1/99
Comprehensive vocabulary flashcards covering longitudinal research designs, data types, analytical models, threats to validity, empirical examples, and attrition mitigation strategies based on lecture notes.
Longitudinal Data (Strict Definition)
A research design characterized by repeated observations collected on the exact same individual units over multiple points in time.
Longitudinal Data (Broad Definition)
A broader research definition encompassing repeated observations collected over time for either the exact same units or comparable probability samples.
Cross-Sectional Design
A research design in which measurement of variables for each individual unit or participant occurs only once.
Panel Data
A longitudinal dataset consisting of repeated observations collected on the same individual units (i) across multiple time points (t).
Balanced Panel
A specific type of panel dataset where all participating units possess measurements taken at the exact same points in time.
Time Series Data
A dataset consisting of repeated observations collected for a single individual unit across many sequential time points.
Repeated Cross-Section
A research design utilizing independent probability samples drawn from the same population at different time points to track population-level changes rather than individual unit changes.
Wave
A distinct, scheduled round or period of data collection conducted within a longitudinal study.
Baseline Wave
The very first round or initial period of data collection administered in a longitudinal study.
Lagged Variable
A variable whose value is measured at a prior wave or earlier time point relative to a subsequent focal outcome.
Time-Varying Variable
A variable whose values can change across successive waves of a longitudinal study, such as employment status or marital status.
Time-Stable Variable
A variable whose value remains constant and fixed across all observation waves, such as race or biological sex.
Attrition
The progressive loss or dropout of participants from a longitudinal panel study over successive waves of data collection.
Temporal Ordering
The chronological sequence in longitudinal designs showing that an alleged causal predictor occurred prior in time to its observed outcome.
Within-Individual Effects
Causal or developmental effects estimated by examining how changes occurring within the same individual over time relate to changes in their outcomes.
Early Time-Stable Confounders
Early background characteristics (e.g., biological sex, race, early upbringing) whose confounding effects can be ruled out in within-individual models if their effects remain constant over time.
Recall Bias
A systematic cognitive error where respondents misremember events from the distant past (e.g., 4 years ago) far more frequently than events from the recent past (e.g., 6 months ago).
Differential Measurement Error
A bias occurring when predictor and outcome variables are assessed simultaneously via the same method (e.g., self-report), causing values on one to distort responses on the other.
Testing Effects
Biases that occur when repeated exposure to a test or survey questionnaire alters a respondent's future responses on subsequent study waves.
Measurement Interval Duration
The length of time elapsed between longitudinal data collection waves (e.g., 6 months), which risks failing to capture rapid causal processes if spaced too far apart.
Interviewer Rapport Bias
An artifact where growing familiarity and trust between participant and interviewer across waves increases candor, creating an artificial appearance of change in sensitive behaviors.
Gottfredson & Hirschi (1987) Critique
A theoretical argument asserting that longitudinal research provides minimal explanatory power if primary causes of crime (like low self-control) are time-stable and has too few cases to estimate complex interactions.
Pathways to Desistance Study
A major prospective longitudinal study tracking adjudicated adolescent offenders into early adulthood to examine criminal offending trajectories and the effects of developmental, social, and legal system processes.
Adjudicated Adolescent Offenders
The specific sampling group recruited in the Pathways to Desistance Study, comprising juveniles who were formally processed and convicted of offenses in court.
Study Sites of Pathways to Desistance
Phoenix and Philadelphia, the two specific urban metropolitan locations from which juvenile offenders were recruited for the Pathways to Desistance Study.
Sample Selection Bias in Pathways to Desistance
The methodological concern that recruiting participants solely on the basis of past criminal adjudication distorts inferences regarding future offending patterns.
External Validity Limits of Pathways to Desistance
The inability of findings from serious adjudicated adolescent offenders in Phoenix and Philadelphia to generalize to all juveniles, minor offenders, unapprehended offenders, or other geographic locations.
Tremblay et al. (2004)
A prospective longitudinal study that analyzed the developmental trajectories and predictive factors of physical aggression during early childhood.
Orth et al. (2012)
A longitudinal investigation that evaluated the life-span developmental course of self-esteem and its reciprocal effects on important life outcomes.
Sampson et al. (2006)
A longitudinal study evaluating whether marriage reduces crime using a counterfactual approach to estimate within-individual causal effects.
Population Trend Research Question
A longitudinal question asking how an overall population changes over time (e.g., trends in public support for same-sex marriage), best answered via repeated cross-sections.
Individual Developmental Research Question
A longitudinal question asking how individual units change as they age (e.g., physical aggression in early childhood), best answered via prospective panel designs.
Age, Period, or Cohort Attribution Question
A longitudinal question examining whether observed shifts stem from aging, historical periods, or birth cohorts (e.g., whether people become more conservative with age), answered using repeated cross-sections or multi-cohort panels.
Circumstance-Outcome Covariation Question
A longitudinal question investigating whether changes in life circumstances alter behavioral outcomes (e.g., marriage reducing crime), answered via prospective or retrospective panel designs.
