Hubbry Logo
Longitudinal studyLongitudinal studyMain
Open search
Longitudinal study
Community hub
Longitudinal study
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Contribute something
Longitudinal study
Longitudinal study
Longitudinal study
View on Wikipedia
from Wikipedia

A longitudinal study (or longitudinal survey, or panel study) is a research design that involves repeated observations of the same variables (e.g., people) over long periods of time (i.e., uses longitudinal data). It is often a type of observational study, although it can also be structured as longitudinal randomized experiment.[1]

Longitudinal studies are often used in social-personality and clinical psychology, to study rapid fluctuations in behaviors[2], thoughts[3], and emotions from moment to moment or day to day[4][5]; in developmental psychology, to study developmental trends across the life span; and in sociology, to study life events throughout lifetimes or generations; and in consumer research and political polling to study consumer trends. The reason for this is that, unlike cross-sectional studies, in which different individuals with the same characteristics are compared,[6] longitudinal studies track the same people, and so the differences observed in those people are less likely to be the result of cultural differences across generations, that is, the cohort effect. Longitudinal studies thus make observing changes more accurate and are applied in various other fields. In medicine, the design is used to uncover predictors of certain diseases. In advertising, the design is used to identify the changes that advertising has produced in the attitudes and behaviors of those within the target audience who have seen the advertising campaign. Longitudinal studies allow social scientists to distinguish short from long-term phenomena, such as poverty. If the poverty rate is 10% at a point in time, this may mean that 10% of the population are always poor or that the whole population experiences poverty for 10% of the time.

Longitudinal studies can be retrospective (looking back in time, thus using existing data such as medical records or claims database) or prospective (requiring the collection of new data).[citation needed]

Cohort studies are one type of longitudinal study which sample a cohort (a group of people who share a defining characteristic, typically who experienced a common event in a selected period, such as birth or graduation) and perform cross-section observations at intervals through time. Not all longitudinal studies are cohort studies; some instead include a group of people who do not share a common event.[7]

As opposed to observing an entire population, a panel study follows a smaller, selected group - called a 'panel'.[8]

Advantages

[edit]

When longitudinal studies are observational, in the sense that they observe the state of the world without manipulating it, it has been argued that they may have less power to detect causal relationships than experiments. Others say that because of the repeated observation at the individual level, they have more power than cross-sectional observational studies, by virtue of being able to exclude time-invariant unobserved individual differences and also of observing the temporal order of events.[9][failed verification]

Longitudinal studies do not require large numbers of participants (as in the examples below). Qualitative longitudinal studies may include only a handful of participants,[10] and longitudinal pilot or feasibility studies often have fewer than 100 participants.[11]

Disadvantages

[edit]

Longitudinal studies are time-consuming and expensive.[12]

Longitudinal studies cannot avoid an attrition effect: that is, some subjects cannot continue to participate in the study for various reasons. Under longitudinal research methods, the reduction in the research sample will bias the remaining smaller sample.[citation needed]

Practice effect is also one of the problems: longitudinal studies tend to be influenced because subjects repeat the same procedure many times (potentially introducing autocorrelation), and this may cause their performance to improve or deteriorate.[citation needed]

Examples

[edit]
Study name Type Country or region Year started Participants Remarks
45 and Up Study Cohort Australia 2006 267,153 The 45 and Up Study is a longitudinal study of participants aged 45 years and over in New South Wales conducted by the Sax Institute. Researchers are able to analyze Study data linked to MBS and PBS data, the NSW cancer registry, State hospitalizations, and emergency department visits and mortality data.

The Study is used by both researchers and policymakers to better understand how Australians are aging and using health services to prevent and manage ill-health and disability and guide health system decisions. 45 and Up is the largest ongoing study of healthy aging in the Southern Hemisphere.

Alzheimer's Disease Neuroimaging Initiative Panel International 2004 n/a –
Australian Longitudinal Study on Women's Health (ALSWH) Cohort Australia 1996 50,000 Includes four cohorts of women: born between 1921 and 1926, 1946–1951, 1973–1978 and 1989–1995
Avon Longitudinal Study of Parents and Children (ALSPAC) Cohort United Kingdom 1991 14,000 –
Born in Bradford Cohort United Kingdom 2007 12,500 –
1970 British Cohort Study (BCS70) Cohort United Kingdom 1970 17,000 Monitors the development of babies born in the UK in one particular week in April 1970
British Doctors Study Cohort United Kingdom 1951 40,701 Monitored the health of British male doctors. It provided convincing evidence of the link between smoking and cancer.
British Household Panel Study Panel United Kingdom 1991 5,500 households (~10,000 individuals) Modeled on the US Panel Study of Income Dynamics PSID study
Building a New Life in Australia : The Longitudinal Study of Humanitarian Migrants (BNLA)[13] Cohort Australia 2013 2,399 A longitudinal study of the settlement experience of humanitarian arrivals in Australia
Busselton Health Study[14] Panel Australia 1966 10,000 –
Caerphilly Heart Disease Study Cohort United Kingdom 1979 2,512 Male subjects (Wales)
Canadian Longitudinal Study on Aging (CLSA-ÉLCV)[15] Cohort Canada 2011 51,388[16] All research participants will be followed until 2033 or death.[17]
Child Development Project[18] Cohort United States 1987 585 Follows children recruited the year before they entered kindergarten in three US cities: Nashville and Knoxville, Tennessee, and Bloomington, Indiana
Children of Immigrants Longitudinal Study (CILS) Cohort United States 1992 5,262 Florida
Congenital Heart Surgeons' Society (CHSS) Cohort Canada – 5,000 Various studies, managed by the Data Center Studies on Congenital Heart Diseases
Colombian Longitudinal Survey by Universidad de los Andes (ELCA)[19] Panel Colombia 2010 15,363[20] Follows rural and urban households for increasing the comprehension of social and economic changes in Colombia
Copenhagen General Population Study (CGPS) Cohort Denmark 1976 170,000 The study is an ongoing prospective cohort study, that investigates the epidemiology of a wide range of diseases in a representative sample of the Danish population. Now integrated with and expanding upon the earlier and less extensive sister study, the Copenhagen City Heart Study.[21][22]
Dunedin Multidisciplinary Health and Development Study Cohort New Zealand 1972 1,037 Participants born in Dunedin during 1972–73
Early Childhood Longitudinal Study (ECLS) United States
Study of migrants and squatters in Rio's Favelas Cohort Brazil 1968 n/a The work of Janice Perlman, reported in her book Favela (2014)[23]
Footprints in Time; the longitudinal study of Indigenous children[24] Cohort Australia 2008 1,680 Study of Aboriginal and Torres Strait Islander children in selected locations across Australia
Fragile Families and Child Wellbeing Study Cohort United States 1998 n/a Study being conducted in 20 cities
Framingham Heart Study Cohort United States 1948 5,209 Massachusetts
Genetic Studies of Genius Cohort United States 1921 1,528 The world's oldest and longest-running longitudinal study
Grant Study Cohort United States 1939 268 A 75-year longitudinal study of 268 physically and mentally healthy Harvard college sophomores from the classes of 1939–1944.
Growing Up in Australia; the longitudinal study of Australian children[25] Cohort Australia 2004 10,000 –
Growing Up in Ireland (GUI) Cohort Ireland 2006 8,000 children
10,000 infants 		Growing Up in Ireland is an Irish Government-funded study of children being carried out jointly by the Economic and Social Research Institute and Trinity College Dublin. The study started in 2006 and follows the progress of two groups of children: 8,000 9-year-olds (Child Cohort/Cohort '98) and 10,000 9-month-olds (Infant Cohort/Cohort '08).
Growing Up in New Zealand (GUiNZ) Cohort New Zealand 2009 6,846 children

