1.)
The primary sources in survey research come from three major components. They are, sampling error, measurement error, and specification error. When sampling error becomes a problem it is usually because the population is irrelevant to the question at hand. Either the chance for a purely random sample has been skewed because not everyone in the population had an equal chance to be chosen for the sample or response rates have made it hard to have a truly random sample. Defining the universe is also a major component of sampling error. When conducting a phone survey you always run the risk that the same person will answer the phone in every home, usually the mother. To avoid surveying the same group of people from each house hold (i.e. middle age women), you must implement a line of questioning that will lead you to another member of the household. Another problem is that many people refuse to take the survey once they are contacted; this is where call back come into play. It is crucial that the interview sets up a callback to the residence that refused the survey. Over the past several decades refusal rates have been on the rise, and it has been found that response rates to surveys directly correlates with their participation in politics. The more involved the citizen the more likely they will be willing to respond to your survey questions. Another major risk in sampling error is the disclosure of sampling information. News stations have a much lower chance of producing a truly scientific poll because of their need for something fast and fresh before their competition gets ahold of similar information. This has led to less than scientific polls on a daily basis, although in recent years there has been a crackdown by some networks and newspapers to conduct polls more accurately.
Measurement error is even more common than sampling error, the way the survey is written can have a huge impact on its outcome. The wording of the question and the wording of the response choices can have a major effect on the outcome of the poll. Studies show that questions that offer a balanced answer instead of simply a yes or no, or agree/disagree format received more accurate results. There is much discussion on whether a middle of the road answer should be offered as an option, as well as a “don’t know”. It is said that people can use these answers to hide their true views on the topic regardless of the motive behind hiding their views. Many questions asked in surveys can leave the respondents feeling uncomfortable, or ill-informed leaving them embarrassed and unlikely to answer the question truthfully. Attitudes and nonattitudes can have an effect on measurement error as well. According to a study done by Philip Converse many people do not have a true opinion on many major issues and just pick answers at random when asked. It was also found that even within one survey there is a chance that there can be very strong attitudes that mix with a few nonattitudes. Sometimes the knowledge gap comes between the interviewer who is especially knowledgeable about the topic and the interviewer who does not have the same amount of knowledge. As human beings we are organic creatures and that applies to our minds as well. We have the ability to change our views on a topic simply based off something we heard in passing or knowledge gathered over many months of research. Although there is a greater chance that it will take us a longer amount of time to decide firmly on a subject but we tend to stick with that view. Which is why polls taken months before an election can prove to have the exact opposite results a week before Election Day. The context in which questions are asked can make or break a survey. Whether it be the placement of a question about the approval of an elected official or the redundancy of questions people are likely to change their answers to conform to what was asked of them earlier in the interview. Attitudes about issues are complex and the interviewer must realize this before conducting the survey. Careful consideration must be taken to be sure that the wording is not overlapping, confusing, contradictory, or vague. A split sample technique can be used to assure this.
Specification error comes about when the theory behind the survey in the first place is flawed. A major problem is that many people are under the assumption that everyone has clear opinions on many issues when its actually almost opposite. Many people have clear opinions on a few issues that are important to them and use those issues to form opinions on other issues when asked about them. Many problems arise in surveys because of faulty assumptions about the population. Surveyors should try to target major issues for the sample they are studying, in order to get a more accurate picture. The way people vote is a complex process in which when people are confronted with an issue their minds may think of it in three different ways. An almost black and white way in which this issue is on this side and so am I or valence, or a concern for performance of a certain issue. Many politicians believe when they position their issues as valence issues they are more likely to be accepted by the majority of the public. Many people can feel uncomfortable showing their true colors when it comes to issues of race, which can lead to skewed polls. The Bradley effect demonstrates this well. Theory specification is essential when putting together a poll, you need to know your audience, and understand the way their mind is going to work.
In the Feldman and Zaller article response instability is listed as a cause for error in survey research. Response instability can be caused by many things, one of them being that many people do not have cut and dry opinions on many of the issues that are asked of them , but continue to answer anyway at random. People may also change their responses because of the questions answered prior.

2.) In the article by Iyengar, Peters, and Kinder they describe the minimal effects model was constructed and showed that propaganda rarely changes the public’s view of the issues at hand. Instead of persuading the public to think a certain way the wanted to get the issues to the public and then the public would think about them, this is called agenda setting. Essentially the media choses which stories to run and therefore choosing which stories people are going to be taking about. Priming is when the media chooses to run one story over another regardless of actual importance, therefore putting the story in the forefront of the public’s mind. The experiment was designed so that two different groups of people would be shown the nightly news only one of the broadcasts would be edited to no detection to show stories about one topic. Before the week of nightly broadcasts they were given a survey of views, and then given the same one at the end of the week of broadcasts. The psychological foundations of their findings show that salience has much to do with the process, but does not always determine the outcome. Salience rarely shifts moral beliefs, although repetition of information does automatically bring it to the forefront of the mind and does affect the outcome. In Zallers Chapter 4, Table 4.1 shows the RAS model. The Receive-Accept- Sample Model is based on the following: the more the person knows about a particular issue the more likely the person will be willing to accept information about it, also people tend to resist what they do not know or do not wish to know. The more recently an issue has been brought to mind the easier it will be for the respondent to answer a question about it. It is much easier for a respondent to answer a question accurately about information they have recently heard or talked about.
3.)
Partisan Activation is the idea that no matter what the respondents views are throughout the election process, on Election Day they will most like retreat back to their original party. Party loyal is extremely important to the election process. This differs from persuasion because according to Bartels definition of persuasion, it has to do with the issues and not the person doing the persuading. The response to the issues and the view of the candidate persuades the voter to change his or hers mind. Polls can be very unreliable during presidential elections and their campaigns. The causes for this can be as mentioned earlier in question one a series of measurement errors when it comes to surveys, including but not limited to question wording, non-response rates, and the ability of the voter or respondent to change their mind over the course of the year. In January of an election year many people consider themselves to be undecided as the year moves forward that number of undecided people drops. The enlightened preference hypothesis aims to prove that when originally polled respondents knew very little of the campaigns or the candidates and as time wears on and they learn about the campaigns or candidates from TV commercials, news stories, or social groups they become enlightened to an extent on the topic more than they had before and their views may change from the original choice.