Cross-Sectional and Longitudinal Observational Designs

Flashcards covering concepts of cross-sectional and longitudinal observational designs, cohort-sequential studies, confounding, developmental psychopathology models, and related empirical examples.

What are the core characteristics of an observational (non-experimental) research design?

Observational designs lack random assignment to conditions, involve no direct manipulation of variables, and observe relationships between or among variables, between-group differences, or within-individual changes.

How are age effects, period effects, and cohort effects distinguished in observational research?

Age effects represent changes due to growing older; period effects represent environmental or social influences that vary through historical time; and cohort effects stem from being born or coming into existence at different times.

Why can observational designs never completely rule out unmeasured confounding?

Because observational designs do not use random assignment, making it impossible to ensure that all potential modifiers of effect and disturbing variables are equally distributed across comparison groups.

What is the defining feature of a cross-sectional research design?

It measures variables and gathers information about covariation at a single point in time across individuals or groups.

What are the main advantages of a cross-sectional study design?

It is relatively inexpensive and quick, measures participants at roughly the same time, has no participant attrition, and minimizes repeated measurement, testing, and period effects.

What are the primary disadvantages of a cross-sectional study design?

It cannot determine causality or the direction of effects, cannot evaluate within-individual change over time, cannot directly test developmental theories, and can confound age differences with cohort effects.

In the study by Floyd et al. (2008), what observational instrument and metric were used to assess physical activity in public parks?

A modified version of the System for Observing Play and Leisure Activity in Youth (SOPLAY) was used, with activity codes transformed into energy expenditure per person ($kcal/kg/min$).

What proportion of park users were observed in sedentary behavior in the Floyd et al. (2008) study across Tampa, FL and Chicago, IL?

$70\%$ of park users in Tampa and $51\%$ of park users in Chicago were observed engaged in sedentary behavior.

How did park energy expenditure differ by neighborhood racial and ethnic composition in Floyd et al. (2008)?

In Tampa, parks located in neighborhoods with the highest concentration of Hispanic residents had the highest mean energy expenditure ($0.069\,kcal/kg/min$), whereas in Chicago, parks in neighborhoods with the highest concentration of African Americans had the highest mean energy expenditure ($0.087\,kcal/kg/min$).

What are the key defining characteristics of a longitudinal design?

A longitudinal design collects repeated measures ($2$ or more occasions) over time on either the same individuals (to assess within-person change) or the same population (to assess trends).

What are the primary advantages and disadvantages of longitudinal research designs?

Advantages include mapping developmental trajectories, establishing temporal precedence for risk factors, and evaluating within-person change. Disadvantages include high financial cost, participant attrition, potential testing effects, and selection bias.

What distinguishes a trend study from a panel study?

A trend study takes repeated cross-sectional samples of different individuals from the same population over time (e.g., BRFSS, NHANES) to observe aggregate changes, whereas a panel study tracks the exact same individuals over time.

What is a major limitation of two-time-point prospective follow-up studies compared to studies with three or more time points?

Two time points can only model linear change and are vulnerable to regression artifacts and measurement error, whereas three or more time points are required to detect nonlinear growth and developmental trajectories.

What is a cohort-sequential (accelerated cohort) design?

A design that takes limited repeated measurements of several temporally overlapping age cohorts simultaneously, linking adjacent data segments to approximate a long-term developmental trajectory in a shorter calendar period.

What primary assumptions must be satisfied for a cohort-sequential design to accurately approximate a long-term longitudinal study?

It assumes a common developmental trend exists across cohorts, that there is no age-by-cohort interaction ($\text{age} \times \text{cohort}$ effect), that individuals without data can be represented by those present, and that the growth function correctly describes trajectories.

In the observational cohort of the SPORT study (Weinstein et al., 2006), how did baseline patient health trends illustrate confounding by indication?

Patients whose problem was self-assessed as getting worse were significantly more likely to receive surgery ($52\%$) than nonoperative care ($20\%$), demonstrating that patients with worse prognoses preferentially selected surgical treatment.

Which developmental psychopathology concept is illustrated by this diagram of a single starting condition diverging into multiple distinct outcomes?

Multifinality, which states that diverse developmental outcomes are likely to result from any single initial source of influence or beginning.

Which developmental psychopathology concept is illustrated by this diagram of multiple different starting conditions converging into a single distal outcome?

Equifinality, which states that the same final developmental outcome may emanate from diverse routes, antecedents, or trajectories.

In the 22-year cohort study on leisure-time physical activity trajectories by Barnett et al., what four trajectory classes and distribution percentages were identified?

Inactive ($56\%$), Increaser ($25\%$), Active ($12\%$), and Decreaser ($7\%$).

What are two methods used to handle participant attrition in longitudinal studies, and what issues does each present?

1) Following only initially enrolled individuals, which causes loss of statistical power and selection bias because dropouts differ from remaining participants; 2) Refreshing the sample with new individuals, which introduces bias if incoming participants systematically differ from stayers and movers.