GUiNZ is New Zealand's largest ongoing longitudinal study. It follows approximately 11% of all NZ children born between 2009 and 2010.[26] The study aims to look in depth at the health and well-being of children (and their parents) growing up in NZ.

Growing Up in Scotland (GUS) Cohort Scotland 2003 14,000 –
Health and Retirement Study Cohort United States 1988 22,000 –
Household, Income and Labour Dynamics in Australia Survey Panel Australia 2001 25,000 –
Irish Longitudinal Study on Ageing (TILDA) Cohort Ireland 2009 8,500 Studies health, social and financial circumstances of the older Irish population
The Jyväskylä Longitudinal Study of Personality and Social Development,[27] (JYLS) Cohort Finland 1968 369 The sample was drawn from 12 complete school classes. Data has been collected when the participants were 8, 14, 20, 27, 33, 36, 42 and 50 years old.
Life Paths into Early Adulthood (LifE Study) Cohort Germany 1979 3,000 Study tracks participants over an extended period to examine processes of development, education, socialization, and intergenerational transmission.
Longitudinal Study of Young People in England (Next Steps) Cohort England 2004 16,000 Large-scale panel study collecting information about young people of England aged 13 to 14 in 2004
Midlife in the United States Cohort United States 1983 6,500 –
Manitoba Follow-Up Study (MFUS) Cohort Canada 1948 3,983 men Canada's largest and longest running investigation of cardiovascular disease and successful aging
Millennium Cohort Study (MCS) Cohort United Kingdom 2000 19,000 Study of child development, social stratification, and family life
Millennium Cohort Study Cohort United States 2000 200,000 Evaluation of long-term health effects of military service, including deployments
Minnesota Twin Family Study Cohort United States 1983 17,000 (8,500 twin pairs) –
National Child Development Study (NCDS) Cohort United Kingdom 1958 17,000 –
National Educational Panel Study (NEPS) Cohort Germany 2009 60,000 Study on the development of competencies, educational processes, educational decisions, and returns to education in formal, nonformal, and informal contexts throughout the life span
National Longitudinal Surveys (NLS) Cohort United States 1979 12,686 (NLSY79),
9,000 (approx., NLSY97) 		Includes four cohorts: NLSY79 (born 1957–64), NLSY97 (born 1980–84), NLSY79 Children and Young Adults, National Longitudinal Surveys of Young Women and Mature Women (NLSW)
National Longitudinal Survey of Children and Youth (NLSCY) Cohort Canada 1994 35,795 Inactive since 2009
National Health and Nutrition Examination Survey (NHANES) Cohort United States 1971 8,837 (since 1999) Continual since 1999
Nature vs Nurture study Cohort United States 1960 11[29] Concluded in 1980. Controversial study by Peter B. Neubauer of twins and triplets separated at birth. Never published.
New Zealand Attitudes and Values Study – New Zealand 2009 n/a –
Northern Ireland Longitudinal Study (NILS)[30] Panel Northern Ireland 2006 500,000 (comprises about 28% of the Northern Ireland population and approximately 50% of households). The NILS is a large-scale, representative data-linkage study created by linking data from the Northern Ireland Health Card Registration system to 1981, 1991, 2001 and 2011 census returns and to administrative data from other sources. These include vital events registered with the General Register Office for Northern Ireland (such as births, deaths, and marriages) and the Health Card registration system migration events data. The result is a 30-year-plus longitudinal data set which is regularly being updated. In addition to this rich resource, there is also the potential to link further Health and Social care data via distinct linkage projects (DLPs). The NILS is designed for statistics and research purposes only and is managed by the Northern Ireland Statistics and Research Agency under Census legislation. The data are de-identified at the point of use; access is only from within a strictly controlled 'secure environment' and governed by protocols and procedures to ensure data confidentiality.
Nurses' Health Study Cohort United States 1976 275,000 Most expensive and largest observational health study in history
ONS Longitudinal Study[31][32] Panel England and Wales 1974 (data from 1971) 500,000 (1% sample of the population of England and Wales). The LS contains records on over 500,000 people usually resident in England and Wales at each point in time) The sample comprises people born on one of four selected dates of birth and therefore makes up about 1% of the total population. The sample was initiated at the time of the 1971 Census, and the four dates were used to update the sample at the 1981,1991, 2001 and 2011 Censuses and in routine event registrations. Fresh LS members enter the study through birth and immigration and existing members leave through death and emigration. Thus, the LS represents a continuous sample of the population of England and Wales, rather than a sample taken at a one-time point only. It now includes records for over 950,000 study members. In addition to the census records, the individual LS records contain data for events such as deaths, births to sample mothers, emigrations and cancer registrations. Census information is also included for all people living in the same household as the LS member. The LS does not follow up household members in the same way from census to census. Support for potential users and more information available at CeLSIUS
Pacific Islands Families Study Cohort New Zealand 2000 1,398 –
Panel Study of Belgian Households[33] Panel Belgium 1992 11,000[34] –
Panel Study of Income Dynamics Panel United States 1968 70,000 Possibly the oldest household longitudinal survey in the US
The Raine Study Cohort Australia 1989 5,768 (Gen1 + Gen2)
750 (Gen3)
100 (Gen0) 		The Raine Study is based in Perth, Western Australia. It has followed the same group of pregnant women (Gen1) and their babies (Gen2) who were born into the study between 1989 and 1992. Its original aim was to investigate the benefits of more frequent ultrasound scans on infant health.[35] It now studies the impact that early life factors (from the womb onwards) have on health throughout life.[36] The Raine Study now includes 4 generations of cohort members.
Rotterdam Study Cohort Netherlands 1990 15,000 Focus is on inhabitants of Ommoord, a suburb of Rotterdam
Scottish Longitudinal Study (SLS)[37] Panel Scotland 1991 274,000 (comprises 5.3% sample of the Scottish population, with records on approximately 274,000 individuals using 20 random birthdates) The SLS is a large-scale linkage study built upon census records from 1991 onwards, with links to vital events (births, deaths, marriages, emigration); geographical and ecological data (deprivation indices, pollution, weather); primary and secondary education data (attendance, Schools Census, qualifications); and links to NHS Scotland ISD datasets, including cancer registrations, maternity records, hospital admissions, prescribing data and mental health admissions. The research potential is considerable. The SLS is a replica of the ONS Longitudinal Study but with a few key differences: sample size, commencement point and the inclusion of certain variables. The SLS is supported and maintained by the SLS Development & Support Unit with a safe-setting at the National Records of Scotland in Edinburgh. Further information and support for potential users is available at SLS-DSU
Seattle 500 Study Cohort United States 1974 500 Study of the effects of prenatal health habits on human development
Socio-Economic Panel (SOEP) Panel Germany 1984 12,000 –
Stirling County Study Cohort Canada 1952 639 Long-term study epidemiology of psychiatric disorders. Two cohorts were studied (575 from 1952 to 1970; 639 from 1970 to 1992).[38]
Study of Health in Pomerania Cohort Germany 1997 15,000 Investigates common risk factors, sub-clinical disorders and manifest diseases in a high-risk population
Study of Mathematically Precocious Youth Cohort United States 1972 5,000 Follows highly intelligent people identified by age 13.
Survey of Health, Ageing, and Retirement in Europe (SHARE) Panel Europe 2002 120,000 Multidisciplinary and cross-national panel database of micro data on health, socio-economic status and social and family networks of individuals aged 50 or over
Study on Global Ageing and Adult Health (SAGE) Cohort International 2002 65,964 Studies the health and well-being of adult populations and the ageing process in six countries: China, Ghana, India, Mexico, Russian Federation and South Africa
Seattle Longitudinal Study Cohort United States 1956 6,000 [39] –
Understanding Society: The UK Household Longitudinal Study Panel United Kingdom 2009 100,000 Incorporates the British Household Panel Study
Up Series Cohort United Kingdom 1964 14 Documentary film project by Michael Apted
Wisconsin Longitudinal Study[40] Cohort United States 1957 10,317 Follows graduates from Wisconsin high schools in 1957

