Developmental Psychology: Research Issues Intractable Variables –Difficult or impossible to manipulate Heredity/Genes Environment Age –Age is a “proxy”

Developmental Psychology: Research Issues

• Intractable Variables– Difficult or impossible to manipulate

• Heredity/Genes• Environment• Age

– Age is a “proxy” for causal variables—i.e., age co-varies with these causal agents, but it is not a causal variable

• Measurement Equivalence

– When constructs change with development, measures that are appropriate at one age (time) may be inappropriate at another age

• Ex: Assessment of attachment in preschoolers

– Is a separation-reunion procedure appropriate?

• Self-report data are limited

– Parents, teachers, and other adults often complete “self-report” measures of children’s behavior

– Greater reliance on observational techniques

• Inferring the meaning of behavior is difficult

– Ex: Infants’ understanding of object permanence

• Complexity of causal influences

– Ethical concerns preclude manipulation of many variables

• Ex: maltreatment and children’s development

– Laboratory analog studies may compromise external validity

• Ex: Marital conflict and children’s development

• Direction of causation

– Biases/assumptions about the direction of influence

• Ex: parents influence children rather than the reverse

– Bidirectional influences are more likely than unidirectional influences

General Research Designs

• Experimental Designs

– Manipulation of hypothesized independent variable

– Random assignment of participants to different conditions (between-subjects designs) OR other control procedures (within-subjects designs, small-n designs)

– Allow strong inferences about causal relationships

• Potential Limitations of Experimental Designs

– Participant non-compliance in the “treatment” or “intervention” condition (e.g., dropping out, failure to participate fully in the treatment)

– Generalization (external validity)

• Non-experimental (Correlational) Designs

– No manipulation of variables

– No random assignment or other comparable control procedures

– Not possible to make strong causal inferences

Why not?

• Selection Bias (Confounding Variables)– Refers to third variables that are correlated

with both the predictor variable and the outcome variable

• Ex: Does high-quality child care cause improved school readiness?

– Children in high-quality child care (and their families) are likely to be different in many ways from children in lower-quality child care (socioeconomic status; high-quality parental care)

– These “confounding” variables are likely to be related to school readiness

• Most common approach to reducing selection bias:

– Identify, measure, and control for possible confounding variables either in the research design or in the statistical analysis

Developmental Designs

• Designs in which age-related change is examined

– Normative development (developmental functions)

– Individual differences

Variables Involved in Developmental Designs

• Cohort: Groups of participants who are born or experience some other common event in the same time period

– Ex: children born in 1980 are a cohort; individuals growing up during the Great Depression are also a cohort

• Age

• Time/Point of Assessment

Simple Developmental Designs

• Longitudinal Designs

– A single cohort is examined at multiple ages (and thus at multiple times of assessment)

– Age and time of assessment are confounded

• An event may occur between points of assessment that produces differences in the dependent variable

– Ex: Sept. 11 may affect rates of psychological disorders in children

» If we see increases in psychological disorders in a longitudinal design, are they due to age or to time of assessment differences?

• Advantages

– Can examine stability and change in individual children’s characteristics and behavior over time

• Disadvantages

– Non-random participant loss (selective attrition)

• Participants who finish the study differ in systematic ways from participants who drop out

– Final sample is not representative of the group

(population) researcher wanted to study—findings may not generalize

– Practice effects

• Change due to familiarity with data collection procedures rather than change due to development

– Time-consuming and expensive

• Cross-sectional Designs

– Multiple cohorts (and multiple ages) are examined at a single time of assessment

– Cohort and age are confounded

• Differences across cohorts may produce changes in the dependent variable

– Ex: Cohorts born in 1970 and 1990 are likely to differ with respect to early child care experiences

» If we see differences in social competence in a cross-sectional design, are they due to age or to cohort differences?

• Advantages

– More efficient than a longitudinal design (faster, less expensive)

– No participant loss

– No practice effects

• Disadvantages

– Cannot examine stability or change in individual children’s characteristics or behavior over time

Complex Developmental Designs

(Sequential Designs)

• Involve complete crossing of 2 of 3 variables (cohort, age, time of assessment)

• Interpretation of data from these designs is still ambiguous– Results cannot be clearly attributed to one of the

three variables (confounding is still present)

• Baltes (1968) argued for the use of the cohort-sequential design in studies of development

• Allows for the separation of cohort and age effects – But time of assessment is still confounded

with both factors• Baltes argues that time of assessment is unlikely

to affect data in developmental studies

• Cohort-sequential design

– Different cohorts compared at the same ages (but at different times of assessment)

Cohort Time of Assessment

1975 1980 1985

1960 15 20

1965 15 20

• Allows comparison of children of the same age from different cohorts

– Ex: Two groups of 15-year-olds (different cohorts); two groups of 20-year-olds (different cohorts)

• If the same-age groups are different from one another with respect to the dependent variable(s), have evidence for cohort effects

• If not, can attribute any differences to age rather than to cohort

• But both are confounded with time of assessment

Data Collection Techniques

•  Systematic Observation (2 Types)

– Naturalistic Observation

• Observe child’s behavior in a natural environment

–Exs: playground, school, home

– Structured Observation:

• Design a situation that will elicit relevant behavior(s)

• Typically conducted in a laboratory setting (but not always)

• Observe different children in the same situation

Coding Observational Data

– Event sampling: Every occurrence of a behavior(s) during a specified observation period is recorded

– Time sampling: The observation period is divided into intervals and the occurrence of a behavior(s) is recorded if it occurs during an interval; the same behavior is not coded twice in the same interval

• Likely to under- or over-estimate the frequency of behaviors depending on the base rate of the behavior and the size of the interval

– Ratings: Likert-type scales are used to rate behavior(s) during a specified observation period

• Often used for “molar” behaviors (e.g., maternal sensitivity)

• Typically require a higher level of inference on the part of observers

General Disadvantages (Observation):

• Observer Bias

– Observer records/judges behavior inaccurately in order to make it consistent with hypotheses or with other beliefs (unintentional!)

• Participant Reactivity

– Observer’s presence affects behavior of those being observed

• Self-report Measures

– Clinical Interviews

• More “open-ended” questions—response choices are not limited

• Participants may be asked different questions (depending on their answers)

– Structured interviews and questionnaires

• More “close-ended” questions—response choices are limited

–Ex: yes/no questions, rating scales, multiple choice questions

• All participants are asked the same questions

• Parents, child care providers, and teachers often provide information about infants and younger children

– Ex: infant/child temperament; behavior problems; social skills

• General Disadvantage (self-report or report by others):

– Data may be inaccurate due to• Deliberate (or semi-deliberate) deception• Misinterpretations of questions • Lower verbal skills• Memory limitations• Lower observational skills• Less knowledge about relevant behaviors

Psychophysiological Methods

• Record physiological responses – Exs: heart rate, hormone levels (cortisol),

brain wave activity

• Infer psychological “states” from these responses (e.g., perceptions, emotions)

Advantage:

• Can be used with preverbal infants and young children (limited language)

Disadvantage:

• Changes in physiological responses can be caused by many “irrelevant” factors (e.g., hunger, boredom, movement)