and compare outcomes across population sub-groups. Large sample sizes are particularly important for affect modules that use a “yesterday” framing (refer to Box 2.5 for more details). In addition, more heterogenous populations require larger sample sizes to ensure outcomes for population groups can be reliably estimated. Sampling size and methodology should be made available for data users.
Survey mode
The first edition of the Guidelines noted that computer-assisted personal interviewing (CAPI) with show cards is considered the ideal mode for collecting data on subjective well-being (show cards that include verbal labels for scale end points are particularly useful when the meaning of the scale and end points change across questions; without show cards, the cognitive burden on respondents can increase). Additionally, some subjective well-being questions touch on sensitive subjects, which means that confidentiality – and ensuring the respondent is in a private space – is important; this can be more difficult to assess when conducting interviews over the phone. However, in the years since the first edition, more national statistical offices have moved to mixed-mode methods for practical (and cost saving) reasons, and the shift to web collection has accelerated in the years since the onset of the COVID-19 pandemic.
Research published in recent years has corroborated findings from the first edition of the Guidelines that subjective well-being outcomes are higher, on average, for respondents answering over the phone compared to those answering in-person (Dolan and Kavetsos, 2016[54]). However, existing evidence as to the effects of web-based survey modes reveals less definitive conclusions. When comparing web-based to in-person or phone-based surveys, new research shows that responses are consistently lower for web-based surveys (Piccitto, Liefbroer and Emery, 2022[55]; Wavrock, Schellenberg and Boulet, 2023[56]), although the magnitude of the differences may not be large, or even statistically significant (Sarracino, Rililo and Mikucka, 2017[57]). On-going investigation into web-based mode effects will help to strengthen and/or corroborate these findings, and provide stronger steers as to good practice moving forward.
Mixed-mode surveys have become more common over the past decade and have since been established as good survey practice – embodying a resource-efficient way to collect household statistics (Eurostat, 2022[58]; Carletto et al., 2022[59]; INSEE, 2022[60]). The COVID-19 pandemic also revealed vulnerabilities in leaning on a single data collection mode (Luiten et al., 2022[61]): for example, data producers relying exclusively on in-person interview methodologies had to quickly create entirely new remote surveying protocols in the face of mandatory social distancing and confinement policies, while those already employing mixed-mode methods could more easily pivot. New publications provide national statistics offices with guidance on how to effectively implement mixed-mode surveys (Schouten et al., 2023[62]).
Whenever possible, national statistical offices and other official data producers are encouraged to collect sufficient information to be able to estimate the impact of mode effects for subjective well-being modules, and the results should be published. The mode used to collect data should always be reported alongside results. When using mixed modes, it is important that questions and response formats are as similar as possible across modes to ensure data comparability. Subjective well-being modules and answer scales in this publication have been designed to work in a variety of formats, including in-person, telephone and web-based surveys.
Question placement
Question placement, and resulting contextual clues, can potentially lead to priming effects that influence responses; for this reason, subjective well-being items should be placed near the start of surveys to minimise these effects. Ideally, the core module (Box 2.2) should be asked immediately following initial screening questions that lead to a respondent’s inclusion in the survey. Where this is not practical – for example, if a survey module ordering is not modifiable, or if subjective well-being measures have historically been included as a part of a larger module of other well-being outcomes and data producers do not want to change the ordering for fear of introducing new order effects – then the most vital recommendation is that subjective well-being measures should not immediately follow questions that may elicit strong emotional responses, or that respondents might use to interpret how to respond to subjective well-being questions. Prominent examples include questions about income, political beliefs, employment status, victimisation, discrimination or anything that alludes to social rankings. The best questions to precede subjective well-being modules are relatively neutral demographic questions, such as age, gender or household composition.
Enumerators or survey designers are recommended to use introductory text at the start of the subjective well-being module to distinguish between question topics. This text – whether stated verbally by an enumerator or presented as text in a self-led survey – can serve as a buffer between subjective well-being items and other sensitive questions. Each of the modules in Chapters 2 and 3 contain introductory text.