See also

[edit]

References

[edit]
[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Longitudinal study

View on Grokipedia
from Grokipedia
A longitudinal study is a research design that involves repeated observations of the same variables, such as exposures and outcomes, over extended periods—often years or decades—to track changes in individuals or groups. These studies are typically observational, though they can include experimental elements, and they collect quantitative or qualitative data without directly influencing participants. By following subjects over time with continuous or repeated monitoring of risk factors or health outcomes, longitudinal studies enable researchers to establish temporal sequences of events and detect patterns of change that cross-sectional designs cannot capture. Longitudinal studies encompass several types, including prospective cohort studies, where groups defined by exposure status are followed forward to observe outcomes; panel studies, which repeatedly survey the same fixed sample; and retrospective studies, which analyze existing historical data to reconstruct past events. Repeated cross-sectional studies, a variant, involve surveying different samples from the same population at multiple time points to infer trends, though they do not track individuals. These approaches are particularly valuable in fields like epidemiology, psychology, and sociology for investigating chronic disease progression, developmental trajectories, and the long-term impacts of interventions. Among the advantages of longitudinal studies are their ability to reduce recall bias by collecting data in real-time, account for cohort effects across generations, and adjust for confounding variables when estimating attributable and relative risks. They excel at linking specific exposures to outcomes and monitoring individual-level changes, making them essential for prognosis in clinical settings and understanding disease etiology. However, challenges include high attrition rates due to participant dropout, which can introduce bias; substantial time and financial costs for long-term follow-up; and difficulties in disentangling reciprocal causation between variables. Notable examples include the Framingham Heart Study, initiated in 1948, which prospectively followed over 5,000 residents to identify cardiovascular risk factors like hypertension and smoking. The Hertfordshire Cohort Study retrospectively linked birth records to later health data, revealing associations between fetal growth and adult coronary heart disease. Such studies have profoundly influenced public health policies and underscore the method's role in advancing evidence-based knowledge.

Overview

Definition and principles

A longitudinal study is a research design that involves repeated observations of the same variables, such as individuals, groups, or phenomena, over multiple time points to examine changes, developments, or trends. This approach contrasts with one-time snapshots, like cross-sectional studies, by capturing dynamic processes rather than static associations at a single point. Typically, it employs continuous or repeated measures to follow participants over prolonged periods, often years or decades, allowing researchers to track exposures, outcomes, and their evolution. Central to longitudinal studies is the principle of temporality, which positions time as the key variable for establishing the sequence of events and understanding causal directions or developmental trajectories. Unlike designs focused on between-subjects differences, these studies emphasize within-subjects changes, analyzing how the same entities vary over time to reveal intraindividual growth or decline. This focus requires a long-term commitment to tracking subjects, ensuring consistent data collection to minimize biases from attrition or external influences. Core elements include treating time not as a cause but as a metric for change processes, with measurements taken at fixed or varying intervals tailored to the phenomenon under study—such as annual assessments for slow-developing traits or more frequent ones for rapid changes. At least two repeated observations are needed to detect and model change effectively, enabling the detection of linear or nonlinear patterns that single observations cannot discern. These principles underpin the study's ability to provide robust insights into temporal dynamics, distinguishing it from static methods.