Time-to-Event Research Question
A longitudinal question assessing the duration until a specific event takes place (e.g., time to rearrest following release), answered using panel or retrospective data containing exact event dates.
Policy Impact Research Question
A longitudinal question testing whether a legal enactment or event altered societal outcomes (e.g., the effect of a new law on crime rates), answered via repeated cross-sections, panels, or time series.
Repeated Cross-Sectional Design
A study design where separate probability samples are drawn from the same population at two or more time points using comparable sampling methods and measurement tools.
General Social Survey (GSS)
A nationally representative survey cited as a classic example of a repeated cross-sectional study tracking population trends over time.
Cost and Speed Advantage of Repeated Cross-Sections
The practical benefit that repeated cross-sectional studies are cheaper and faster to conduct than panel studies because researchers do not track participants over time.
Attrition Avoidance in Repeated Cross-Sections
The structural design benefit of repeated cross-sections where participant dropout across waves is eliminated because an entirely new sample is drawn at each time point.
Nonresponse in Repeated Cross-Sections
A form of sampling bias in repeated cross-sections where selected individuals refuse or fail to respond at a specific wave, even though panel attrition does not occur.
Developmental Limitation of Repeated Cross-Sections
The inability of repeated cross-sectional studies to analyze developmental patterns or trajectories within individual units because different units are measured at each wave.
Individual Causal Ordering Limitation of Repeated Cross-Sections
The limitation that repeated cross-sectional designs cannot establish the temporal order or causal sequence of relationships at the individual unit level.
Prospective Panel Design
A longitudinal research design where participants are recruited, measured at baseline, and then followed forward into the future over multiple successive measurement waves.
Contemporaneous Measurement
Assessing a participant's current status or behavior at the exact moment of survey administration (e.g., asking 'What is your GPA right now?').
Interval Measurement
Assessing events, behaviors, or conditions that occurred during the specific window of time elapsed between the prior survey wave and the current wave.
Temporal Order Advantage of Prospective Panels
The capability of prospective panel designs to establish that causes precede their effects in time, provided measurements occur with sufficient frequency.
Within-Unit Change Advantage of Prospective Panels
The primary analytical advantage of prospective panel designs enabling researchers to measure and examine developmental trajectories occurring within the same individual over time.
Practical Limitations of Prospective Panels
The major drawbacks of prospective panel studies, including high financial costs, long completion times, and susceptibility to participant attrition.
Retrospective Panel Design
A study design where participants are recruited at a single point in time and asked to report backwards on their past activities, life events, or beliefs.
52-Subject Follow-Up (Sampson et al., 2006)
A specific retrospective interview follow-up component highlighted in Sampson et al. (2006) used to gather past life-event histories from former study subjects.
Cost and Speed Advantage of Retrospective Panels
The logistical benefit of retrospective panels being relatively inexpensive and fast to complete because all historical data are gathered in a single measurement session.
Absence of Attrition in Retrospective Panels
The structural feature of retrospective panels where multi-wave dropout is eliminated because participants are surveyed only once.
Selection Effects in Retrospective Panels
A major threat to validity in retrospective designs where only individuals who are alive, accessible, and non-institutionalized at the interview date can be sampled.
Recall Issues in Retrospective Panels
The degradation of data quality in retrospective panels caused by participant memory decay, distortion, or inaccuracy regarding events occurring years prior.
Lack of Baseline Measures in Retrospective Panels
The limitation of retrospective designs where prospective baseline measurements do not exist, making objective verification of pre-existing conditions impossible.
Life History Calendar
A timeline matrix tool utilized in retrospective studies to record monthly histories across life domains such as residence, marriage, children, education, employment, and criminal justice contact.
Age Effect
A systematic change in attitudes, behaviors, or health that occurs primarily because an individual is growing chronologically older.
Cohort Effect
Variations in attitudes, traits, or behaviors resulting from the unique historical, generational, and cultural milieu shared by individuals born in the same era.
Period Effect
A broad historical event or macro-level societal shift that impacts all individuals in a population simultaneously, regardless of their chronological age or birth cohort.
Confounding of Age and Cohort
The mathematical reality that age and birth cohort are perfectly collinear at a single point in time (e.g., every 70-year-old in 2026 was born in 1956), making them impossible to separate without longitudinal data.
Multi-Cohort Multi-Time Point Design
An analytical research strategy that measures multiple distinct birth cohorts across multiple calendar time points to disentangle age, period, and cohort influences.
Same Age, Different Cohorts Comparison
An analytical comparison holding age constant across different calendar waves (e.g., comparing 40-year-olds in 1996 to 40-year-olds in 2026) to evaluate cohort effects.
Same Cohort, Different Ages Comparison
An analytical comparison tracking members of a single birth cohort across time (e.g., evaluating people born in 1956 at age 40 in 1996 and age 70 in 2026) to evaluate age effects.
Period Effect Assumption in APC Disentanglement
The necessary analytical assumption that no major macro-level historical events or secular shocks occurred between observation points that could confound age and cohort comparisons.