Question order is also important within subjective well-being modules. It is recommended to flow from general to specific questions and to use consistent ordering of individual affect items (alternating between positive and negative items) to reduce the risk that asking either a positive or negative measure first influences subsequent responses. The measures in the recommended modules in Chapters 2 and 3 are designed to be asked in the order in which they appear.
Choice of measures, question wording and answer scales
Question wording matters, and for data to be comparable measures should use standardised phrasing and answer scales. Chapters 2 and 3 provide specific measure recommendations; it is advised to use the exact phrasing and answer formats provided. For data producers currently fielding different iterations of subjective well-being questions, it is recommended that future changes – implemented in order to align with OECD recommendations – be phased in, ideally using parallel split samples, so that the effects of the change can be understood. This will provide data producers with needed information to determine how to address disruptions to the time series and to adjust previous data sets if needed.
For the majority of the subjective well-being questions recommended in these guidelines, the OECD recommends using a 0-to-10 point numerical scale, anchored by verbal labels which represent conceptual absolutes: e.g. not at all satisfied / completely satisfied. To obtain more comparable results both between people and in particular across countries, it is advised to label scale interval points – between the anchor points – with numerical (2, 3, 4, etc.), rather than verbal or written (somewhat satisfied, somewhat dissatisfied, etc.) labels, since the latter can prove particularly difficult to translate consistently and may be more susceptible to different interpretations by different people. 0 to 10 numerical scales enable nuanced responses with high analytic value for data users, and (compared to labelled Likert scales) they more strongly prime respondents to consider the scale intervals as equal in size. The order of response categories may be particularly important in the case of phone-based surveys, in which each response category is given a verbal label and no visual aid can be provided. This poses less of a problem for numerical scales – as opposed to Likert scales – and the consistent presentation of the scale from lowest (0) to highest (10) values can reduce respondent burden.
Unipolar scales (containing a continuous scale focused on a single dimension) are recommended for affect measures: for example, anchored by never/not at all at one end and all the time/completely on the other. Conceptually, it is cleaner to measure affect separately, rather than combining positive and negative affective states into a single bipolar continuum: for example, very sad to very happy. Data on negative vs. positive affective states should be measured separately, as it is possible for a respondent to report feeling both emotions strongly (or not) over the course of a 24-hour recall period.
The length of the reference period used in subjective well-being questions is particularly critical for measures of affect. When trying to understand affect as it was experienced, as opposed to how a respondent feels in general (an evaluative measure), or how they remember feeling over a longer period of time (which can be subject to recall bias), reports over a period of 24 hours or less (in the case of time use diaries) are recommended. Conversely, mental health screening tools (see Box 2.7) use a longer reference period – typically the past two, or four weeks. These measures are designed to pick up the persistent patterns of affect over a sustained time period, to thereby identify respondents at risk for poor mental health (that is, most everyone experiences momentary feelings of sadness or worry, the signal for poor mental health is very high frequency and/or intensity of such feelings over a prolonged period of time – refer to Box A A.1 for an extended discussion on mental health screening tools vs. affect measures).
Lastly, when cleaning and reporting data from a longer series of items using the same response scale, it is necessary to screen for response sets (when a respondent provides identical responses to a series of items, regardless of the content of the question). These are most visible when the respondent scores at the top or bottom of the scale for all measures (regardless of the measures’ direction – e.g. positive or negative valence). Response sets may indicate either a lack of understanding on the part of the respondent, boredom, disengagement or an unwillingness to respond meaningfully. The first best guard against response sets is good survey design, which includes strong consideration for respondent burden and experience as well as keeping batteries of items as short as possible (a practice used consistently throughout the question modules that follow), coupled with transition text that provides a break between items with similar response scales. This procedure cannot correct for the more subtle influences of response sets/social desirability biases, but it can mitigate risks.
Translation
Given the importance of question wording on how people interpret, understand and respond to subjective well-being questions, high quality translations into local languages are critical. Translations should capture the construct the question is designed to measure. A robust translation process, including back translation, is essential. New measures introduced in this edition of the Guidelines have, to the extent possible, been tested for translatability and cross-cultural adaptability.