Comparison with other designs

Longitudinal studies differ fundamentally from cross-sectional studies in their approach to time and subject tracking. While longitudinal designs involve repeated measures on the same individuals over extended periods—often years or decades—to observe changes and trajectories, cross-sectional studies collect data at a single point in time from different subjects, offering a static snapshot of a population but unable to distinguish individual-level changes from group differences. This temporal distinction allows longitudinal studies to avoid confounding by cohort effects, such as generational differences in experiences or exposures that can bias cross-sectional comparisons across age groups, as the same cohort is followed throughout. In contrast to experimental designs, longitudinal studies are inherently observational and non-manipulative, relying on the natural progression of variables without researcher intervention, whereas experiments actively manipulate independent variables—often through random assignment—to isolate causal effects and establish stronger internal validity. Longitudinal approaches thus prioritize real-world dynamics and long-term patterns in unmanipulated settings, making them complementary to experiments when ethical or practical constraints prevent variable control, though they yield weaker causal inferences due to the absence of randomization. Longitudinal studies also diverge from case-control designs in directionality and scope. Prospective longitudinal (cohort) studies follow exposed and unexposed groups forward to identify emerging risk factors and outcomes, enabling the assessment of multiple effects from a single exposure, in opposition to case-control studies that retrospectively compare individuals with and without a specific outcome to pinpoint prior risk factors, which is particularly efficient for rare diseases or outcomes with long latency periods. This forward-looking nature of longitudinal designs supports the establishment of temporality—where potential causes precede effects—reducing issues like recall bias inherent in the backward-tracing of case-control methods. Researchers select longitudinal designs over alternatives when investigating developmental processes, such as aging or behavioral evolution, or when temporal precedence is essential for causal inference, as these studies provide sequenced data that cross-sectional snapshots or retrospective case-control analyses cannot replicate. They are ideal for fields like epidemiology or psychology where understanding change direction and individual variability is paramount, but less suitable for scenarios requiring quick results, where cross-sectional or experimental methods offer faster insights.

Types

Prospective studies

Prospective studies, also known as prospective cohort studies, are a type of longitudinal design in which researchers recruit participants at a baseline point in time, typically before any outcomes of interest have occurred, and then follow them forward to collect data as events unfold. This setup allows for the observation of natural changes and developments in real time, starting from an initial assessment where participants are selected based on shared characteristics or exposures, such as age, health status, or environmental factors, while ensuring they are free of the outcome at the outset. Key features of prospective studies include the capture of data prospectively as outcomes develop, which enables the establishment of temporality—demonstrating that exposures precede outcomes—and supports stronger inferences about potential causal relationships compared to other designs. These studies are particularly common in cohort research, where groups exposed to specific factors (e.g., lifestyle habits or environmental risks) are tracked alongside unexposed groups to monitor incidence rates and associations over time. For instance, the Framingham Heart Study, initiated in 1948, recruited residents of Framingham, Massachusetts, and has followed them through multiple generations with baseline cardiovascular assessments, illustrating how prospective designs can reveal long-term patterns in disease development. The structure of prospective studies typically begins with comprehensive baseline assessments, followed by periodic follow-ups at predetermined intervals, such as annual surveys or clinical examinations, to track changes systematically. To address attrition, which can introduce bias if participants drop out differentially, researchers implement planned retention strategies, including large initial sample sizes, incentives, regular contact to build rapport, and statistical adjustments like weighting to account for losses. These measures are essential, as attrition rates can exceed 20-30% in long-term cohorts, potentially skewing results toward healthier or more compliant subgroups. Unique considerations in prospective studies revolve around ethical challenges, particularly obtaining and maintaining long-term informed consent, as participants may not fully anticipate future study demands or evolving risks over decades. This requires dynamic consent processes, such as ongoing re-consent or broad initial permissions for unforeseen analyses, to uphold autonomy while minimizing burden, especially in vulnerable populations like children or the elderly. Additionally, the extended timelines—often spanning years or lifetimes—imply significant cost implications, including expenses for repeated data collection, participant tracking, and infrastructure, which can make these studies resource-intensive compared to retrospective alternatives that reconstruct past events more quickly.

Retrospective studies

Retrospective studies represent a backward-looking approach within longitudinal research, where investigators analyze pre-existing records, databases, or participant recollections to reconstruct the timeline of exposures, events, and outcomes from the past up to the present state. This design allows researchers to identify cohorts based on historical criteria—such as birth years or employment records—and trace the progression of conditions without initiating new data collection. Unlike forward-tracking methods, it leverages already available information to establish temporal relationships, making it particularly suited for examining long-term effects where prospective follow-up would be impractical. Key features of retrospective studies include their efficiency in time and cost, as they utilize existing data sources like medical archives, employment logs, or administrative databases, avoiding the need for prolonged participant monitoring. These studies often rely on electronic health records, historical registries, or retrospective self-reports to compile longitudinal profiles, enabling rapid analysis of large populations. They are especially prevalent in epidemiological research for investigating rare events or conditions with extended latency periods, where assembling sufficient cases prospectively would require decades or substantial resources. Execution of retrospective studies faces several challenges, primarily related to data quality, such as incomplete or inconsistent records stemming from variations in historical documentation practices. Verifying the accuracy of timelines can be difficult due to potential gaps in archival data or reliance on memory-based reports, which may introduce errors in event sequencing. Additionally, selection bias arises from the availability and accessibility of data sources, as only certain populations or records may be represented, potentially skewing results toward those with better documentation. A representative example is the use of retrospective studies to trace disease progression from past exposure logs to current outcomes, such as analyses linking occupational asbestos exposure—documented in historical employment and health records—to the development of mesothelioma in affected workers. These investigations reconstruct exposure timelines from decades prior to assess incidence rates and progression patterns in rare asbestos-related cancers. Such approaches can complement prospective designs by providing historical validation of risk factors observed in ongoing cohorts.

