Frame work and size of sample.

Kinds of the social scale.

There are various scales commonly used in social researchers.

1. Point scale.
2. Borgodous scale.

Point scale.

This is a type of scale developed to classify persons by two extreme criterions. For classification, Words or situations representing the criterions are selected. The respondents are asked to cross the one that is unfavourable to him/her. A point is given to every word that has not been crossed out. The attributes of the respondents are determined by the number of points and types of items not crossed out.

Borgodous scale.

This is the type of scale which is developed by Emery.S Borgodous which measure social distance. Social distance is being defined as nearness or liking between two social groups. In this case, a feeling of Americans towards different races such as English, Swedish, Poles, Koreans, are considered social distances. He prepared several items which show the varying relationships of a social distance of Americans with other races.

Frame List.

Frame list is the list of the population units which the sample units are selected. The list is sometimes called as the frame. If the list is not available, it should be before conducting the main survey.

In order to be useful, the list must have following qualities.

1. The list must be reliable.
2. The list must be maintained by an authority that can be relied upon.
3. The list should contain full information about the units so that an appropriate sample design may be planned.
4. The list should be up to date.i,e it must include latest additions to the population (Birth, In-migrants, etc and exclude Deaths, Outmigrants).
5. The list should be exhaustive.i,e all population units must be included in the list.
6. In the list, the unit must not be repeated in the list. For example, if the list of guardians is maintained, the name of the persons may appear at more than one place according to the number of school going children.

The problem of a list.

In the problem of a list, the perfect list is rare. Most of the available list suffers from following problems.

1. Missing of elements , non-coverage, incompleteness,etc.
2. A cluster of elements together in one listing (Names of a member in a family).
3. Blanks or foreign elements.
4. Duplicate listing.

Ways of avoiding the frame problems.

There are three ways of avoiding the problems.

They are as follows.

1. Ignore and discard the problem.
2. Redefine the population to fit the frame.
3. Correct the entire population list

The size of the sample.

The size of the sample is an important factor to be considered in sampling. This is because the size has a direct bearing on the accuracy of Estimation. Cost and administration of the survey. Though large samples give a smaller standard error, they are general, difficult to manage and unfit for the study. On the other hand, small sample tends to give the higher standard error but avoid unnecessary expenses. Therefore, an optimum sample size is required. An optimum sample size is one which fulfils the requirements of efficiency, representativeness, reliability, and flexibility.

The criteria are: The sample should be small enough to avoid unnecessary cost and large enough to avoid intolerable sample errors.

The technique of testing the reliability of the sample chosen.

While testing the reliability of the sample, the size of the sample is an important factor to be considered. This is because the size has a direct bearing on the accuracy of estimation, cost and administration of the survey. Through large samples give smaller standard errors, they are generally difficult to manage and unfit for detail study, on the other hand, small samples tend to give higher standard errors but avoid unnecessary expenses. Therefore, an optimum sample size is required. An optimum sample size is one which fulfil the requirements of efficiency. Representativeness reliability and flexibility.

If the sample design is simple random sampling, the simple procedure is first to estimate.

$$n_0=\frac{t^2_a/_2σ^2}{d^2}$$

Then to estimate ‘n’ by using the relation.

$$n=\frac{n_0}{1+n_0/N}$$

Where (σ)=population standard deviation.

ta/2=critical value for two tail test at α level of significance.

D=described error in estimation of the population means =\(\overline{X}\)-μ

Since n0>n1 higher so sample size I,e no is better than smaller one I,e n1

For attributes with two possible outcomes, it can be shown that σ²=PQ.

Then the relation takes the form

$$n_0=\frac{t^2_a/_2PQ}{d}$$

Where P= the proportion of an outcome of a dichotomous attribute in the population.

Q=1-P.

Reference

Kerlinger, F.N. Foundation of Behavioural Research. New Delhi: Surjeet Publication, 2000.

Kothari, C.R. Research Methodology. India: Vishwa Prakashan, 1990.

Singh, M.L. and J.M Singh. Understanding Research Methodology. 1998.

Singh, Mrigendra Lal. Understanding Research Methodology. Nepal: National Book centre, 2013.