Interviewer training
When subjective well-being data are included in interviewer-led surveys, enumerators should be well-briefed, both on the constructs that the recommended modules aim to capture (so they can answer clarifying questions from respondents) but also on how the collected information will be used and why it is valuable. This enables enumerators to build rapport with interviewees, which diminishes non-response and refusal rates and enhances data quality.
Collecting subjective well-being data in minority populations
The subjective well-being measures and modules included in this publication are designed as far as possible to be relevant to all populations and cultures, across the globe. To further this, in the process of conducting this update and ensuring global inclusivity of the ensuing recommendations, measurement tools and well-being frameworks developed by, or in collaboration with, Indigenous groups and minority communities were reviewed. Indeed, official data producers in many OECD countries have developed well-being surveys tailored to specific communities.
When collecting subjective well-being measures in general, but in particular for minority populations, data producers should be mindful of the following guiding principles (refer to Box A A.3 for an extended discussion of each):
Foster community involvement. Building relationships and trust with the community is important; participatory research methodologies are recommended in that they elevate community voices, priorities and values.
Take a strengths-based approach. There should not be a heavy or unbalanced focus on well-being deprivations or too much emphasis on values that may not be as relevant to the local community (i.e. individual achievement) while devaluing aspects that may be important in other cultural contexts (spirituality, cultural vitality, connection to land). Instead, surveys should seek to find a balance between capturing the resilience, capabilities and assets of communities – as well as vulnerabilities, without too heavy a focus on deficits and problems.
Develop local ethical guidelines. Research practices should be both scientifically sound as well as culturally appropriate and equitable. Attitudes towards privacy, informed consent and the role of community leaders should be tailored to the local context.
Ensure data sovereignty. The local community should govern their data and autonomously make decisions about methods, management and dissemination. Community consultations can help establish these norms.
Reporting central tendency, level and distribution of subjective well-being measures
Summary statistics of central tendency include mean, median and mode: each provides a different and useful way of presenting population averages and comparing levels of subjective well-being across population groups using a single value. Beyond averages, different approaches to reporting the distribution of subjective well-being outcomes highlight inequalities in outcomes across the top and bottom of the distribution and across different population groups. OECD recommendations for reporting the central tendency, level and distribution of subjective well-being data are outlined in Box 1.2.
Use the mean to report average levels of subjective well-being. Given the limited number of scale categories (eleven discrete options, when using a 0-10 scale), the median and mode are less sensitive to changes over time or between population groups, in comparison to the mean. When possible, report both the mean as well as the share of the population below a stated threshold. Specific thresholds are described in the implementation details accompanying each module in Chapters 2 and 3; details on thresholds can also be found in the OECD’s Well-being Database metadata document (OECD, 2020[52]).
When reporting deprivations, it is recommended to check first that the change in the share of the population above and below the threshold paints a consistent picture with that of changes in the distribution as a whole. This is because thresholds can create artificial data cliffs that under- or over-estimate the overall change by focusing on just one cut-point in the distribution. There can also be significant changes in the data on either side of a threshold that are worthy of note, even if the total share of the population above and below a given threshold remains stable.
Summary measures of distribution (also described as “vertical inequalities”) are also of high policy value, and trends in such summary measures should be routinely monitored alongside mean average scores (refer to (OECD, 2017, pp. 68-71[63]) for a detailed discussion of different inequality types and how to measure them in the context of well-being metrics). To provide a summary of the distribution of responses, it is recommended to report the ratio between the top 20% (the quintile with the highest subjective well-being outcomes) and the bottom 20% (the quintile with the lowest). This methodology is used in the OECD Well-being Database (OECD, 2020[52]) and in the accompanying How’s Life? series of publications (OECD, 2024[14]) when reporting on vertical inequalities. To provide a more complete view of the distribution of subjective well-being outcomes, it is also possible to present the share of the population who chose each discrete answer option on the 0 to 10 scale.
Reporting of “horizontal inequalities”, i.e. average results for different population groups, are also of high policy value, especially where time series can illustrate whether gaps in outcomes are narrowing or widening over time. Demographic covariates included in the survey (see above) together with sample sizes will determine the data disaggregations that are possible to report, with age, gender and education breakdowns being a minimum baseline. Additional groups can be considered, should sample sizes permit. Differences in outcomes for vulnerable population groups (including groups at risk of discrimination) can be of particularly high policy relevance. When reporting horizontal inequalities, especially for relatively small populations, care should be taken to assess the statistical significance of differences in outcomes between groups, rather than making simple comparisons between point estimates.
Analysis of subjective well-being data
The first edition of the Guidelines contained a full chapter outlining different approaches to analysing subjective well-being data, with different recommendations depending on the ultimate goals of the analytical exercise. Interested readers can refer to Chapter 4 of the 2013 Guidelines for a detailed discussion of the following topics:
Reporting on subjective well-being summary statistics: how to interpret changes over time and differences between population groups, and how to handle cross-country comparisons given concerns around cultural differences in question interpretation and answer patterns.
How to analyse the drivers and determinants of subjective well-being, and how to manage issues with shared method variance, omitted variables, reverse causality, frame and reference effects, and hedonic adaptation.
Integrating subjective well-being data into policy processes, including designing, implementing and evaluating policy interventions. Updated examples can also be found in Mahoney (2023[5]).
References
[48] Abdallah, S. and J. Mahoney (2024), “Measuring eudaimonic components of subjective well-being: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 30, OECD Publishing, Paris, https://doi.org/10.1787/667fbe08-en.
[32] Algan, Y., C. Blanc and C. Senik (2025), “Trusting others: How unhappiness and social distrust explain populism”, in Helliwell, J. et al. (eds.), The World Happiness Report 2025, University of Oxford: Wellbeing Research Centre, https://worldhappiness.report/ed/2025/trusting-others-how-unhappiness-and-social-distrust-explain-populism/.
[25] Alimujiang, A. et al. (2019), “Association between life purpose and mortality among US adults older than 50 years”, JAMA Network Open, Vol. 2/5, https://doi.org/10.1001/JAMANETWORKOPEN.2019.4270.
[53] Ametepe, F. et al. (2024), PIBien-être 2023 Report, Institut national de la statistique et des études économiques (STATEC), https://statistiques.public.lu/fr/publications/series/analyses/2024/analyses-01-24.html.
[35] Blanchflower, D. and A. Oswald (2020), “Trends in extreme distress in the United States, 1993–2019”, American Journal of Public Health, Vol. 110/10, pp. 1538-1544, https://doi.org/10.2105/AJPH.2020.305811.
[59] Carletto, C. et al. (2022), “Positioning household surveys for the next decade”, Statistical Journal of the IAOS, Vol. 38/3, pp. 923-946, https://content.iospress.com/articles/statistical-journal-of-the-iaos/sji220042.
[26] Case, A. and A. Deaton (2021), “The great divide: Education, despair and death”, NBER Working Papers, No. 29241, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/W29241.
[21] Chida, Y. and A. Steptoe (2008), “Positive psychological well-being and mortality: A quantitative review of prospective observational studies”, Psychosomatic Medicine, Vol. 70/7, pp. 741-756, https://doi.org/10.1097/PSY.0B013E31818105BA.
[27] Clark, A. (2001), “What really matters in a job? Hedonic measurement using quit data”, Labour Economics, Vol. 8/2, pp. 223-242, https://doi.org/10.1016/S0927-5371(01)00031-8.
[42] Clark, A. et al. (eds.) (2019), The Origins of Happiness: The Science of Well-being over the Life Course, Princeton University Press, Princeton, NJ, https://books.google.com/books?hl=en&lr=&id=6ciXDwAAQBAJ&oi=fnd&pg=PP9&dq=the+origins+of+happiness&ots=ZKdX_U8pjm&sig=xRD7b95BcsQkfQfGMsURR2ZQu6Y.
[28] Clark, A., Y. Georgellis and P. Sanfey (1999), “Job satisfaction, wage changes and Quits: Evidence from Germany”, SSRN, https://papers.ssrn.com/abstract=132109.
[24] Cohen, R., C. Bavishi and A. Rozanski (2016), “Purpose in life and its relationship to all-cause mortality and cardiovascular events: A meta-analysis”, Psychosomatic Medicine, Vol. 78/2, pp. 122-133, https://doi.org/10.1097/PSY.0000000000000274.
