Children's Web magazine...
Entertaining , Educational, Fun,Informative and MORE

Surveys are useful tools of political research in which you have facts to then compare your results with. Good survey research is not just about designing the question in a neautral manner which does not lead the person to learn towards a certain prefixed answer; sampling, questionnaire design and fieldwork are all important components of surveys.

Surveys combine two things: the ancient but extremely efficient method of obtaining information from people by asking questions and the modern random sampling procedures that allow a relatively small number of people to represent a much larger population.

The purpose of surveys is fundamentally to help us to shed light on what ordinary people think, say and do. It also helps us try and understand why people think/ say/ do as they do. (e.g. Why do some people vote for the BNP?).

These are the elements of survey research:

1) We start with a clear idea of what is it we want to find out, describe or explain. E.g. why did people take part in the London riots?

2) We then develop a theory or explanation. E.g. political alienation, breakdown of family values).

3) next we write a question to try and measure our explanation.

4) We conduct a survey to collect evidence.

5) We use this evidence to establish whether our theory is correct or not.

If our explanation is ‘good’ then it tells us something meaningful about the real world. If our explanation is ‘bad’ there are three main possibilities:

- The first is to do with who we ask questions: we call this our sample, and we want it to be representative of the population

- The second is to do with what questions we ask: we call this questionnaire design, and we want it to be valid and reliable

- The third is to do with how we ask questions: this refers to the fieldwork

There are two main sources of survey error:

- Sampling errors: which can be a coverage error, sampling error, non-responsive error

- Measurement errors:  that is how valid the measurement and response is.

A sampling frame is a group of people we want to target and make inferences about whereas a random probability sample is where everyone has a non-zero chance of being selected, nobody gets excluded before starting whereas other types of samples include snowball samples and internet samples.

After the sample interviews are called for, the respondents are usually around sixty-five percent of samples.A good sample is representative of the population; it allows us to generalise our findings to the population  from which our sample was drawn. To ensure this a random probability sample is good.

The sampling frame needs to define population, a sampling frame is needed in order to select our sample and is a list of all the units in the population - for example to Electoral Register is one possible sampling frame. Once we have a sampling frame we can select our sample. Probability samples are the best methods for getting a representative sample. If a sample is drawn at random everybody in the pop has a known, non zero probability of being sampled. And sample size needs to be large enough, typically about over 1000people, then we can make reasonably reliable generalisations. Most social science surveys achieve a response rate of about 65%. there are multiple ways to maximise response rate: advance letters, face to face interviews, repeat visits on different days at different times or even payment.

Other methods of sampling are quota, purposive, and snowballing.

Measurement errors

There can be multiple errors involved in measuring in your surveys and can include:

- Question wording: ambiguous, lengthy, double-barrelled questions (asking about more than one thing), word choice, loaded questions.

- Social desirability, (agreeing with what is to be considered socially correct), knowledge questions, threatening questions, memory and respondent recall.

- Open or closed questions: by giving them list of alternatives you focus their attention on what you want

- Question order effects

Question wording affects the answer and the question order is linked to reciprocity. Surveys are very sensitive measurement instruments and one should be aware that the interviewer could make the mistake of manipulating and leading the participant. There are many sources of error or bias; these call for careful design and implementation which can reduce but not eliminate many of these problems. 

Image: http://www.dykes-hall.co.uk/wp-content/uploads/2015/02/Survey.jpg