BRM (Fach) / Types of research (Lektion)
In dieser Lektion befinden sich 14 Karteikarten
Types of research • reporting: summarize information • descriptive: describe an issue • explanatory: explain the reasons for a phenomenon • predictive: predict the likelihood or magnitude of a future phenomenon → Most studies are explanatory and/or predictive
Diese Lektion wurde von Juma erstellt.
Diese Lektion ist leider nicht zum lernen freigegeben.
- Types of research reporting: summarize information descriptive: describe an issue explanatory: explain the reasons for a phenomenon predictive: predict the likelihood or magnitude of a future phenomenon → Most studies are explanatory and/or predictive
- Criteria for good research research purpose clearly defined details of procedures provided research design throughoughly planned high ethical standards applied research limitations frankly revealed appropriate data analysis delivering sufficient findings research findings clearly and logically presented conclusions justified and matched with findings researcher's experience and credentials reflected
- Positivism early 19th century, philosophy, sociology and natural science, only observable phenomena can provide credible data, focus on causality and law-like generalizations, value-free way of research, researcher objective and independent of data, high structured, large samples, measurement, mostly quantitative, may use qualitative data for explanations
- Interpretivism: various 20th century philosophy and sociological streams, symbolic interactionism and social constructionism subjective meanings and social phenomena are of interest focus upon details of a situation and a reality behind these details, subjective meanings motivating actions value-bound research, researcher subjective and part of what is being studied small samples, in-depth investigations, narrative qualitative data, rarely supplemented by quantitative data
- Deduction Top-down testing theory reasons (premises) lead to conclusions about specific cases general premises explain specific experience (data)
- Induction Bottom-up building theory specific pieces of evidence suggest a conclusion and explain general theory
- 4 Aims of a literature review demonstrate knowledge on the field develop and refine research questions build on and evaluate previous work indentify gaps, inconsistencies and research suggestions
- Four types of critique Rhetoric: evaluate whether arguments are logical and if language is used in a misleading way, use reflective skeptism, offer alternatives, have a purpose with critic Tradition: question fundamental assumptions and conventional wisdom, reduce limits set by the powerful of what may be debated Authority: question dominant views of the privileged and appreciate genuine disagreements, recognize the plurality of different perspectives, try to see the world through others eyes Objectivity: be skeptical of available information, not value-free usually, is the information suppressed by interest group?
- Research design Qual. vs. quant. degree to which research question has to be formalized: exploratory vs. formal, power of researcher to influence variables: experimental vs. non-experimental, purpose of the study: descriptive vs. causal vs. predictive, time dimension: cross-sectional vs. longitudial, method of data collection: monitoring vs. interrogation topical scope: case vs. statistical analysis participants perceptions: actual routine vs. modified routine
- Sampling simple random sampling: randomly draw individual elements from the entire population complex probability sampling: more quantitative, more biased o systematic sampling (determine sampling ratio e.g. from whole population) o stratified sampling (determine different strata e.g. lines of business in a company) o cluster sampling (clusters & randomly sample entire clusters, families of one block) o double sampling (at least two stages, two phases of sampling) non-probability sampling: more qualitative o convenience sampling (choose from friends, neighbors) o purposive sampling (judgment sampling – one predetermined criterion) o snowball sampling (initial identification of ind. Who then locate others )
- Reliability consistency of a measure across repeated administrations o test-related reliability (same measure to same respondents two different times) o inter-rater reliability (extend to which two or more raters agree) o equivalent-form related reliability (two parallel forms of test show same results) o internal consistency reliability · split-half reliability: divide questionnaire into odd numbers and even numbers, two scales per person, assess degree of similarity · Cronbach's alpha reliability: how well multiple measures on the same person agree (e.g. questionnaire items), response to each item is related to response to every other item in the same measure, degree of homogeneity across items (from 0.7 to 1 the measure is seen as reliable)
- Validity degree to which a measure captures what is intended to measure, accuracy of drawing inferences regarding what a score means construct validity (confidence in the interpretation of what a measure means, degree of which a measure captures the underlying theoretical construct) o measure should be related to measures of similar constructs = convergent validity o unrelated to measures of different constructs = divergent validity) face validity (from a laypersons perspective, is really measured what was intended?) content validity (degree to which a measure captures the entire domain of a variable) criterion related validity (degree to which a measure forecasts a criterion = predictive valid.)
- Exploratory Factor Analysis data reduction method (e.g. grouping of many questionnaire items into groups) Eigenvalue: How much of the total variance of all variables (items) is accounted for by a specific factor o Kaiser criterion (retain all factors with an Eigenvalue >1) o Scree test criterion (retain all factors before the first one where Eigenvalue starts to level off) Factor loading: correlation between a variable and a factor communality: extent t which all factors together explain one variable (questionnaire item) Rotation: tends to clean up the factor structure
- Confirmatory Factor Analysis measurement model of a structural equation analysis: tests the fit between the obtained data and a predetermined theory-based factor structure as compared to other models good fit: chi-square x2 should be as small as possible, chi-square divided by degrees of freedom df <2 relative fit indexes such as GFI (Goodness of Fit Index) and NNFI (Non-Normed Fit Index), the same as TLI (Tucker-Lewis Index) should be >0.9 (How much variance is accounted for by your model – mehr als 90% ist gut) RMSEA (Root Mean Squared Error of Approximation) <0.08