Methodology

Design and sampling

The design of a longitudinal study begins with clearly defining the research questions and hypotheses, which guide the overall structure and focus on key outcomes such as changes in health status or behavioral patterns over time. Timelines are established based on the study's objectives, often spanning years or decades to capture long-term trajectories, with planning phases including protocol development and staff training that can take at least one year before data collection starts. Researchers must choose between fixed intervals, where assessments occur at predetermined regular times (e.g., annually), and event-based intervals, where follow-up is triggered by specific occurrences like health events, to align with the study's aims and minimize biases from unobserved changes. Power calculations are essential for determining sample size, accounting for expected attrition to ensure sufficient statistical power for detecting meaningful changes. A common approach adjusts the base sample size formula for proportions by inflating it for anticipated loss to follow-up. The attrition-adjusted sample size NN can be calculated as: N=Z2⋅p⋅(1−p)E2⋅11−rN = \frac{Z^2 \cdot p \cdot (1-p)}{E^2} \cdot \frac{1}{1 - r} where ZZ is the z-score corresponding to the desired confidence level (e.g., 1.96 for 95% confidence), pp is the estimated prevalence or proportion of the outcome, EE is the margin of error, and rr is the expected attrition rate. This adjustment helps maintain power despite participant dropout, which is common in extended studies. Sampling methods prioritize representativeness to support generalizable inferences. Probability sampling, such as random or stratified selection from a defined population, ensures each individual has a known chance of inclusion, facilitating unbiased estimates of population parameters. Cohort-specific sampling targets groups sharing a common experience, like birth cohorts following individuals born in a particular period to study developmental trajectories. To minimize loss to follow-up, strategies include oversampling underrepresented or high-risk subgroups at baseline, such as ethnic minorities, to compensate for potential differential attrition and preserve sample balance. Additional retention efforts, like collecting detailed contact information and offering flexible assessment modes, can further reduce dropout rates. Ethical planning is integral, requiring ongoing informed consent to uphold participant autonomy in multi-year commitments, with processes that reaffirm understanding of study purpose, risks, voluntariness, and withdrawal rights at regular intervals to address potential forgetting over time. Institutional Review Board (IRB) approval is mandatory, evaluating risks, benefits, and protections under principles of respect for persons, beneficence, and justice as outlined in federal regulations. Practical considerations include budgeting for extended durations, estimating costs for personnel, equipment, and participant incentives across out-years while justifying variations, such as increased analysis expenses in later phases, to secure sustainable funding.

Data collection techniques

In longitudinal studies, data collection relies on a range of methods to capture repeated measures from the same participants over time, ensuring the reliability of tracking changes in variables such as health outcomes or behaviors. Common techniques include surveys and interviews for self-reported data, biomarkers for objective physiological indicators (e.g., blood samples or wearable sensor readings), and administrative records for verifiable historical information like medical or employment histories. These approaches enable the gathering of both quantitative metrics, such as frequency of events, and qualitative insights, such as personal experiences. Mixed-methods designs, integrating surveys with biomarkers or records, facilitate triangulation to cross-validate findings and reduce biases inherent in single-method reliance. To maintain consistency in repeated measures across multiple waves, researchers implement standardized protocols that use identical instruments, question wording, and procedures at each time point, often employing unique coding systems to link data to individuals. Technology, particularly mobile applications, supports real-time logging by allowing participants to input data via smartphones, such as daily symptom tracking or ecological momentary assessments, which minimizes recall errors and enables frequent, low-burden collections over periods ranging from weeks to years. For instance, apps with push notifications and automatic synchronization have been shown to improve adherence in health-related longitudinal tracking, though challenges like digital literacy must be addressed. Quality control is paramount to uphold data integrity, involving rigorous training for data collectors to ensure uniform administration of methods and regular monitoring to detect deviations. Non-response, a frequent issue in repeated measures, is managed through strategies like personalized reminders via email or phone and monetary or gift incentives, which have been found to boost retention rates in cohort studies. Any changes in measurement tools, such as updates to survey software, are meticulously documented to allow for adjustments in data interpretation and to preserve comparability. Specific techniques address common challenges in longitudinal data gathering. Panel conditioning, where repeated participation alters respondents' behaviors or responses (e.g., increased awareness leading to behavioral changes), can be mitigated by extending intervals between waves to reduce cumulative effects and using statistical adjustments like weighting to account for experienced versus new participants. For retrospective elements within prospective designs, event history calendars improve recall accuracy by providing a graphical timeline anchored to landmark events, prompting sequential and parallel retrieval of life details; studies show this method reduces inconsistencies in event dating by enhancing completeness and agreement with prior reports, for example achieving 87% agreement between concurrent and retrospective reports of school attendance in a longitudinal study.

