Quantitative Karina General Flashcards
what are the four main types of quantitative data gathering?
Survey
Observation
Behaviour / implicit
Biological / physiological / neurological
What are the two main sources of bias?
participant
researcher
what kinds of participant bias?
- social desirability
- demand characteristics
what kinds of researcher bias?
- experimenter expectations
- selective reporting of data
- rare occasions – falsifying data
what types of validity are there?
internal
external
ecological
what is Internal Validity of Design?
extent of cause-effect claims,
and ability to rule out alternative explanations
what is external Validity of Design?
External Validity of Design – extent to which findings can be generalised
what is a Non-experimental Research Designs
?
• Examine relationships between naturally occurring characteristics of individuals or events
however in Non-experimental Research Designs… you cannot….
No cause-effect claims
what are the advantages ?
- Studying phenomena that cannot be experimentally manipulated
- Examining ‘real behaviour’ in ‘natural setting’ (although can be conducted in lab)
- Avoids demand characteristics, participant roles, experimenter bias
what are the disadvantages?
• Cannot form confident causal inferences • Direction of causality:
X → Y or Y → X?
¥ Third variable problem: Z → X and Y
¥ Person, environmental, and operational
confounds (think of some of the confounds in a study examining the relationship between exposure to violent media and aggressive behaviour)
3 main types of non-experimental research?
- Correlational research (observation/survey research)
- Archival research
- Case studies
what are the key points to observational research?
• Unobtrusive observations best
- If possible, make audio and visual recording
- Quantifying data:
Ð Develop a coding scheme; clearly and simply define behaviours; use clear operational definitions
• Have at least 2 independent observers – measure inter-rater reliability
what are the Measures used to quantify behaviour in obs research?
- frequency – number of times in a period
- duration – how long behaviour lasts
- interval – whether behaviour occurs in specific period
How/when to sample behaviour?
¥
- time sampling – observation/recording/observation
- individual sampling – individual selected
- event sampling – observe one behaviour, record all instances
what is cross sectional design?
• Create ‘cohort’ groups based on different ages of participants ‘quasi-IV’
advantage of cross sec design?
¥ Quick, cheap, collect developmental data in short period
disadvantage of cross sec design?
¥ Generation effects may be a problem (e.g., measure substantial differences in IQ with age - 20 to 70 - because of education)
what is longitudinal design?
Ð Same set of participants measured several times (e.g. over months, years)
what are the 3 advantages of long design?
Ð Avoids possible generation effects
Ð Development trajectory clear
Ð Stronger claim for causality
4 main issues with long design?
1) participant attrition/mortality
2) time + money
3) multiple observation effects
- threat to internal validity e.g. increase in Iq might be because of education and training toward test used – also, many factors are confounded with age
4) cross generational effects (conclusion drawn from one study generation may not generalize (e.g 1910–>1950 vs 1990–>2017)
- Possible cross-generational problem
- Results from one generation may not generalise to another
what is Cohort-Sequential Design
?
¥ Combines cross-sectional and longitudinal component in same design
benefit of Cohort-Sequential Design?
¥ Allows some disentangling of age, cohort, time
you can also test for …
¥ Possible to test for generation effects, though does not eliminate them (lets you detect and consider them)
age is a xxxxxx variable so no xxxxxx can be attributed to age
age is a quasi- experimental variable so no cause can be attributed to age
what is a single - case design?
Ð Consists, in the limit, of one participant (N=1 design) but typically 3-6 participants used to establish generalisability/external validity
Ð Includes stringent control
Ð Systematic effect of treatments/conditions over multiple observations establishes cause and effect
what are the details of the case-study approach? 4 points
• Case-study approach
Ð Observational technique of observing single participant
Ð Often used in clinical settings
Ð Descriptive or observational approach
Ð No manipulation or control of variables; simple recording of observations
when to use single case design?
Ð Single-case designs powerful where behaviour of single cases/participants can be observed under different treatments/conditions, with multiple observations per treatment/condition, allowing momentary fluctuations to average out effects of treatments are reversible and/or carryover
what is baseline design?
- Individual participants observed under each of several ‘treatment’ phases or conditions
- Multiple responses recorded in each phase before next phase (i.e., time series)
- Baseline phase of extensive observations
- Repetition, or intra-participant replication, of treatment effects establishes reliability of findings
in baseline design you need to establish
• Adopt a stability criterion
- guarantees participant will stay in a phase until observations no longer show systematic changes
- criterion should not be too strict or loose
- removes ‘transitional’ behaviours from analysis
• Deal with uncontrolled variability
- sources need to be identified and controlled
and determine a level of internal
Determine generality of findings - inter-participant replication
• Generality across situations requires study replication
- direct replication – exact replication
- systematic replication – incorporate variations on original
what are the four issues with baselines?
• Drifting baselines
- baseline that does not stabilise but continues to show systematic variations (drift)
- take drift into account, and effects of treatment can maybe still be determined
• Unrecoverable baselines
- behaviour cannot be returned to original baseline after treatment (carryover effects; treatment effect may still be manifest)
• Unequal baselines between participants
- different participants level off at very different baseline values (e.g., rats spontaneously press lever different number of times depending on individual levels of motivation)
- steps may be taken to equalise baselines (e.g., increase deprivation of all individuals so lever pressing high in all, and all have opportunity for learning)
• Inappropriate baseline levels
- baseline levels that are too high or low may mask effects of treatment (similar to range effects)
- levels have to be adjusted to appropriate levels
most common types of baseline designs?
Ð Single-Factor Design
Ð Most common–ABA(baseline ,treatment, baseline)
Ð Demonstration of treatment effect requires return to baseline
Ð To establish intra-participant replication (and avoid ending on baseline in clinical contexts) can extend to ABAB
Ð ABAB can be extended to accommodate multiple levels of IV – e.g. ABACBC
- need to counterbalance to deal with carryover effects
Single-case Designs - Advantages ?
¥ Focus on tightly controlling error variance
¥ Focus on individual behaviour:
- makes identifying and controlling sources of error variance easy
- may reveal critical mechanistic details lost with group approach (e.g., attention & transfer of learning about compound stimuli)
¥ Causal relationships can be established with very few participants
Single-case Designs - Disadvantages ?
¥ Making multiple observations is time- consuming, can be tedious
¥ Not appropriate for all research questions (e.g., developmental and social psychology!)
¥ Results may be of limited generality
¥ All variables that can cause error variance cannot be identified and controlled
what is Construct validity?
Construct validity is “the degree to which a test measures what it claims, or purports, to be measuring
what is Test validity?
Test validity is the extent to which a test (such as a chemical, physical, or scholastic test) accurately measures what it is supposed to measure.
what is content validity ?
In psychometrics, content validity (also known as logical validity) refers to the extent to which a measure represents all facets of a given construct.
what are Convergent and discriminant validity ?
Convergent and discriminant validity are both considered subcategories or subtypes of construct validity. The important thing to recognize is that they work together – if you can demonstrate that you have evidence for both convergent and discriminant validity, then you’ve by definition demonstrated that you have evidence for construct validity. But, neither one alone is sufficient for establishing construct validity.
measures of constructs that theoretically should be related to each other are, in fact, observed to be related to each other (that is, you should be able to show a correspondence or convergence between similar constructs)
and
measures of constructs that theoretically should not be related to each other are, in fact, observed to not be related to each other (that is, you should be able to discriminate between dissimilar constructs)
