Politics being a very sensitive national issue, there is a need to exhibit high levels of accuracy in such opinions as the opinion poll. As a result, the polling companies should ensure that their results have a very low standard deviation and consequently as low a margin error as possible. Although the most commonly used methods like quota sampling yield results, the techniques are not purely random, and hence it becomes quite difficult when trying to estimate the distribution of the statistics. This bars this technique from achieving its goal of giving its output as purely predictable in any distribution given time difference.
Population into classes like those who work in the public sector, those in the private sector, college students, university students, lecturers, teachers, farmers, lawyers, engineers et cetera. The number of samples in each group will be equal. I shall dispatch a team of trained interviewers to the field for the survey. A comprehensive sampling can then be carried out by an interviewing team that will get the opinions of the sample population in each group. Say, for example, an opinion poll shall be carried out on university students. Each group shall be sampled put individually until all the groups have been covered. The outcomes of each group shall then be printed accordingly, and thereafter the average of all the groups is calculated. This shall then be published as the national opinion poll on the presidential race.
As seen before, this method is more accurate in the poll results because the number of the sample of the population is large enough to minimize the error in standard deviation. When the standard error is minimized, the data tends to have more precision or less imprecision, meaning that the accuracy of the estimate is high. In order to reach the greatest heights of precision, techniques like strata-based sampling and group sampling may be used singly or in combination. Stratified sampling occurs when the need arises to divide the population into groups or subpopulations, and therefore the interests of the groups are subjected to variance with respect to each stratum. Here the interests of the whole population are represented and therefore giving higher reliability in the opinion polling. Group sampling or Cluster sampling is a sampling technique where the entire population can be divided into groups or clusters, and a random sample of these clusters is selected. All observations in the selected groups are integrated into the sample.