2008 Recession Period Effect Example
A historical macroeconomic crisis cited as an example of a period effect that may have altered political conservatism across all cohorts and age groups simultaneously.
Trajectory
The developmental path, course, or pattern that a specific behavior, characteristic, or attitude follows across time.
Continuous Trajectory Variation Assumption
The modeling assumption that individual unit developmental trajectories fluctuate continuously around an overarching population average trajectory.
Multi-Level Modeling (Hierarchical Linear Modeling)
An analytic modeling framework used to evaluate longitudinal trajectories by partitioning variation into within-individual and between-individual components.
Random Effects (Mixed) Models
Statistical models that incorporate both population-average fixed parameters and unit-specific random effects to explain continuous variation in developmental trajectories.
Growth Curve Models
A structural modeling technique used to estimate the average developmental trajectory of a construct over time and explain individual deviations around that trajectory.
Group-Based Trajectory Modeling (GBTM)
A semi-parametric statistical method assuming that the population is composed of distinct, latent sub-groups that each follow unique developmental trajectories.
Growth Mixture Modeling (GMM)
An advanced statistical method combining growth curve modeling with latent class analysis to identify distinct developmental trajectory classes that possess within-class variation.
Trajectory Intercept
The baseline level or starting value of an outcome variable at the beginning of an observed developmental trajectory.
Trajectory Trend
The rate, slope, and direction of change in an outcome variable over time across an observed developmental trajectory.
Differential Attrition
A form of non-random attrition occurring when participant dropout rates differ systematically across sub-groups, introducing sample selection bias.
Sources of Longitudinal Attrition
The primary causes of participant loss over time, including participant mortality, incarceration, inability to relocate moved individuals, active refusal, and data dropped for poor quality.
Participant Remuneration
Providing financial payments or incentives to participants at each wave to foster study retention and minimize panel attrition.
Inter-Wave Tracking Resources
Allocating dedicated staff and funding between measurement waves to maintain updated contact information and track participants.
Repeat Contact Strategy
A retention procedure that directs resources toward making multiple, persistent contact attempts across different times and modes to locate participants.
Short Form Survey Strategy
The practice of administering an abbreviated questionnaire focusing only on core questions to reduce respondent burden and decrease wave dropout.
Oversampling for Attrition
A design strategy where researchers recruit a substantially larger baseline sample than statistically required to buffer against anticipated future sample attrition.
Rich Baseline Data Collection Strategy
Gathering extensive demographic, contact, and background information during wave one to aid in tracking participants and analytically adjusting for non-random attrition.
Missing Data Methods
Statistical procedures used to account for and analyze missing values caused by participant nonresponse and attrition in longitudinal panel studies.
Measurement Invariance
The methodological requirement that a survey instrument or operationalization measures the exact same underlying construct in the exact same manner across all study waves.
Scale and Wording Invariance
The requirement that question phrasing and response options (e.g., avoiding switching from Likert scales to numeric ladders) remain identical over time.
Mode of Data Collection Invariance
Maintaining a consistent administration medium (e.g., in-person interview versus telephone interview) across study waves to prevent measurement artifacts.
Construct Non-Invariance
A threat to validity occurring when the underlying meaning captured by a measure shifts over time or across developmental stages.
Physical Aggression Non-Invariance (Tremblay et al.)
A developmental measurement problem where behaviors like biting indicate normative development at age 5 but signify severe behavioral pathology at age 10.
Attachment to Parents Non-Invariance
A developmental measurement problem where the behavioral expressions and operational meaning of parental attachment change fundamentally between early childhood and adolescence.
Religiosity Non-Invariance
A developmental measurement problem where religious attendance is parental and imposed during childhood but voluntary and self-chosen in adulthood.
Self-Esteem Reference Shift
A developmental measurement shift where the cognitive reference group for self-worth moves from external social peers to internal standards and comparisons with one's prior self.
Panel Conditioning
A bias in longitudinal studies occurring when participants' previous survey experience or measurement exposure alters their responses or behaviors in future waves.
Survey Shortening Conditioning
A panel conditioning pattern where respondents realize affirmative answers trigger lengthy follow-up questions, leading them to falsely answer 'no' in future waves to finish faster.
Self-Monitoring Reactivity
A panel conditioning phenomenon where the very requirement to log a behavior (such as daily soda consumption) causes participants to alter that behavior.
Change Score
A basic metric for describing longitudinal change calculated by subtracting a unit's score at an earlier wave from its score at a subsequent wave (Yt2−Yt1).
Time Series Forecasting
An analytic modeling technique that uses sequential past observations of a single unit or population over many time points to project future trends.
Differences-in-Differences (DiD)
A quasi-experimental longitudinal method that estimates causal policy effects by comparing the pre- and post-treatment change in an outcome for a treated group against that of an untreated comparison group.
Synthetic Control Method
A quasi-experimental statistical method used to evaluate policy effects by constructing a weighted combination of untreated comparison units to approximate the counterfactual trajectory of the treated unit.
Interrupted Time Series
A quasi-experimental design evaluating the impact of an intervention or policy by analyzing shifts in longitudinal trends before and after a specific event date.