Analysis

Statistical approaches

Longitudinal studies generate repeated measures over time, necessitating statistical methods that account for within-subject correlations, temporal dependencies, and heterogeneity across individuals. Primary approaches include multilevel modeling, growth curve analysis, time-series techniques, generalized estimating equations, and causal inference methods adapted for time-varying factors. These models enable estimation of trajectories, average effects, and causal relationships while handling the nested structure of data where observations are clustered within subjects. Multilevel modeling, also known as hierarchical linear modeling, is a cornerstone for analyzing longitudinal data with nested structures, such as repeated measures within individuals. It partitions variance into fixed effects (common across subjects) and random effects (varying by subject), allowing for individual-specific intercepts and slopes in trajectories over time. This approach accommodates unbalanced data and missing observations under certain assumptions, making it suitable for studying change processes like cognitive development or health outcomes. A basic two-level multilevel model for outcome YijY_{ij} at time jj for subject ii can be expressed as: Yij=β0+β1⋅Timeij+u0i+u1i⋅Timeij+eijY_{ij} = \beta_0 + \beta_1 \cdot \text{Time}_{ij} + u_{0i} + u_{1i} \cdot \text{Time}_{ij} + e_{ij} where β0\beta_0 and β1\beta_1 are fixed effects for the intercept and slope, u0iu_{0i} and u1iu_{1i} are random effects capturing subject-specific deviations (assumed normally distributed with mean zero), and eije_{ij} is the residual error. Seminal developments in this framework emphasize its flexibility for continuous outcomes and extensions to categorical data via generalized linear mixed models. Growth curve analysis, often implemented within multilevel frameworks, focuses on modeling individual developmental trajectories and population-level patterns of change. It estimates latent growth parameters, such as initial status and rate of change, while testing for covariates influencing these trajectories, such as age or intervention effects. This method is particularly useful for hypothesis testing about acceleration or deceleration in growth, as seen in studies of child language acquisition or disease progression, and handles non-linear forms through polynomial or spline specifications. Key advantages include its ability to incorporate time-invariant and time-varying predictors without assuming equal spacing of measurements. For individual-level trends, time-series analysis methods like autoregressive integrated moving average (ARIMA) models capture autocorrelation and non-stationarity in sequential data. ARIMA, originally developed for univariate forecasting, adapts to longitudinal contexts by modeling trends, seasonality, and shocks at the subject level, such as in intensive repeated measures from ecological momentary assessments. It specifies a process as ARIMA(p,d,q), where p is the autoregressive order, d the differencing for stationarity, and q the moving average order, enabling prediction of future values based on past errors and observations. While computationally intensive for large panels, it excels in detecting abrupt changes, like intervention impacts in single-subject designs. Generalized estimating equations (GEE) provide a robust alternative for estimating population-averaged effects in longitudinal data, particularly when interest lies in marginal associations rather than subject-specific predictions. Introduced for correlated responses, GEE extends generalized linear models by specifying a working correlation structure (e.g., exchangeable or autoregressive) to account for within-subject dependencies, yielding consistent estimators even under misspecification of the correlation. It is widely applied to non-normal outcomes, such as binary or count data in clinical trials tracking symptom severity over time, and focuses on average trends across the population. The method's sandwich variance estimator ensures valid inference for clustered data without requiring full likelihood specification. Causal inference in longitudinal settings often employs propensity score methods adapted for time-varying exposures to balance confounders at each time point. These approaches, such as inverse probability weighting, estimate the probability of exposure given past history and covariates, then weight observations to create pseudo-populations mimicking randomization. This mitigates bias from time-dependent confounding, as in studies of dynamic treatment regimens for chronic conditions, where exposures like medication adherence fluctuate. Similarly, instrumental variable (IV) approaches address unmeasured confounding by leveraging variables that affect exposure but not the outcome directly, such as policy changes or genetic markers. In longitudinal data, two-stage least squares or GMM estimators extend IV to time-series cross-sections, isolating exogenous variation while controlling for fixed effects. Both methods enhance causal validity but require strong assumptions, like no unmeasured confounders affecting the instrument. Recent advances as of 2025 integrate machine learning techniques, such as recurrent neural networks and transformer models, with traditional statistical methods and causal inference for analyzing intensive longitudinal data, particularly in psychological and clinical research. These hybrid approaches improve prediction of complex trajectories and handling of high-dimensional time-varying covariates, enhancing scalability for large-scale studies while maintaining interpretability through causal frameworks.

Addressing challenges

Longitudinal studies often encounter missing data, which can arise due to participant dropout, skipped assessments, or other factors, and must be addressed to avoid biased estimates. Missing data mechanisms are classified into missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR); the latter two are particularly prevalent in repeated measures designs where missingness depends on observed or unobserved variables, respectively. For MAR data, multiple imputation (MI) is a widely recommended technique that creates multiple plausible imputed datasets based on observed data patterns, analyzes each separately, and pools results to account for imputation uncertainty, reducing bias compared to single imputation methods. Inverse probability weighting (IPW) is another approach suitable for MAR assumptions, where weights are assigned based on the inverse probability of observing the data given observed covariates, effectively upweighting complete cases to represent the full sample. Combining MI and IPW can further enhance robustness when both outcome and covariate missingness occur, as demonstrated in simulations showing improved efficiency over either method alone. Attrition, a form of selective dropout, introduces selection bias by systematically excluding certain subgroups, potentially distorting associations between variables over time. To correct for this, weighting methods adjust for inclusion propensity by estimating probabilities of retention based on baseline and time-varying covariates, then applying inverse weights to balance the sample toward the original population. Sensitivity analyses are essential for evaluating dropout impacts, involving scenario-based testing of assumptions (e.g., varying MNAR patterns) to assess how results change under different missingness mechanisms, thereby quantifying potential bias without assuming a single truth. Empirical evaluations indicate that such post-hoc corrections, while not eliminating bias entirely under MNAR, often outperform complete-case analysis in maintaining generalizability, especially when attrition exceeds 20-30%. Time-varying confounders, which change over the study period and are affected by prior exposures, pose challenges in estimating causal effects, as standard regression adjustments can induce bias by blocking mediator pathways. Marginal structural models (MSMs) address this by using IPW to create a pseudo-population where exposures are independent of confounders, allowing unbiased estimation of dynamic treatment effects through weighted regression. For handling measurement error in repeated assessments of these confounders, simulation studies show that regression calibration or simulation-extrapolation methods can correct MSM estimators, reducing bias by up to 50% in scenarios with moderate error variance, though uncorrected errors may attenuate effects toward the null. Implementation of these techniques relies on specialized software for efficient computation in longitudinal settings. In R, the nlme package supports linear and nonlinear mixed-effects models with built-in options for handling correlated errors and missing data via maximum likelihood estimation. The lme4 package extends this for generalized linear mixed models, offering scalable fitting for large datasets with unbalanced repeated measures and integration with MI via the mice package. In SAS, PROC MIXED provides comprehensive procedures for mixed models, including REML estimation and weighting for attrition, while PROC GENMOD accommodates generalized outcomes with IPW for MSMs. In Python, libraries such as statsmodels offer mixed linear models for longitudinal data analysis, and PyMC enables Bayesian implementations of multilevel models, supporting modern workflows for reproducible research as of 2025. These tools facilitate multilevel modeling extensions, enabling researchers to incorporate the addressed challenges directly into analysis pipelines.

Strengths and limitations

Strengths

Longitudinal studies offer a key advantage in establishing causality by providing temporal precedence, which allows researchers to observe the sequence of events and better infer cause-and-effect relationships compared to cross-sectional designs that capture data at a single point in time. This design facilitates the identification of how exposures precede outcomes, reducing the ambiguity inherent in simultaneous measurements and enabling more robust causal inferences through techniques such as natural experiments and advanced statistical modeling. A primary strength lies in tracking change over time, as these studies follow the same individuals repeatedly, capturing intra-individual variability, developmental trajectories, and aging effects with high accuracy. By observing the same across multiple time points, researchers can assess the duration, , and timing of , distinguishing between age, cohort, and period effects to reveal dynamic patterns that static analyses cannot detect. Longitudinal designs also reduce certain biases, particularly in prospective setups where data collection occurs in real time, minimizing recall bias that arises from retrospective reporting of past events. Furthermore, they allow control for time-invariant confounders—such as inherent individual traits like genetics or baseline characteristics—through analytical approaches like fixed-effects models, which isolate within-person changes and mitigate the impact of unobserved stable factors. Finally, these studies hold significant policy and predictive value by enabling the forecasting of trends, such as disease progression or behavioral shifts, based on observed trajectories and long-term patterns. This capacity to project future outcomes from historical data supports evidence-based decision-making in areas like public health and social policy, offering insights into the long-term implications of interventions or exposures.

