Surveys and accidental sampling

Opinion polls carried out by accidental sampling of the survey is not random.

Accidental sampling are characterized method is selected without the person being surveyed, cases are chosen to have at hand.

Examples of accidental sample surveys are what make the radio and television journalists.

These surveys are not representative and can never know how reliable the sample is selected.

As a research about public opinion is unreliable.

Provided that presents the opinion of the way we run a double risk of having a biased sample.

First, as the sample is casual, may not be at all representative, and even that interested people try to be surveyed.

In that case, the bias of the sample increases.

But secondly, if the survey is planned or executed from a media or institution with a clear profile or defined interests, who agree to be interviewed by such means shall be persons of similar ideological.

An example of this is the surveys of online newspapers, where readers vote who they are, a universe is clearly not representative.

A definition of survey

We discuss the definition of "survey" that provides the academic Manuel Garcia Ferrando (1992:141), who says:

"The survey is a technique that utilza a set of standardized research procedures by which collects and analyzes a series of data from a sample of cases representative of the wider population or universe, of which aims to explore, describe, predict and or explain a number of characteristics. "

Here García Ferrando is very ambiguous to use words like "method" and "procedures."

The author defines survey as a method or quantitative research, which in turn can use different "standard procedures", which actually means data recording techniques subordinates to research.

It is highly questionable whether the survey to predict, because the investigation is complex but never based on final data from more than one time in a reality of dynamic phenomena and causes by many variables.

There is no epistemological justification for claiming that a survey to predict, since in any case, should be a highly confirmed theory that allows predictions or forecasts in science.
Unless, of course, to pursue practical or political objectives.

The types of survey: individual survey

There are, strictly speaking, various types of opinion polls and attitudes registration.
But if we analyze the different types of surveys in processes for the collection of data we can distinguish three major types of survey: staff survey, the self-administered survey and telephone survey.
From now on, we must clarify, we are talking about surveys that apply to people or survey investigating the opinions and attitudes of individuals.
Discussed in this post the individual survey.

INDIVIDUAL SURVEY

The personal interview is the best known of the three types, and today almost no one living in urban centers has not been sought to answer some.
This is a typical face-to-face survey.
In this type of survey requires a team of trained interviewers who, after receiving an instruction concerning the nature of the survey, how to administer the questionnaire, some problems with certain questions, how to access the survey, should contact the persons elected to be interviewed.
Sometimes when it is a multistage sample survey, the last stage, which involves selecting people who have to answer, is undertaken by the interviewer.
At other times, however, already has the address and name of the person, as when performed on a standard sample, eg a professional society.
It is the task of the interviewer approached the respondent with a structured or semi-structured questionnaire, or with any questions closed or open and closed questions (discuss this later).
One of the methodological issues is often not sufficiently appreciated is that all questions must be made strictly in the order they appear in the questionnaire, in the same tone and, above all, without adding or omitting anything, even a comma in the text survey questionnaire.
This ensures the objectivity of the results from the application of the same stimulus to all respondents. Let's be clear, as always, that these considerations are made on the assumption that the questionnaire is objective and is built by professionals, that the questions measure what they purport to measure, and so on.
Some types of personal opinion polls are coincidental survey, the survey at random or random (or which includes, as recorded at a final stage of sampling a random selection process) and survey fees.
The coincidental survey is a type of non-probability sampling (later we will analyze it), or a sample in which no one knows the error of estimate (we'll explain later) or error resulting from the spread of the sample data population. In the survey, the interviewer takes coincidental random people in a busy city and survey.
It's that simple, as they make bad journalists and pollsters (and political operatives, why deny it), although some research may be used responsibly. In the survey at random or random pollster has a table of random numbers that tells you based on the number of questionnaires to be completed, for example, which household member should interview and may in some cases replace him in case of absence for an individual of the same sex and similar socioeconomic level and age.
Finally, the quota sampling is one in which the interviewer and out of the office of the institute or consultancy or agency with a guideline on how many men interviewed, a few women, what ages, what you already know what you want and crossing is going as the respondent.
As can be seen, the random survey individual may lead to the replacement of absent or rejecting entevistado a fee.

Opinion polls as a way of sampling

Today mention an error that is usually found frequently in the literature on surveys and opinion polls.
The same is to confuse all the surveys for opinion polls or social surveys. If we wish to clarify what is a survey, or provide a definition, we must make it clear that by no means something that applies to people necessarily.
In fact, the survey is a technique of data collection (some might question how far the data are "collected" as one who takes something that's already been out there and are not created, at least to some extent by those who do research, but that is another problem that will address later) that derives or is an application of the sampling procedure.
Specifically is the use of certain standardized and validated procedures under which the research takes information on a representative sample of a population or universe, as is usually called the total haw investigated.
It is clear that the purpose of such research procedures is to improve our insight of some phenomenon, say a good explanation.
One can conduct a survey in which sample is selected books from a library upon which a questionnaire is completed in some information, because what matters is that the procedure and representative sample.
This, as already mentioned, results from the application of the mathematical theory of probabilities to the investigation of phenomena, whether or not social.
A sample survey differs from a census in which it is to administer a questionnaire to all people concerned, while the survey is always applied to a sample, the best representative.