1) Count 4-6 possible confounding factors in the realization of a Research designs on.
- unobserved heterogeneity
- misspecification
- between temporal events
- change the person
- test-retest effect
- Instrumentation
- selection (random, mortality, missing data)
2) What is meant by an independent variable and what it means by
dependent?
There is a causal relationship between variables. The cause is then defined as independent variables, the effects however, as the dependent variable. Since causality only in a time sequence of cause (before) and effect (after) is possible still applies: The independent variable must occur before the time-dependent. need to correlate the variables and they must form a closed system, ie, no apparent causal variables created by third
3) What is referred to as "selection bias" bei der Realisierung eines
Forschungsdesigns?
Als Selektionsfehler bezeichnet man bei der Realisierung eines Forschungsdesigns eine grundsätzliche Veränderung der Zusammensetzung der Stichprobe. Dies kann durch Zufall geschehen oder systematisch. Besonders in letzterem Falle besteht die Gefahr, dass eine erneute Befragung nicht durch den Einfluss einer unabhängigen Variable anders ausfällt, sondern durch eine unrepräsentative Stichprobe (im Sinne der Vergleichbarkeit) verfälscht wird.
4) Examine to the internal and external validity of social scientific investigation.
Considerations internal validity deal with questions such as:
- How far is the variation in the AV on the variation in the UV due?
- How good is it managed to find in the study only the effects of UV in the AV?
- can be ruled out possible alternative hypotheses for the explanation of the UV-V variation?
An experimental plan to be valid internal if the variation of the AV (up to unsystematic error variance FV) goes back only to the variation of the UV.
core problems of external validity the Populationsvalidität and ecological validity (eg, If something else came out, when other people, at different times in different situations (Umgebungen) an anderen Plätzen untersucht worden wären?). Populationsvalidität bezieht sich auf Versuchspersonenrepräsentativität und kann stichprobentheoretisch noch einigermaßen hergestellt werden. Schwieriger wird es, über Situationen zu generalisieren. Dies ist praktisch unmöglich, da die Population der möglichen Situationen in den wenigsten Fällen geklärt erscheint.
5) Was ist Untersuchungsmortalität? In welchen Forschungsdesigns spielt sie eine Rolle?
Das Problem der Untersuchungsmortalität goes back to the selection bias in longitudinal studies or multiple surveys. This raises the question whether the mortality changes the sample in its basic composition, such as systematically as they occur. Then, the survey results would be falsified.
6) What are reactive measurement effects?
various factors of a social scientific research design: reactivity or reactive effects of measurement: Through the implementation of the pre-test or the first survey, the susceptibility of the proposed subject for the stimulus and will lead to changes in measurement results.
7) Describe four strategies to minimize confounding factors in the realization of an empirical research design.
- real elimination : eg presence of third parties not allow
- constant : To ensure that the observed effect is due to the variation of the independent variables, tried all other factors held constant. As the natural light from day to day and varies throughout the day, tests should be used to visual perception in a trial of all passages of time equal to the illuminated laboratory are carried out.
- randomization : the assignment of subjects to experimental and control group randomly happens. This ensures that the differences between groups at a sufficiently großen Stichprobe ausmitteln. Durch Randomisierung wird ausgeschlossen, dass es durch die Aufteilung der Versuchspersonen in Experimental- und Kontrollgruppe zu systematischen Verzerrungen der Ergebnisse kommt.
- Matching : Verfahren zur Bildung von Gruppen, die bezüglich eines Störfaktors oder mehrerer Störfaktoren homogen sind. Soll zum Beispiel eine Lehrmethode evaluiert werden, so können durch Parallelisierung zwei hinsichtlich ihrer Noten möglichst ähnliche Schülergruppen gebildet werden
- replication : repetition of experiments, because the more often an experiment was repeated, the more reliable are the results (law of large numbers)
8) Describe the differences and advantages and disadvantages of field and
laboratory research?
- independent of the design question
- laboratory is artificial, but also field research does not cover everyday behavior
- Often in the laboratory, a context-relative- independent behaviors studied (but general laws should not let it affect)
- contribute confounding variables everywhere
- laboratory studies can be achieved easily, but there is the problem of large numbers (then often too complex and therefore expensive)
9) What are the 2-3 problems have to deal with a survey design
?
- problem of finding variance : enough cases with / without feature, not skewed distribution of continuous variables, rare features, disproportionate sampling, analysis problems in stark inequality à weighting need
- causality problem : Does A to B or B to A? Genesis of the problem setting, it is only a link and possible Kovariationsaussagen
- control of third-, intervening and Supressorvariablen: not observed UVs that affect the AV secretly
o antecedent variables: X à UV à AV
o intervening variables: UV à X à AV
o Supressorvariablen: UV à Xi (+) and Xi (-) à AV (UV can have different effects in 2 groups, so that in average no change comes out. This hides the Supressorvariable the effect of the treatment)
10) What is the causality problem in survey design? What other problems have survey designs?
One of the problems of ex post facto arrangements as the survey design is that the independent and dependent variables can not be uniquely determined (Does A to B or B to A?). For the assessment of causality a clear chronology of cause and effect is necessary, by using a single measurement is not possible. At most one can attempt to simulate in the survey to retrospectively a before-survey, which is not always possible and also a high source of error that can lead to distortions forms. Other problems include the problem of finding variance and the problem of introducing control variables. (See previous question!)
11) How influential between the influence of a third variable and the empirical content of a theory?
third variable (distorted choices and failures) can provide a false causality, and thus distort the empirical content of a theory. (See question 9, item 3)
0 comments:
Post a Comment