Test 3 Part 1 Flashcards
Qualitative vs quantitative data
Qualitative: rich data, finding meaning (identifying themes) through interpretation
•rarely has hypothesis, limited to specific contexts (chosen for a purpose-small)
•ex: history, english, languages, counselling psych, philosophy
•case studies (psychobiography), naturalistic observation, phenomenology, appreciative inquiry (search for the best in people)
Quantitative: numeric data, finding measurement (turn behaviour into #) through prediction
•almost always hypothesis, broad/generalizable - large/ representative of population
•ex: psychology dept., bio, chem, physics, econ, math
•experiments, close ended surveys, quasi-experiments
Action research •qualitative: focus groups •quantitative: numerical records Content analysis of archives •qualitative: themes •quantitative: systematic coding
Blurry:
Qualitative data can be coded quantitatively
•systematic observation: counting frequency of certain behaviour
Qualitative data isn’t necessarily objective
•rating scales
Naturalistic Observation
Researcher immerses self in a natural setting to collect information overtime
Goal: accurately describe experiences in a particular setting
Qualitative approach: observes without hypothesis in mind (can use mixed gathering some qualitative data - income, family size)
Concealed vs non concealed: balance ethics vs participant reactivity
•features: field notes, narrative approach (reflect chronological order), grounded theory - organized around the theory developed by researcher
Issues
1. Participation and concealment
•participant observation: allows the researcher to observe the setting from the inside, experiencing the events like participants
*issue: may lose objectivity
•concealed observation: may be preferable because the presence of the observer may influence the behaviour of those being observed (less creative), and doesn’t come with potential ethical issues
Limits
•qualitative: most useful when investigating complex social settings
•quantitative: gather data in real life settings and generating hypothesis for later experiments
*the inability to control the setting makes it challenging to test well defined hypotheses under such specified conditions
*can’t control time and place (very time consuming because researcher must determine what information is important in so much info and create themes/code data into meaningful categories)
Systematic observation
Specify observable categories of behaviour by observing people in a relevant setting/situation
•situation: can be contrived or natural
•location: can be lab or field
Code behaviour into specified categories
•can be live or videoed
Qualitative (typically involves hypothesis)
ex: infant attachment styles (strange situation)
•avoidant ambivalent secure
•1(baby makes no effort) to 7(baby makes extreme effort) scale measuring proximity, contact, resistance, and avoidance
Coding system: researchers decided whats important and develop a coding system to measure such behaviours
•purpose: to summarize qualitative observations
•should be as simple as possible
•can use new coding systems, or ones developed by others (already valid)
*FACS (facial espresso coding system) or MICS (interactions of family members during meals
Issues in systematic observation
•Reliability: uses inter rater - requires a lot of training, difficult to be high unless u have a really simple coding system (videos are better)
•Reactivity: the presence of the observer can change behaviours (can hide cameras)
•Sampling: researchers must make decisions about how to sample behaviours
Case studies in qualitative research
Intense study of a single person, group, incident, or community
•can be qualitative or qualitative
•ex: interviews/focus groups, surveys, naturalistic observation, analysis of archival data and single case quasi experimental designs
Generally, case studies have
•low internal validity because they can’t make causal claim (too many alternative explanations)
*adding reversals may help if possible
•low external validity (generalizability)
*adding multiple baselines helps generalize to other situations/people
Compare/contrast systematic and naturalistic observation
- Systematic observation is much less global, and is used more often in quantitative studies
- before systematic observations/coding, can have a naturalistic observation to help them come up with categories for the coding system
Research designs for special circumstances
- Single case: ABA reversal design, ABAB reversal design, and multiple baseline design
•ex: therapist testing if theory produces patients symptoms. wanting to train your dog to stop peeing, researcher exploring decision making of someone with a specific brain injury - Quasi-experiments: dress the need to study the effect of an IV in settings which the control of true experimental designs can’t be achieved
•can’t make causal claims (lack control conditions and random assignment) - low internal validity
•Book: one-group post-test only, one group pretest-posttest, nonequivalent control group, nonequivalence control group pretest-posttest multiple repeated measures (interrupted time series, control series)
•ex: community agency evaluating how well an intervention program works, teacher evaluating whether a classroom activity enhances learning, gov, evaluating new law effectiveness
•involves one group only post test design (participant, IV, DV), one group pretest post-test design (participants, DV pretest, IV, DV post-test), non-equivalent control group design (has a control group that isn’t equivalent), non-equivalent control group pretest posttest design) - Developmental: cross sectional, longitudinal, sequential
•ex: researcher comparing attention span of older vs younger people, university tracking critical thinking skills as students progress through their degrees
features that help them strive toward internal validity
Single case experimental designs
- Single case experimental designs
•baseline measurement (test 1 and 2 scores) then treatment (test 3 score) - was there an improvement
*alternative explanations: - Single case reversal design: baseline (test 1) treatment (test 2 - improvement?) and baseline again (test three - was there a drop)
- Single case multiple baseline design
•course 1: baseline measurement (test 1) and treatment (test 2 - improvement?)
course 2: baseline measurement (test 2 score) and treatment (test 3 score - improvement?)
3 criteria needed to establish causal relationship (continuum) 1. Temporal precedence •theres an order 2. Covariation of cause and effect •presence=presence, absence = absence 3. Eliminate alternative explanations
Threats to internal validity
Eliminating plausible alternative explanations
1. History: something happens in the world that would affect most/all of our participants at the same time as the onset of treatment and could have caused the effect
•alternative explanation for changes: multiple groups who experience the event and it affects them differently, becomes confound
•ex: natural disaster, media event
- Maturation: most/all participants change between pretest and post-test for reasons not of interest to the researcher
•alternative explanation for changes: if there are multiple groups and the groups mature differently
•ex: smoking- caring more about health as you age - Testing: the act off taking the pretest changes how most/all participants respond to post-test
- Instrument decay: over repeated use, treatments or measurements change, making it look like your treatment had an effect when really it was the operational definitions that were changing
•forgetting to record a smoke towards the end of the study
Archival data: content analysis and interpretation
- Content analysis is the systematic analysis of existing documents
- Used to quantify information in archival documents
- Often done with computers now (LIWC)
- Archival data allows researchers to study interesting questions: enhanced ability to generalize results
Problems with archival data: desired records may be difficult to obtain, we have no control over what data were collected and the way it was recorded (accuracy?), and many alternative explanations for observed relationships exist - so we cannot make causal claim
Why true experiments have higher internal validity
- having a control group (can be helped with replicability)
- having random assignment
- harder to detect a true effect
