Please enable JavaScript.

Coggle requires JavaScript to display documents.

M&T - Coggle Diagram

- - - - relates to external validity
      - sample size
    - - variant: multistage
    - - e.g internet resaerch/animal research
      - college students
      - pretty common, just take threats into account
    - - matching
      - blocking
  - - - make participation interesting for all sorts of people :check:
  - - - recruit from avalialable population
      - sample pool
    - - selection in natural environment
      - good external validity :check:
      - internal validity :red_cross:
  - - - random assignment
      - restricted random assingment=> matching & blocking
    - - assignment bias
      - non-specific events
- - - - basic method
      - test-retest reliability
      - parallel forms reliability
      - split-half reliability
  - - - ceiling effect
      - floor effect
  - - - performance cues
      - role attitude cues
      - cooperative attitude
      - apprehensive attitude
      - negative attitude
      - validity threat! :red_cross:
    - - expectancy effects :red_cross:
        
        => treats groups unequivalently
      - solution
        
        blinding techniques :check:
    - - solution: blind observer :check:
- - - - correlational = obersavtional research
      - naturalistic obseravtional
        
        ethnography
        
        sociometry
        
        case history
        
        =demonstration
        
        can be single subject design
      - archival research (?)
      - quantifying behaviour
        
        frequency method
        
        duration method
        
        interval method
      - coping with complexity (?)
        
        time sampling
        
        individual sampling
        
        event sampling
        
        recording
    - - time-series designs
      - equivalent time samples
      - developmental designs (age= q.i. var.)
        
        cohort sequential
        
        longitudinal
        
        carryover & history effects :red_cross:
        
        = threat to internal validity
        
        cross-sectional
    - - control groups
        
        no-treatment
        
        placebo
      - baseline designs
  - - - variant: pretest-posttest design
        
        no counterbalancing possible-> carryover! :red_cross:
        
        control group
        
        = automatically mixed design
        
        gets some internal validity :check:
        
        still= pretest sensitization :red_cross:
        
        solomon four group design
        
        or leave out pre-test
        
        true-experimental design
      - carryover effects! :red_cross:
        
        counterbalancing
      - single factor two-level design
      - progressive error :red_cross:
        
        enough time betw. trials
    - - solomon four group design
      - two independent variables
        
        between-subjects: what differs groups
        
        within subject var:: what groups have in comon
    - - baseline design
        
        ABAB - intrasubject replication
        
        use multiple persons to establish external validty
        
        similar: pretest-posttest design
        
        AB- design
        
        risk of confounding! :red_cross:
      - watch out: uncontrolled variability :red_cross:
        
        tight experimental control!
    - - single factor (random assignemt)
        
        randomized multigroup
        
        variation: multiple control group design
        
        randomized two group
      - ways to manipulate ind. var.
        
        quantitative manipluation; parametric designs-> ordinal
        
        qualititative manipulation: nonparametric -> nominal
      - matching to groups
        
        matched two-groups / pairs
        
        matched multigroup
        
        why? ruling out confounding variables betw. subjects
        
        internal validity :check:
  - - - q.i. reduces error variance :check:
      - don´t conclude causality from q.i. designs !
    - - main effects (on average on all levels)
        
        simple main effect= only on specific level of i. var.
      - interaction
      - more than 2 var. = higher order f.d.
  - - - remedies: debrief/-hoax, role playing & prior consent
      - when used?: cost-benefit analysis or social contract theory
      - types: passive/active & intentional/unintentional
    - - APA ethical guidelines
      - belmont report
        
        respect for persons
        
        beneficence
        
        justice
      - local/national/international laws
    - - information-consent issues
      - anonimity & confidentiality
    - - IACUC
      - cost-benefit assessment
    - - how?
        
        data fabriction
        
        falsification
        
        plagiarism
      - look for reasons in notes
      - solutions
        
        grant-view process
        
        peer-review process
        
        replication
        
        education
- - - - history
      - maturation
      - testing -> pretest/posttest
      - instrumentation
      - statistcal regression
      - biased selection of subjects
      - experimental mortality
    - - rule out confounding var.
        
        matching/blocking
        
        hold extraneous var. constant
      - identify rival hypothesis beforehand
      - check valididty, reliability, accuracy, adequacy of measures used
      - pilot study
      - be aware of volunteer bias
      - rule out pretest-sensitization
        
        solomon four group design
        
        control groups
  - - - reactive testing
      - interaction betw. participant selection bias nd independent var.
      - reactive effects oxperimental arrangements
        
        laboratory settings
      - multiple treatment inference
    - - avoid pretests- if needed solomon four group design
      - use different settings- (field settings, simulations)
      - right sampling methods
      - avoid carryover effects-
        
        practice sessions
        
        counterbalancing!
        
        make treatment order an independent var.
    - - usually contradict each other
  - - - reduce erroring variance
        
        hold extraneous variables constant
        
        match subjects on crucial characteristics :check:
        
        can reduce external validity
        
        for within subject design- use same subject for all levels of independent variable
      - increase effectiveness of independent variable
        
        select correct levels of independent var.
        
        choose sensitive enough dependent var.
      - randomize error variance across groups
      - use statistical analysis to infer error variance
    - - sources
        
        learning
        
        fatigue
        
        habituation
        
        sensitization
        
        adaption
        
        contrast
      - dealing with it
        
        counterbalancing :check:
        
        partial: random order with latin square
        
        complete balancing: all possibilities
        
        make treatment order an independent var.
- - - - no interaction- study main effects
      - interaction: study simple effects
- - - - simple language
      - infleunces on interpretation
        
        time interval e.g.
        
        order of questions
      - question format
    - - perception
      - memory
    - - partial memory
      - wording of Q:
      - context of Q
      - subjective theories
    - - response scales
        
        visual analogue scale
    - - demand cahracteristics :red_cross:
        
        cooperative attitude
        
        negative attitude
      - social desireablility
  - - - temporal consistency
        
        test-retest reliability
      - internal consistency
        
        split-half reliability
    - - face validity
      - content validity
      - criterion validity
      - criterion validity
        
        concurrent validity
      - construct validity
  - - - reduces validity :red_cross: not reliability
    - - reduces validity and reliability :red_cross: