Introduction
Sampling involves selecting a section of a total population (the group being investigated) in order to evaluate and come up with a general view of the group’s traits. Sampling, in essence, reduces the cost of research; data collection becomes quick, and there is correlation and near accuracy since the group involved in the research is small. There are five different types of sampling, that is: stratified sampling; quota sampling; cluster sampling; systematic sampling; and simple random sampling (Virgin.net).
A stratified sample is a sample that has been grouped in a stratum, in plural strata. A stratum happens when the members of a population vary and are grouped into uniform groups for the sampling of each individual differently. Clustered sampling refers to when samples are divided into groups called clusters and the groups are sampled other than the individual elements of the survey. Intraclass correlation coefficient is the quantity of the consistency of estimations or rankings.
Confidence Interval: Survey Accuracy
Why the accuracy of a confidence interval calculated from survey data increases as the survey sample size increases
Since in sampling, the researchers just use a small cluster or representative of the total population, it is likely that different clusters of the population may result in slightly varied results. Due to the variations the surveyors normally gives a Confidence Interval (CI), which is a range of values or a limit; both upper and lower limit within which the findings should lie (Institute for Health and Work 2007). In order to tell how accurate or precise an estimate is we use the CI.
According to Vaus (2002 p.81) in the article “Surveys in Social Research”, the author talks about the number of things that need to be noted about the connection of correctness and the amount of the sample. Through working with small samples, the improvement in accuracy only increases considerably. Increasing the size of the sample, for example, from an amount of 100 to 156 decreases the ultimate sampling mistake from a percentage of 10% to 8%. To decrease the sampling error from 2.5% to 2%, that is, the confidence interval, the amount of the sample needs to be increased by nine hundred cases. The author goes on to add that the law governing this states that by decreasing the sampling error by half the sample size needs to be multiplied by 4. This is basically because there is an added response rate if there is a larger sample size as the percentage of people who respond to the survey is increased (Bennekom, 2007).
Stratified Sampling
Recommendations as to when stratified sampling can be useful when conducting a survey
In stratified sampling, the population is first divided into groups, and an equal sample is selected from each group. As a result, this type of sampling produces more accurate findings and also brings out the different characteristics within the various categories. According to McCormack & Hill (1997 p.53) in the article “Conducting a Survey: The Spss Workbook”, the author talks about how stratified sampling tries to guarantee that there is no chance that the main parts of the sample might be signified below standards in the sample because of random errors.
Stratified sampling tends to make sure that people of each age group are very well represented in the sample. Stratification can improve the efficiency of the survey as sampling each individual provides approximations with lower inconsistency. It allows a better comparison of the samples. It is mostly suitable to cases where the subjects of the sample are naturally with identifiable characteristics within their groups.
Cluster Sampling
Recommendations as to when cluster sampling can be useful when conducting a survey
Cluster sampling is a method that selects a whole sample from a population of survey instead of specific persons after the whole population has been grouped into different clusters. The main advantage of this type of survey is that the analyst does not have to get into the sampling model which has a record of each person of the population under survey. The most suitable application of this type of sampling is where the samples stand for different parts of topographies (McCormack & Hill, 1997 p.53). A cluster sampling survey offers a cost advantage as well.
Cluster Sampling: Online Piracy in the United Kingdom
If you were conduct a survey about piracy for a large population, spread over the whole of the United Kingdom, what clusters (i.e. geographical areas) would you recommend if cluster sampling were to be used. In your answer, give justifications for your recommendations
I would conduct a survey in Scotland, Ireland and England geographic regions. This is because these regions are historically known for holding a large population of a group that practices online piracy of videos and music. Scotland has a 66% rate of piracy. Ireland has a 48% rate while England has a 72% rate of piracy. The ratings are as per each country’s developments in technology.
In determining the sample size we use tables and power function charts. Each survey should be guided by a set confidence interval which is as well based on the sample size. Different sampling methods require different sample size with regard to the percentage and the required width.
List of References
Bennekom, FV 2007,Statistical Confidence in a Survey: How Many is enough? Web.
De Vaus, DA 2002, Surveys in Social Research, 5th Edition, Routledge, St Leonards, NSW, Australia.
Institute for Health and Work, 2007, Research Excellence Advancing Employee Health: Confidence Intervals. Web.
McCormack, B & Hill, E 1997, Conducting a Survey: The SPSS Workbook, Cengage Learning EMEA, Cambridge, United Kingdom.
Virgin.net, Different Types of sampling. Web.