Monday, July 21, 2014

Probability vs Non probability sampling

Probability sampling means using any sample that has used random selection from the population and non probability sampling is the contrary process. Sample selection is a very important part of a survey research study and a unsuitable sample selection can lead to incorrect inferences about the population under consideration (Groves at el, 2009).
Probability sampling assigns an equal or a nonzero chance of selection to all elements of population (Groves et al, 2009) and this is also the best advantage of this method. According to Groves et al (2009) the benefit is that we can use one sample ‘realization to estimate the characteristic of the frame population’. Frankfort-Nachmias & Nachmias (2008) write that probability sample is the ‘only’ method that can be used representative sample designs. Representative sample designs is an analysis made using sample units that give results similar to those results that would have been obtained from the whole population. The distinguishing factor of probability sample designs that it gives the same probability to each unit to be included in a sample is also its best advantage.
A well designed sample makes certain that if the study was conducted repeatedly with different samples then the same results will be obtained with minimum variation (Frankfort-Nachmias & Nachmias, 2008). Probability sample designs make it possible for the researchers to estimate the degree of deviation if the study was repeated. Rao,Hartley & Cochran(1962) reports that the probability method is the simplicity of calculation and calculating the variance of the estimator.
Top of Form
Bottom of Form
Top of Form
Bottom of Form
I would like add an interesting point to this comparison of probability and non probability sampling: the probability sampling techniques used in a quantitative research are ‘rarely appropriate’ when conducting a qualitative research study of the same nature (Marshall, 1996). I would also like to add that it is not uncommon for social scientists to use non probability designs for various reasons: study of an unspecified population or convenience, cost outweighs other factors.
My final project aims at understanding if a relation exists between success in life and growing up as an immigrant or in a cross culture. I prefer to use a proportionate stratified sample design as in my research the variable of ‘being raised by immigrant parents/ cross culture’ is already defined. I would use that as the basis of division as according to Frankfort-Nachmias & Nacmias (2008), when dividing sample into a homogeneous strata the basis of division is related to the variable. The sub samples of the sample design would be based upon age of the participants as well. We are aiming to measure success at various stages in life: high school, college and then eventually mid life.

References

Frankfort-Nachmias, C., & Nachmias, D. (2008). Research Methods in the Social Sciences (7th ed.). United States of America: Worth Publishers.
 Groves, R. M., Fowler, F. J. Jr., Couper, M. P., Lepkowski, J. M., Singer, E., & Tourangeau, R.
(2009). Survey methodology (2nd ed.).  Hoboken, NJ: John Wiley & Sons.
Marshall, M. N. (1996). Sampling for qualitative research. Family practice, 13(6), 522-526

Rao, J. N., Hartley, H. O., & Cochran, W. G. (1962). On a simple procedure of unequal probability sampling without replacement. Journal of the Royal Statistical Society. Series B (Methodological), 482-491.

No comments:

Post a Comment