The Poll Methodology by Fairfax/IPSOS - Assignment Example

Research Evaluation: Fairfax/IPSOS poll, published Monday 15th February 2016 Name Class Unit Introduction This research evaluation is based on Fairfax/ IPSOS poll which was published on 15th February 2016. This will involve determining how the researchers collected the data, sampling method used, where the data was collected, time of data collection and why they selected the type of data. The analysis will also evaluate the methodological approach, what is reported and lastly discuss strengths and limitations of the methodology. Data collection The researchers in this poll used telephone surveys. This is through random digit dialling (RDD). In this method, the researcher uses the area code on software which generates the last numbers in a random manner. The research centre selects the random digit sample based on both cell phone and landline numbers. The principle behind the use of RDD is use of exchange information numbers (EIN). The EIN are based on a given area while the rest of numbers are randomised. The sample was then surveyed and data recorded for analysis. The data in this case is respondent answers. Sampling The random digit sample used in this case comprised of both landline and cell phone users. 32% of the sample was made up of mobile phone users. The sampling of both landline and cell phone ensures that the surveys are able to represent all those who have access to either landline or cell phone. The ratio used on the landline and cell phone helps in ensuring that the sample has a balanced demographic composition and also balance the costs of fieldwork consideration. The landline sample using this method ensures that both listed and unlisted numbers are represented. This sample of 1403 respondents was selected who were aged over 18 years. Data collected The data collected through the phone survey are the respondents answers based on the party approval, party support, candidate approval, candidate preference. This is through asking questions over the telephone on the same. This led to having data on the number of people who supports each party (ALP and LNP), number of people supporting a given parties in primary (LNP, ALP, Greens, PUP, and others) candidate approval ratings (Turnbull and Shorten) and lastly the candidates preference (Turnbull and Shorten). The data was based on the respondent opinion which was tallied as a percentage of the total. For example, the percentage of the respondents supporting Prime Minister Turnbull is 62%. Time of data collection poling preference in Time of data collection Time used for polling matters a lot. This is due to fact that the time determines the possibility of polling a given group. For example, a poll done on Friday evening may leads to young voter missing out hence introducing bias. This leads to the need for carrying poll in the entire week or normalising the results to cater for everyone. The telephone interviews were carried out from 11 to 13 February 2016. We can make an assumption that the poll was normalised to cater for everyone. This is the fieldwork date where a sample of 1403 was interviewed. The research by Fairfax/IPSOS was aimed at gaining public opinions on Australia politics in 2016. The poll required to determine the support for the two party LNP and ALP based on the preference flow from the last election. This data helped in determining the strength of coalition under Turnbull. The poll also wanted to determine the primary support for the coalition since the last election based on poling preference. To determine the approval rating and preference for Turnbull and Shorten, they had to interview on the same based on respondent support. All the data was obtained based on interview questions where respondent answers were recorded. Methodological analysis RDD is used in the research due to its capability to reach a wide population. RDD and has the capability of including all eligible persons in a given population. The method utilises both landline and cell phones in the target population. The rise in the use of cell phones makes it vital to ensure that the research methodology is able to balance the sample between the landlines and cell phones. This has been achieved through ensuring that 32% of the sample was made up of cell phone numbers. Despite this, use of cell phone surveys is costly than the traditional landline. This may also have led to the lower percentage of cell phone in the sample. The rate of response from the cell phones is lower than when using the landlines. The polling methodology used by Fairfax IPSOS has the capability of estimating the population response based on close ended options given during the interviews. Also, the sampling method used in this poll is capable of representing the whole population in Australia. This is due to fact that the sample is generated in a random manner. The sample has been able to meet the minimum number which is required for accuracy. It is important to note that beyond a given minimum number (1000), the number of participants used in a sample is less important to poll accuracy compared to the process of random selection. The margin of sampling error in this research is ±2.6% is within the limits. For a number of 1000 people in a sample, the accepted limit is within 3% either way. This shows that the sample selection has ensured equal and random selection. The main aim of measuring reliability is to ensure that the error is reduced as much as possible. This is a margin of error that has been determined based on the assumption that the sample was properly collected and the questions well designed. In this case, it can be assumed that the use of RDD helped in selecting a sample that is a cross sectional representation of the Australian population. Despite this, failure to have a well selected sample may lead to an aura of precision which is lacking. If this sample is not well collected and the questions are not properly worded, there is a chance that the results are highly unreliable. On the poll profile, the report fails to show the type of questions which were asked. This would have helped in determining whether the questions are well designed and also explain the underlying concepts which might be of interest. The wording determines the response. This may also help in determining whether the margin of error given is accurate. Lack of questions used makes it hard to analyse whether the poll had measurement errors. Despite this, polls contain coding errors which are inevitable. It is also difficult to analyse the questionnaire design. This is due to fact that there is no actual wording of the questions asked. The wording of the questions has a great impact on the final results. Wording determines how the participants contribute especially in open ended questions. The methodology used by the poll has been able to adhere to the logic of generalising the survey finding from an agreed sample to the large population. This is through adhering to the strict statistical principles in random probability sampling. The RDD methodology is also used due to fact that it is fast hence able to keep pace with changing political environment. This is due to fast turnaround time in this method. The method provides a tool which is capable of understanding the trends and characteristics of the general population. Use of face to face is highly costly and is also subject to bias which is not possible with use of telephone. Strengths and limitations of methodology Strengths Fairfax/IPSOS has used RDD in sample selection which ensures that the sample is randomly selected and everyone has an equal chance of being selected. RDD is easy to administer compared to other methods. The sample can be selected automatically in less time as compared to other sample selection methods. Use of telephone through RDD by Fairfax/IPSOS makes it possible to conduct the interview remotely. This reduces the geographical dependence hence reducing the overall costs. The method is thus cheaper than conducting the interviews in given geographical regions. When the sample used in RDD is valid it gives a true representation of the large population. The method is highly efficient when collecting information from a large sample. It also reduces the number of errors due to standardisation. It is possible to collect a wide range of information on topics such as values, beliefs, attitudes and behaviour as seen in the case of Fairfax/IPSOS. Limitations The poll methodology by Fairfax/IPSOS has several limitations. The data reliability in this poll is affected by various factors. This includes the level of motivation for the respondents, ability to respond and their memory. The error margin of ±2.6% used in this poll is based on assumptions on the sample collection and the question design. When the two assumptions are violated, there are chances of getting the wrong margin of error. This may lead to results whose precision being predicted is wrong. When collecting the sample used in this poll, there is a possibility of an error in over representing a given type of respondents. There are chances that the poor, homeless and those without telephone connection are left out when using RDD. Also, the population which speaks foreign language are not able to participate. This includes the migrants or ethnic minorities. For example in Australia, there are Aborigines and migrants who may not be fluent in the national language. The method is exposed to the coding errors. There are chances that some questions can be miscoded or there is a reporting error. Another limitation is social desirability bias. This occurs when those being interviewed tries to create a favourable social image. The respondent may try to give a desirable answer through lying. Despite this, the occurrence of this error is low. Also, there is a possibility of the interviewer in RDD to come up with inappropriate cues or change the questions wording due to bias. Read More