Mathematics, Research Paper Example

George W. Bush had been leading in the poling surveys that were major according to Kudlow (2000). It has also been reported that the Texas governor leads with a twelve percentage which indirectly means that this might cause a change in the House GOP and a Senate.  After checking with different polls like the new CNN/USA Gallup poll, CBS News poll, LA Times, Newsweek and AP poll , they were all giving the same results  which showed that Bush had outdone Gore. However, on Election Day Bush ended up losing the popular vote by a small margin (although winning the electoral vote).

So what do the economists say about these variances? What could be the explanation behind these unexpected results? Is this where we need the statistical probability tables which Dobson,Young & Gibberd (n.d.) keeps talking about?  This is where Republican Ed Goeas comes in. Goeas is a publisher of articles that are highly regarded like the battleground survey. He cautions about poll reading and says that the time the polls are conducted really matters when it comes to determining results. It matters if they are conducted over the weekend or during the week. During the week, from Sunday night to Thursday night to be specific, is when the political pollsters are seen to poll. Unlike the common belief and practice, the weekends are more based on the location of different races.  It is considered abnormal for a person to spend between twenty to thirty minutes during the weekend trying to even talk to a pollster.

Another theory by Goeas is that the surveys covering registered voters are much less compared to those that are meant to canvas likely voters. The previous statement is supported by another line that talks about an anti-Clinton feeling. The democrats are said to have a very low intensity compared to republicans.

The director in charge of media of Mr. Bush also agreed with Goeas.  He mentioned that some of the things that can determine the accuracy of a poll are voter samples and weighing (even for women) which includes the geographic balance. It is therefore essential to have the right percentages from every region and state.

What sampling techniques would you use to get the most accurate estimate of which candidate would win on Election Day? The most effective sampling techniques that would be used for the purpose of an election would therefore be the probability sampling methods namely: – simple random sampling, stratified random sampling, systematic random sampling, area random sampling and multi-stage sampling. Probability sampling as explained by Web Center for Social Research methods as a technique that makes use of random selection. To have this method of selection that is random, it is important to set up procedures that make sure that the different aspects in the population are captured and all have equal probabilities of being selected.

The simplest of them all is the simple random sampling which according the Web Center,  uses a way to randomly generate numbers mechanically which is computerized and also have a table which is used for coming up with random numbers. Its objective to have an equal chance in selecting the random numbers is also another of its best characteristics. Selecting a simple random sample can therefore be so easy especially if the business firm can either produce a list of the company’s records or if it has computerized records then a sample is drawn. This method would represent the views of the wider population hence quite accurate.

Stratified sampling, also known as quota or proportional sample is also accurate considering that the population is subdivided into same groups and a sample is taken from each group. This is a very good way of sampling because all the subgroups are represented in the main sample. This method might actually be more accurate than simple random sampling because its representation goes beyond just the population. It represents even the subgroups. In the polls, this method would be most effective since it focuses on a smaller part of the population and would therefore give a true and more honest view.

Then there is the systematic random sampling which is done by looking at the size of the population and deciding the sample size needed. The interval size is then got by dividing the population size by the sample size after which an integer is selected from one to the interval size number and the unit that falls on the interval size number is selected. This method works best with a randomly ordered population and compared to the above methods, it is way easier to do unlike systematic random sampling and more accurate than simple random sampling. It would therefore pass to be the best between the three in determining the polls.

Area or cluster random sampling method is best used when regions that are widely disbursed such that a lot of travelling has been done to finally get to the targeted population. To cover the intended area, the population should therefore divided into clusters, then sampling of the clusters randomly and finally cover all the units in the sampled clusters. This method applies so much to a poll scenario because the statistics might need to represent every part of the country. It can therefore be very helpful to the process.

Then there is multi-stage which covers all the four methods to achieve a stronger variety because separately the methods are quite effective but even more effective when combined. The multi-stage method would therefore the most accurate estimate which would give the final true results of the candidate.

References

Dobson A., & Young A. & Gibberd B. (n.d.). SurfStat Australia. Retrieved November 29, 2009, from http://surfstat.anu.edu.au/surfstat-home/surfstat-main.html

Larry Kudlow (2000) The Weak-End of Polling, Retrived November 29, 2009, from http://www.nationalreview.com/kudlow/kudlow071800.html

Probability Sampling, Retrieved November 29, 2009 from http://www.socialresearchmethods.net/kb/sampprob.htm