PSYC 324 Exam 4

Visual impairment (VI)

A spectrum from low vision to blindness

Types:

• Peripheral: ocular

• Cerebral: brain

Measured by:

• Visual acuity (e.g. 20/400)

• logMAR classification

Etiology of visual impairment

Main anatomical origins:

• Cerebray pathways

• Retina

• Optic nerve

Causes:

• Genetic

• Congenital anomalies

• Perinatal injury

• Infection or illness

Identification of visual impairment

Early signs:

• Poor eye contact

• Lack of tracking

Impacts of visual impairment

Delays in:

• Motor development

• Exploration

• Cognitive milestones

Language:

• Delayed early milestones

• Echolalia

Protective factor:

Neuropsychological profiles for children with visual impairments

Cross-modal plasticity: visual cortex repurposed

• Enhanced auditory processing, memory, and tactile skills

Visual impairments and NDDs

75% have a comorbidity

• Intellectual disability

• Autism

• ADHD

Somatic symptom disorder

Distressing physical symptoms

Excessive thoughts/behaviors

Duration: > 6 months

Functional neurological symptom disorder (FNSD)

Neurological symptoms

Inconsistent with medical findings

Often acute onset

Illness anxiety disorder

Fear of serious illness

Minimal physical symptoms

Reassurance-seeking

Factors for somatic disorders

Factors:

• Predisposing

• Precipitating

• Perpetuating

Predictive coding model

Theory for somatic disorders

• Brain predicts bodily states

• Symptoms = perception + evaluation

• Misinterpretation of signals

Replication crisis

Many classic findings could not be reproduced

Harmed validity and trust in psychological research

Key contributors to failed replicability

• Small sample sizes (low power)

• Publication bias

• Researcher degrees of freedom

• Overreliance on p-values

Combination of strong bias toward statistically significant finds and flexibility in data analysis can lead to irreproducible research

Publication bias

Only studies with statistically significant findings are published

Open science

Increasing openness and transparency in research through various practices

• Pre-register studies

• Increase documentation

• Open data and materials sharing

• Change norms and incentive structures

• Open access publishing

Preregistration

Time-stamped, read-only version of your research plan created before you begin data collection

Makes clear distinction between confirmatory (hypothesis testing) and exploratory research (hypothesis generating)

Pre-analysis plans

Document:

• Target sample: size, population, sampling

• Data cleaning and processing

• Exclusion criterion

• Specific analyses to be conducted

P-Hacking

Unreported flexibility in data analysis

Mining data to see significant patterns without first specifying a hypothesis

HARKing

Hypothesizing After Results are Known

Barriers to preregistration

Time

Having previously analyzed the data

Working on more exploratory research projects

PIs

Registered reports

Paper submitted to a journal with just introduction and methods section

Then goes through peer review

If approved, goes to data collection and analysis

Peer reviewed again to make sure conclusions are supported by the data

Open data and materials

Share data (excluding proprietary or sensitive data)

Share materials: manipulation text, survey questionnaires, stimuli, etc

Share analysis scripts

Ideally, all data and materials are located in one place with persistent identifiers

Preprints

Multiple testing problem

AKA family-wise error rate

Can be corrected by setting higher threshold value to reach significance

Type I error

False positive (α)

Type II error

False negative (β)

Statistical power

Probability of a test correctly rejecting a false null hypothesis

Power = 1 − β

Many studies in psychology are underpowered → poor replicability

Meta-analysis

Combining evidence across studies

Pooling results from multiple studies

Increases statistical power

Funnel plots

Addressing bias

Statistical solutions

• Regression adjustment

• Inverse probability weighting

• Study design improvements

Addressing confounding

Analytic approaches

• Regression

• Stratification

• Standardization

Confounder

Distorts relationship through real but noncausal association

Mediator

Explains mechanism

Third variable

Moderation

When effects differ across groups

• Varies by age, gender, context

Path analysis and advanced mediation models

Involves multiple mediators and multiple outcomes

Examines direct effects and indirect effects (via mediators)

Based on a theoretical causal model

Can include confounders and longitudinal (repeated) data

Missing data

MCAR: Missing Completely at Random)

• Unrelated to nay variables

MAR: Missing at Random

• related to observed variables

NMAR: Not Missing at Random

• Related to unobserved or missing values

Can be handled through multiple methods:

• Complete case analysis

• Full Information Maximum Likelihood (FIML)

• Multiple Imputation (MI)

  • Creates multiple datasets

  • Pools results

Measurement error

Attenuates associations

Can bias estimates

Common sources: self-report and observer ratings

Latent variables

Represent unobserved constructs

Measured via indicators (e.g. depression symptoms)

Uses factor analysis

Reduces measurement error