Systematic sampling from finite populations.

摘要: The impossibility to reach an entire population, owing time and budget constraints, results in the need for sampling estimate population parameters. There are various methods of this thesis deals with a specific method probability sampling, known as systematic sampling. Problems within context include: (i) If size is not multiple sample, then conventional (also linear sampling) will either result variable sample sizes, or constant sizes that greater than required; (ii) Linear most preferred design populations exhibit trend; (iii) An unbiased variance cannot be obtained from single sample. I attempt make original contribution current body knowledge, by introducing three new modified designs address problems mentioned above. We first discuss measures compare designs, before providing review Thereafter, methodology examined well two other methodologies overcome problem (i). obtain efficiency related formulas methodologies, after which we demonstrate depends on correlation units, turn arrangement structure population. As result, common under structures. Further (including proposed design), considered optimal trend, resolve (ii). tackle exploring strategies, include designs. Finally, numerical comparisons all discussed thesis, structures, comprehensive report thesis.