Exam II Study Cards Flashcards
Selection of Design
Consider feasibility of conducting study
Intervention Fidelity
Constant data collection
Experimental Design Features
Randomization
Control Group
Manipulation
Advantages: most appropriate for testing CAUSE AND EFFECT RELATIONSHIPS
Disadvantages: costly and difficulty in field
Experimental Design
For testing Cause and Effect relationships
Positivism
Real Natural Truth by observation Context minimized Neutral observer Experimental
Constructivism
Subjective Multiple realities Culture and environment Context emphasized Active participant w/ dialogue
Hawthorne Effect
Trial participants act differently to get attention
or please researcher
Quasi-Experimental Design
MANIPULATION OF A VARIABLE
- not random
Layouts:
- Non-equivalent (foam vs soap)
- After only non-equiv (no baseline data)
- one-group (pre/posttest; nurse residencies)
- Time series (some baselines, interventions applied and info gathered afterward)
Advantages: practical and feasible
Disadvantages: unable to make clear cause/effect statements; may not be able to randomize
Non-Experimental Design
No independent variable b/c already occurred
(smoker already smokes)
- concepts of control still observed w/in a cohort
Advantages: IMPORTANT when randomization, control, and manipulation are not appropriate or possible
Survey Studies (CDE)
Comparative
Descriptive
Exploratory
Correlational Studies
Relationships between variables
but does not determine cause and effect
Solomon-Four-Group Design
Tests AGAINST the THREAT of INTERNAL
VALIDITY of INSTRUMENTATION
Extraneous Variables
AKA Intervening/Mediating
Interfere w/ operations of variables being studied
Control w/:
homogenous sample; consistent data collection; training and supervision; manipulation of IV; randomization
Threats to Internal Validity
History; Maturation; Testing; Instrumentation;
Mortality; Selection Bias
History
Historical events that skew results
Maturation
Respondents change as time passes;
external to investigation
Testing
Repeated testing influences responses in the
future
Instrumentation
Measurements affected by instrument
calibration
Mortality
Loss of subjects from pre to posttest
Selection Bias
Respondents change own mind about
participating
External Validity
Questions conditions of findings
Internal Validity
Asks if IV caused or resulted in change in DV
“Truth Value”
Rules out threats as rival explanations of relationships between variables
Consider BEFORE PLANNING A STUDY to not negate results
Population
Well-defined set
Sample
Subset of units from a population
Accessible Population
Group of people available to make a sample
Exclusion Criteria
Restrict population to HOMOGENEOUS group
of subjects
Inclusion and Exclusion Criteria
Establish control for extraneous variability or bias that would limit the strength of evidence contributed by the sampling plan.
Careful establishment of criteria INCREASE PRECISION of the study and STRENGTH of EVIDENCE, contributing to ACCURACY and GENERALIZATION of the findings
REPRESENTATIVENESS
Key characteristics CLOSELY
APPROXIMATE those of the population;
foremost criterion in appraising a sample.
Who is CONVENIENCE SAMPLING
Any subject available that meets criteria and
will participate
Advantages: easy to draw
Disadvantages: HIGHEST
RISK OF BIAS b/c samples may volunteer based off of interest in study
QUOTA SAMPLING
-Fill spots with proportionate strata
represented (40/40/20)
-Oversampling to satisfy survey size
-BIGGER SAMPLE MORE REFLECTIVE OF TRUTH
-RECRUITMENT and ENROLLMENT.
-Representativeness INCREASED b/c prob of over/underrepresentation addressed.
-Characteristics of population selected according to researcher’s judgment and Lit review
PURPOSIVE SAMPLING (QUAL)
- MORE heterogeneous the population, -GREATER RISK of bias introduced
- Sample HANDPICKED so limited ability to generalize
PROBABILITY (RANDOM SELECTION)
Take the Simple and STRAight way to not get in a Cluster
- SIMPLE, STRATIFIED, AND CLUSTER
- AVOIDS BIAS but more time consuming
- Each subject has an EQUAL CHANCE OF BEING CHOSEN
- CLOSELY ASSOCIATED W/ EXPERIMENTAL AND QUASI
- STRONGEST TYPE OF SAMPLING
- Different then RANDOM ASSIGNMENT*
RANDOM ASSIGNMENT
Subjects ASSIGNED to experimental or control group on a RANDOM BASIS
SIMPLE RANDOM SAMPLING
- CAREFULLY CONTROLLED
- Researcher DEFINES the POPULATION, LISTS the UNITS of the pop (SAMPLING FRAME), and SELECTS a sample of units (SUBSET) from which the sample will be chosen
- LABORIOUS, TIME CONSUMING, INEFFICIENT
- CAUTION w/ reporting generalized findings, esp if pop difficult to list completely
- LOW BIAS
- REPRESENTATIVENESS MAXIMIZED but no guarantee
- Differences function of CHANCE
- Probability of non-representativeness DECREASES with INCREASED SAMPLE SIZE