[22] Cohen, S. et al. (2006), “Positive emotional style predicts resistance to illness after experimental exposure to rhinovirus or influenza a virus”, Psychosomatic Medicine, Vol. 68/6, pp. 809-815, https://doi.org/10.1097/01.PSY.0000245867.92364.3C.
[15] Delhey, J. and C. Kroll (2012), “Inequality and social integration a ’happiness test’ for the new measures of national well-being: How much better than GDP are they?”, Discussion Papers, Research Unit: Inequality and Social Integration, Vol. SP I 2012-201, https://ideas.repec.org/p/zbw/wzbisi/spi2012201.html.
[54] Dolan, P. and G. Kavetsos (2016), “Happy talk: Mode of administration effects on subjective well-being”, Journal of Happiness Studies, Vol. 17/3, pp. 1273-1291, https://doi.org/10.1007/s10902-015-9642-8.
[38] Durand, M. and C. Exton (2019), “Adopting a well-being approach in central government: Policy mechanisms and practical tools”, in Global Happiness Policy Report 2019, Sustainable Development Solutions Network, New York, http://ghwbpr-2019.s3.amazonaws.com/UAE/GH19_Ch8.pdf.
[58] Eurostat (2022), Position Paper on Mixed-Mode Surveys — 2022 edition, Publications Office of the European Union, Luxembourg, https://ec.europa.eu/eurostat/fr/web/products-statistical-working-papers/-/ks-tc-21-009.
[39] Exton, C. and M. Shinwell (2018), “Policy use of well-being metrics: Describing countries’ experiences”, OECD Statistics Working Papers, No. 2018/07, OECD Publishing, Paris, https://doi.org/10.1787/d98eb8ed-en.
[50] Fleischer, L., C. Smith and C. Viac (2016), “A review of general social surveys”, OECD Statistics Working Papers, No. 2016/9, OECD Publishing, Paris, https://doi.org/10.1787/bb54d16f-en.
[29] Graham, C. and S. Pinto (2021), “The geography of desperation in America: Labor force participation, mobility, place, and well-being”, Social Science & Medicine, Vol. 270, p. 113612, https://doi.org/10.1016/J.SOCSCIMED.2020.113612.
[23] Graham, C. and S. Pinto (2019), “Unequal hopes and lives in the USA: Optimism, race, place, and premature mortality”, Journal of Population Economics, Vol. 32/2, pp. 665-733, https://www.jstor.org/stable/48699052.
[17] Hadzi-Vaskov, M. and L. Ricci (2021), “Understanding Chile’s social unrest in an international perspective”, IMF Working Paper, No. 21/174, International Monetary Fund, https://www.imf.org/en/Publications/WP/Issues/2021/06/25/Understanding-Chiles-Social-Unrest-in-an-International-Perspective-461279.
[44] Hey, N. (2018), Improve wellbeing with data infrastructure: Seven actions for organisations, What Works Wellbeing, https://whatworkswellbeing.org/blog/improve-wellbeing-with-data-infrastructure-seven-actions-for-organisations/.
[40] HM Treasury (2021), Wellbeing Guidance for Appraisal: Supplementary Green Book Guidance, Social Impacts Task Force, https://www.gov.uk/government/publications/green-book-supplementary-guidance-wellbeing.
[16] Ianchovichina, E. (2018), “Eruptions of Popular Anger: The Economics of the Arab Spring and Its Aftermath”, MENA Development Report, World Bank, Washington, DC, https://documents1.worldbank.org/curated/en/251971512654536291/pdf/121942-REVISED-Eruptions-of-Popular-Anger-preliminary-rev.pdf (accessed on 22 March 2023).
[60] INSEE (2022), Le multimode dans les enquêtes auprès des ménages : une collecte modernisée, un processus complexifié − Courrier des statistiques N7 - 2022, Institut national de la statistique et des études économiques (INSEE), https://www.insee.fr/fr/information/6035934?sommaire=6035950.
[20] Kaiser, C. and A. Oswald (2022), “The scientific value of numerical measures of human feelings”, PNAS, Vol. 119/42, https://doi.org/10.1073/pnas.2210412119.
