Saturday, January 31, 2009

week 4 measurement

Blog Question: What distinguishes Quantitative from Qualitative designs,
what is the difference between “validity” and “reliability,” and what is
meant by the terms “probability” and “significance?”

According to Goubil-Gambrell, quantitative research tries to attach numerical values to variables, populations, samples, etc. and show their relationship to each other. When conducting quantitative inquiry, researchers perform experiments that manipulate the variables for analysis in order to establish causal links. Qualitative research is descriptive and identifies variables in light of circumstances that frame inquiry. Lauer and Ashe add that qualitative research is not concerned with treatments or manipulating control groups.

Frederick Williams offers an analysis of validity and reliability and provides key distinctions. The main difference is that validity requires some standard outside of the component of measurement to which comparisons are made, whereas reliabiltiy is concerned with "comparing a measure with itself," but one does not necessarily imply the other. In other words, validity questions the fitness of a researchers tools to that thing he/she claims to measure and reliability has to do with the replication of research, which may or may not be consistent based on the component "subparts."

Probability means that an outcome will occur in all likelihood due to the necessary factors being in place, whereas significance occurs based on circumstances that are not necessarily present. In the words of Lauer and Ashe, "significance in statistics is a statement of the degree of rarity of a result based on chance probability alone. In almost all cases of statistical analysis, the calculated statistic is compared with standard, tabled values of chance distribution of that statitistic. If the calculated value is sufficiently larger (for almost all statistics) than the tabled value, the result is called statistically significant, meaning that the statistical relationship between two variables observed is unlikely to have occurred simply by random chance alone. Significance is declared usually at a probability (p) level of five or one percent" (287, emphasis mine).