Limitations

Longitudinal studies are inherently resource-intensive, requiring substantial financial and temporal investments due to their extended duration, which can span years or decades. These designs demand ongoing data collection efforts, participant tracking, and maintenance of research infrastructure, often leading to higher costs compared to cross-sectional alternatives. For instance, the prolonged follow-up periods necessary to observe changes over time escalate expenses related to personnel, equipment, and repeated assessments. A primary challenge is attrition bias, where participants drop out over time, potentially skewing results toward those who remain in the study, often referred to as "survivors" who may differ systematically from dropouts in ways that affect outcomes. This non-random loss can introduce bias, particularly if attrition correlates with key variables like exposure or health status, reducing the representativeness of the sample and threatening the validity of inferences. While statistical methods exist to address attrition, such as imputation techniques, fully correcting for it remains difficult, especially when dropout patterns are unpredictable or related to unobserved factors. Longitudinal studies also face challenges from other biases, including panel conditioning, where repeated participation may influence participants' responses or behaviors, potentially altering the data collected over time. Additionally, disentangling reciprocal causation between variables—where exposures and outcomes mutually influence each other—can be difficult, limiting the ability to establish clear directional causality despite the temporal data. Ethical and logistical issues further complicate longitudinal research, particularly in maintaining participant privacy and consent over extended periods amid evolving personal circumstances. Prolonged involvement can expose individuals to repeated sensitive inquiries, raising concerns about confidentiality breaches as data accumulates and external factors like data breaches or legal changes intervene. Logistically, ensuring consistent follow-up while respecting autonomy requires robust protocols for re-consent and data protection, yet these can strain resources and participant trust. Finally, limitations in generalizability arise from cohort-specific effects and selection biases inherent to the study design. Participants recruited from a particular time and place may experience unique historical or environmental influences—known as cohort effects—that do not apply to other populations, restricting the applicability of findings beyond the original group. Additionally, initial sampling challenges can result in cohorts that underrepresent certain demographics, further limiting how well results extrapolate to broader societies.

Applications

In health sciences

In health sciences, longitudinal studies are pivotal for tracking disease incidence, treatment efficacy, and risk factors over extended periods, enabling researchers to observe how these elements evolve in populations. For instance, the Framingham Heart Study, initiated in 1948, has continuously monitored participants to identify cardiovascular risk factors such as hypertension, smoking, and diabetes, revealing their cumulative impact on heart disease development. This prospective cohort design has provided foundational evidence for understanding atherosclerosis progression and informing preventive strategies. Similarly, these studies assess treatment efficacy by following patient outcomes post-intervention, capturing variations in response due to individual factors like age or comorbidities. Notable examples illustrate the breadth of applications in and . The , launched in as a prospective cohort of over 120,000 nurses, has examined influences on cancer and , establishing between factors like diet, , and postmenopausal with . In parallel, the UK Biobank, established in 2006 with 500,000 participants, integrates genetic, imaging, and health data to map trajectories of diseases, including genetic predispositions to conditions like dementia and diabetes, facilitating large-scale genomic analyses. These studies have profound impacts on and . By analyzing long-term immunity , longitudinal has shaped policies, such as booster recommendations for to sustain against , based on decay patterns observed over months to years. Furthermore, they advance by tracking , such as levels in patients, which correlate with progression and guide tailored therapies. In , findings from cohorts like Framingham have influenced guidelines on and , reducing population-level cardiovascular mortality. Unique to health sciences, longitudinal studies often integrate with clinical trials to extend observation beyond trial endpoints, combining randomized data with real-world follow-up for comprehensive efficacy assessments. They also employ to handle endpoints like mortality, using techniques such as Cox proportional hazards models to estimate time-to-event risks while accounting for censoring in datasets with varying follow-up durations. This approach is essential for prognostic modeling in chronic diseases, where outcomes like cancer recurrence or organ are tracked amid competing risks.

In social sciences

Longitudinal studies in the social sciences are widely employed to examine dynamic processes such as , structures, , and behavioral changes over time, allowing researchers to track how and societal factors evolve and interact. In , these studies facilitate the of course transitions, including , , and outcomes, by following cohorts or panels through repeated observations that capture both stability and variability. For instance, the National Study (NCDS), initiated in , has tracked over 17,000 individuals born in , , and , providing insights into intergenerational and the long-term effects of early-life experiences on adult . In economics, longitudinal designs like panel studies are instrumental for investigating income dynamics, labor market participation, and wealth accumulation, enabling causal inferences about policy impacts on household well-being. The Panel Study of Income Dynamics (PSID), launched in 1968 by the University of Michigan, is the world's longest-running longitudinal household survey, following more than 18,000 individuals across generations to assess economic resilience, poverty persistence, and family resource allocation. This approach has revealed patterns such as the intergenerational transmission of earnings and the role of education in mitigating economic disadvantage. Sociologists and economists also utilize these studies to explore broader social changes, such as shifts in gender roles, migration patterns, and community cohesion. The British Household Panel Survey (BHPS), conducted from 1991 to 2009 by the University of Essex, monitored approximately 5,500 households annually to document evolving family dynamics, employment trajectories, and subjective well-being in response to societal transformations like welfare reforms. By distinguishing short-term fluctuations from enduring trends, longitudinal research in the social sciences supports robust evidence for theoretical models of social stratification and informs evidence-based policymaking.