[36] Kim, E., V. Strecher and C. Ryff (2014), “Purpose in life and use of preventive health care services”, Proceedings of the National Academy of Sciences of the United States of America, Vol. 111/46, pp. 16331-16336, https://doi.org/10.1073/PNAS.1414826111.
[47] Kudrna, L. et al. (2024), “Measuring affective components of subjective well-being: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 31, OECD Publishing, Paris, https://doi.org/10.1787/6c72da70-en.
[61] Luiten, A. et al. (2022), Mixed-Mode Official Surveys: Design and Analysis, Centraal Bureau voor de Statistiek (CBS), https://unstats.un.org/iswghs/events/mixed-mode-webinar/UN-WG-Household-Surveys-webinar-17-11-22.pdf.
[5] Mahoney, J. (2023), “Subjective well-being measurement: Current practice and new frontiers”, OECD Papers on Well-being and Inequalities, Vol. 17, https://doi.org/10.1787/4e180f51-en.
[30] Martikainen, S. et al. (2022), “Moments of meaningfulness and meaninglessness: A qualitative inquiry into affective eudaimonia at work”, Article Group & Organization Management, Vol. 47/6, pp. 1135-1180, https://doi.org/10.1177/10596011211047324.
[45] Measure Wellbeing (2018), Evaluating Wellbeing – Better Evaluations for Social Change, Nottingham Trent University, https://measure-wellbeing.org/.
[19] Morozink, J. et al. (2025), “Educational disparities in 20-year trajectories of psychological well-being in a national sample”, Scientific Reports, Vol. 15/1, pp. 1-12, https://doi.org/10.1038/s41598-025-95058-y.
[43] Murtin, F. et al. (2017), “Beyond GDP: Is there a law of one shadow price?”, European Economic Review, Vol. 100, pp. 390-411, https://doi.org/10.1016/J.EUROECOREV.2017.09.001.
[31] O’Connor, K. (2020), “Life satisfaction and noncognitive skills: Effects on the likelihood of unemployment”, Kyklos, Vol. 73/4, pp. 568-604, https://doi.org/10.1111/KYKL.12226.
[14] OECD (2024), How’s Life? 2024: Well-being and Resilience in Times of Crisis, OECD Publishing, Paris, https://doi.org/10.1787/90ba854a-en.
[37] OECD (2023), Economic Policy Making to Pursue Economic Welfare: Report Prepared for the 2023 Japan Presidency of the G7, OECD Publishing, Paris, https://www.oecd.org/economy/G7_Beyond_GDP_Economic_policy_making_to_pursue_economic_welfare_2023.pdf.
[13] OECD (2023), Measuring Population Mental Health, OECD Publishing, Paris, https://doi.org/10.1787/5171eef8-en.
[6] OECD (2021), COVID-19 and Well-being: Life in the Pandemic, OECD Publishing, Paris, https://doi.org/10.1787/1e1ecb53-en.
[52] OECD (2020), How’s Life? 2020: Measuring Well-being, OECD Publishing, Paris, https://doi.org/10.1787/9870c393-en.
[63] OECD (2017), How’s Life? 2017: Measuring Well-being, OECD Publishing, Paris, https://doi.org/10.1787/how_life-2017-en.
[12] OECD (2017), OECD Guidelines on Measuring the Quality of the Working Environment, OECD Publishing, Paris, https://doi.org/10.1787/9789264278240-en.
[11] OECD (2017), OECD Guidelines on Measuring Trust, OECD Publishing, Paris, https://doi.org/10.1787/9789264278219-en.
[10] OECD (2013), OECD Framework for Statistics on the Distribution of Household Income, Consumption and Wealth, OECD Publishing, Paris, https://doi.org/10.1787/9789264194830-en.
[9] OECD (2013), OECD Guidelines for Micro Statistics on Household Wealth, OECD Publishing, Paris, https://doi.org/10.1787/9789264194878-en.
[1] OECD (2013), OECD Guidelines on Measuring Subjective Well-being, OECD Publishing, Paris, https://doi.org/10.1787/9789264191655-en.
[2] OECD (2011), How’s Life?: Measuring Well-being, OECD Publishing, Paris, https://doi.org/10.1787/9789264121164-en.
