Will consider three aspects of reliability:

1.Sources of unreliability.

• may be due to bad wording
• a person may understand the question differently on different occasion.
• different coders might code the same responses (e.g. occupation) differently.

2. Testing reliability.

3. Increasing reliability

• the best way to create reliable indicators is to use multiple item indicators: they are more reliable.
• asking their sex in different way.

Three basic ways to access validity:

1. Criterion validity.

• compare new question (our own) with the established measure - but the old one is valid/hard to find?

2. Content validity

3. Construct validity.

• evaluate a measures by how well the measure conforms with theoretical expectations.

1. Obtain a complete sampling frame.
2. Give each cause a unique number starting at one.
3. Decide on the required sample size.
4. Select that many numbers from a table of random numbers.
5. Select the cases which correspond to the randomly chosen numbers.

Divide population by required sample size e.g. 50/10 = 1/5. We will select one person for every 5 in the population. Where to start? Since fraction = 1/5 - starting point is somewhere between persons 1 - 5 , so decide from table of random numbers.

• Select the relevant stratifying variable(s).
• Use SRS or systematic sampling to select of people within each strata.

• by judged as typical of some category of cases of interest to the researcher.

• the quotas are given; e.g. 20% of population between 20 - 30 years old etc.
• non - random - select by a certain criteria.

• anyone be useful for pilot testing questionnaire or exploratory research.
• must be used with caution and only for specific purposes.

The required sample size depends on two key factors:

• the degree of accuracy
• the extent to which there is variation in the population.