References

  1. https://pmc.ncbi.nlm.nih.gov/articles/PMC4669300/
  2. https://www.bmj.com/about-bmj/resources-readers/publications/epidemiology-uninitiated/7-longitudinal-studies
  3. https://methods.sagepub.com/ency/edvol/encyc-of-research-design/chpt/longitudinal-design
  4. https://doi.org/10.3978/j.issn.2072-1439.2015.10.63
  5. https://www.iwh.on.ca/what-researchers-mean-by/cross-sectional-vs-longitudinal-studies
  6. https://nobaproject.com/modules/research-designs
  7. https://pmc.ncbi.nlm.nih.gov/articles/PMC2998589/
  8. https://learning.closer.ac.uk/learning-modules/introduction/what-can-longitudinal-studies-show-us/strengths-of-longitudinal-studies/
  9. https://www.simplypsychology.org/prospective-study.html
  10. https://pmc.ncbi.nlm.nih.gov/articles/PMC11009715/
  11. https://learning.closer.ac.uk/learning-modules/introduction/types-of-longitudinal-research/prospective-vs-retrospective-studies/
  12. https://pubmed.ncbi.nlm.nih.gov/39268885/
  13. https://pmc.ncbi.nlm.nih.gov/articles/PMC9120971/
  14. https://www.scribbr.com/methodology/retrospective-cohort-study/
  15. https://www.sciencedirect.com/topics/medicine-and-dentistry/retrospective-cohort-study
  16. https://www.intechopen.com/chapters/59393
  17. https://pmc.ncbi.nlm.nih.gov/articles/PMC10324708/
  18. https://pmc.ncbi.nlm.nih.gov/articles/PMC6321947/
  19. https://www.scirp.org/journal/paperinformation?paperid=146520
  20. https://pmc.ncbi.nlm.nih.gov/articles/PMC3008590/
  21. https://pmc.ncbi.nlm.nih.gov/articles/PMC3878210/
  22. https://pmc.ncbi.nlm.nih.gov/articles/PMC3772653/
  23. https://pmc.ncbi.nlm.nih.gov/articles/PMC4034392/
  24. https://learning.closer.ac.uk/learning-modules/study-design/section-two/
  25. https://www.nlsinfo.org/content/cohorts/nlsy97/intro-to-the-sample/sample-design-screening-process
  26. https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-018-0586-7
  27. https://www.thehastingscenter.org/irb_article/maintaining-informed-consent-validity-during-lengthy-research-protocols/
  28. https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/read-the-belmont-report/index.html
  29. https://grants.nih.gov/grants-process/write-application/advice-on-application-sections/develop-your-budget
  30. https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-022-01732-4
  31. https://www.jmir.org/2021/1/e17691/
  32. https://www.norc.org/content/dam/norc-org/pdfs/Panel%20Conditioning%20Research%20Brief_FINAL.pdf
  33. https://pmc.ncbi.nlm.nih.gov/articles/PMC2798968/
  34. https://pmc.ncbi.nlm.nih.gov/articles/PMC2613314/
  35. https://www.bristol.ac.uk/media-library/sites/cmm/migrated/documents/longitudinal.pdf
  36. https://pmc.ncbi.nlm.nih.gov/articles/PMC3131138/
  37. https://www.sciencedirect.com/topics/psychology/growth-curve
  38. https://pmc.ncbi.nlm.nih.gov/articles/PMC7986567/
  39. https://ppw.kuleuven.be/okp/_pdf/Ariens2020TSAOI.pdf
  40. https://www.jstor.org/stable/2531734
  41. https://onlinelibrary.wiley.com/doi/10.1155/2014/303728
  42. https://pmc.ncbi.nlm.nih.gov/articles/PMC7154493/
  43. https://pmc.ncbi.nlm.nih.gov/articles/PMC3515775/
  44. https://www.annualreviews.org/content/journals/10.1146/annurev-clinpsy-081423-022947
  45. https://pmc.ncbi.nlm.nih.gov/articles/PMC3016756/
  46. https://pmc.ncbi.nlm.nih.gov/articles/PMC11631082/
  47. https://pmc.ncbi.nlm.nih.gov/articles/PMC3412287/
  48. https://pmc.ncbi.nlm.nih.gov/articles/PMC6766975/
  49. https://pmc.ncbi.nlm.nih.gov/articles/PMC9464329/
  50. https://pmc.ncbi.nlm.nih.gov/articles/PMC7429147/
  51. https://academic.oup.com/aje/article/184/3/249/2237540
  52. https://www.stat.uga.edu/sites/default/files/LMMIntro.pdf
  53. https://support.sas.com/resources/papers/proceedings16/5800-2016.pdf
  54. https://www.statsmodels.org/stable/mixed_linear.html
  55. https://www.pymc.io/projects/docs/en/stable/
  56. https://pubmed.ncbi.nlm.nih.gov/2055872/
  57. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6484949/
  58. https://pmc.ncbi.nlm.nih.gov/articles/PMC9980691/
  59. https://pmc.ncbi.nlm.nih.gov/articles/PMC9121819/
  60. https://learning.closer.ac.uk/learning-modules/introduction/what-can-longitudinal-studies-show-us/weaknesses-of-longitudinal-studies/
  61. https://pubmed.ncbi.nlm.nih.gov/15797643/
  62. https://www.frontiersin.org/journals/sociology/articles/10.3389/fsoc.2019.00033/full
  63. https://bmcpublichealth.biomedcentral.com/articles/10.1186/1471-2458-12-918
  64. https://www.framinghamheartstudy.org/
  65. https://pmc.ncbi.nlm.nih.gov/articles/PMC4966216/
  66. https://academic.oup.com/ije/article/44/6/1800/2572656
  67. https://nurseshealthstudy.org/about-nhs/history
  68. https://pmc.ncbi.nlm.nih.gov/articles/PMC4981811/
  69. https://pmc.ncbi.nlm.nih.gov/articles/PMC8948167/
  70. https://www.science.org/doi/10.1126/scitranslmed.adv4214
  71. https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2019.00898/full
  72. https://www.heart.org/en/news/2018/10/10/framingham-the-study-and-the-town-that-changed-the-health-of-a-generation
  73. https://pmc.ncbi.nlm.nih.gov/articles/PMC3057767/
  74. https://pmc.ncbi.nlm.nih.gov/articles/PMC6110618/
  75. https://www.ahajournals.org/doi/10.1161/circulationaha.115.017719
  76. https://cls.ucl.ac.uk/cls-studies/1958-national-child-development-study/
  77. https://psidonline.isr.umich.edu/
  78. https://www.understandingsociety.ac.uk/about/british-household-panel-survey/
  79. https://sru.soc.surrey.ac.uk/SRU28.html
Add your contribution
Related Hubs
Contribute something
User Avatar
No comments yet.