[8] Perona, M. (2025), Le Bien-être des Français, Mars 2025, CEntre Pour la Recherche EconoMique et ses APplications (CEPREMAP), https://www.cepremap.fr/2025/04/le-bien-etre-des-francais-mars-2025/.
[55] Piccitto, G., A. Liefbroer and T. Emery (2022), “Does the survey mode affect the association between subjective well-being and its determinants? An experimental comparison between face-to-face and web mode”, Journal of Happiness Studies, Vol. 23/7, pp. 3441-3461, https://doi.org/10.1007/S10902-022-00553-Y.
[18] Ryff, C. (2024), “The privilege of well-being in an increasingly unequal society”, Current Directions in Psychological Science, https://doi.org/10.1177/09637214241266818.
[57] Sarracino, F., C. Rililo and M. Mikucka (2017), “Comparability of web and telephone survey modes for the measurement of subjective well-being”, Survey Research Methods : SRM, Vol. 11/2, pp. 141-169, https://madoc.bib.uni-mannheim.de/43970/.
[62] Schouten, B. et al. (2023), Mixed-mode official surveys: Design and analysis, Chapman & Hall, https://www.routledge.com/Mixed-Mode-Official-Surveys-Design-and-Analysis/Schouten-vandenBrakel-Buelens-Giesen-Luiten-Meertens/p/book/9781032102962.
[46] Siegerink, V. and F. Murtin (2024), “An OECD survey of employee well-being: An instrument to measure employee well-being inside companies”, OECD Papers on Well-being and Inequalities, No. 24, OECD Publishing, Paris, https://doi.org/10.1787/74f48e24-en.
[49] Smith, C. et al. (2025), “Globally inclusive measures of subjective well-being: Updated evidence to inform national data collections”, OECD Papers on Well-being and Inequalities, No. 35, OECD Publishing, Paris, https://doi.org/10.1787/bd72752a-en.
[4] Stiglitz, J., J. Fitoussi and M. Durand (eds.) (2018), For Good Measure: Advancing Research on Well-being Metrics Beyond GDP, OECD Publishing, Paris, https://doi.org/10.1787/9789264307278-en.
[3] Stiglitz, J., A. Sen and J. Fitoussi (2009), Report by the Commission on the Measurement of Economic Performance and Social Progress, https://ec.europa.eu/eurostat/documents/8131721/8131772/Stiglitz-Sen-Fitoussi-Commission-report.pdf.
[51] UNSD (2025), Guide to Producing Statistics on Time Use, United Nations Statistics Division, https://unstats.un.org/UNSDWebsite/statcom/session_55/documents/BG-3k-Guide_to_Producing_Statistics_on_Time_Use-55UNSC_background-E.pdf.
[34] Ward, G. (2019), “Happiness and voting behavior”, in Helliwell, J., R. Layard and J. Sachs (eds.), World Happiness Report 2019, Sustainable Development Solutions Network, New York, https://worldhappiness.report/ed/2019/happiness-and-voting-behavior/.
[33] Ward, G. et al. (2021), “(Un)happiness and voting in U.S. presidential elections”, Journal of Personality and Social Psychology, Vol. 120/2, pp. 370-383, https://doi.org/10.1037/PSPI0000249.
[56] Wavrock, D., G. Schellenberg and C. Boulet (2023), Survey Framing and Mode Effects in Life Satisfaction Responses on Canadian Social Surveys, Statistics Canada, https://www150.statcan.gc.ca/n1/pub/36-28-0001/2023001/article/00003-eng.htm.
[7] What Works Wellbeing (2021), What helped the UK cope with the Covid-19 pandemic and lockdowns?, What Works Wellbeing Centre, https://whatworkswellbeing.org/resources/what-helped-the-uk-cope-with-the-covid-19-pandemic-and-lockdowns/ (accessed on 15 April 2025).
[41] Wright, L., T. Peasgood and S. MacLennan (2017), A Guide to Wellbeing Economic Evaluation, What Works Wellbeing Centre, https://whatworkswellbeing.org/resources/a-guide-to-wellbeing-economic-evaluation/.