5sampling Techniques

Sampling techniques for Market Research

Basic Terms Universe / Population The complete set of any group e.g. total population of hyderabad. Sample / Panel A sub set which represents the Universe. The selected respondents are called sample and the selection process is called sampling technique.

Universe

Sample

Sample Sample is not a concept solely used by statisticians, We all implement sample for various purposes in our day to day life, Few examples of samples are : Every housewife determines whether the soup she is preparing has an acceptable flavor just by tasting a spoonful(Sample). You may test a swimming pools temperature just by dipping your toe into the water(Sample is the water around your toe). Before buying a new book many people scan a few pages(Sample) to see whether it is interesting. You do not have to drink a whole glass of spoiled milk to ascertain that it is sour. A prospective car buyer test drives an automobile to judge the performance of the vehicle.

Sample size and it’s significance What should be the size of the Sample (How large or small it should be) ? If the sample size(‘n’) is too small, it may not serve to achieve the objectives, and if it is too large, it may incur huge cost and waste of resources . Generally, sample size must be of an optimum size, it should neither be excessively large nor too small . Statistically, Sample size should be large enough to give a desired confidence interval and acceptable error level.

Reasons for Sampling •

A universe survey is time consuming, in most of the cases the decision makers have a time frame.

•

The cost of gathering information from the universe is very towering, this is a compelling consideration In favor of sampling.

•

The accuracy of the information may not be justifiably enhanced by taking a complete enumeration.

•

In some situations the item may not be reusable after the test (e.g..testing the photographic film by exposing it). Testing the sample is the only solution in this kind of situations.

Managerial Objectives of Sampling • How representative is the sample to the population. For e.g. if the decision maker is interested in only a segment of the population, Sample also should represent the same.

•

The sample should be sufficient enough to give stable results.

• Using research resources efficiently within the limits of the time allocated for the project.

Sampling Process Define population from which the Sample is to be drawn. Establish a ‘frame(source List / census)’ of the population. Choose the method of Sampling.

Determine the sample size. Identify/select the actual members of the sample.

Sampling Methods Sampling methods Sampling methods

Non probability methods 1. Convenience Sampling 2. Snowball sampling 3. Quota control Sampling 4. Judgment Sampling

Probability methods 1. Simple Random Sampling 2. Systematic Sampling 3. Area Sampling(cluster) 4. Stratified Sampling

Sampling Methods In probability sampling the population elements have a known chance of being selected for inclusion in the sample. Where as in non probability method, members of the sample are selected purposefully or accidentally. Probability methods are more efficient and scientific in obtaining accurate samples, but in all situations probability methods are not practical.

Non probability Methods 1. Convenience Sampling According to this method, the sample is selected on the basis of convenience or accessibility. for example, for testing a new product one can simply add this to the appropriate product section of a supermarket and observe how well it moves in comparison with other products of the category. The problem with convenience sampling is that we have no exact way of determining the representativeness of the chosen sample.

Non probability Methods 2. Snowball Sampling According to this method, initial units are selected using probability methods, but the additional units are then obtained based on the information of initial units(referrals). Reduced sample size and costs are the major advantages of this method. But the referral selection process can cause some bias .

Non probability Methods 2. Judgment Sampling According to this method, sample is selected based on the opinion of some experts( also called as ‘sampling by opinion’). This form of judgment sampling can be applied only when adequate data are available to describe the whole populations parameters and it’s sub items. 3. Quota Control Sampling In this method, the researcher attempts to ensure the sample selected is representative of the population by selecting sampling units on the basis of certain parameter (e.g.., age, sex, occupation)

Probability Methods(1) 1. Simple Random Sampling Refers to a method of selecting items from a population such that every possible sample of a specified size has an equal chance of being selected.

This method is useful when the entire population is listed and the sample have to be chosen by some randomizing method. The result produced from this method tends to produce more larger standard errors compared to the other random sampling methods.

Probability Methods(2) 2. Systematic Random Sampling Refers to a sampling technique that involves selecting the kth item in the population after randomly selecting a starting point between 1 and k. The value of k is determined as the ratio of the population size over the desired sample size.

The representativeness of a systematic sample depends on the items diverse from each other are spread out on the list. How ever this sampling method can produce a better sample than the simple random sampling.

Probability Methods(3) 3. Stratified Random Sampling Refers to a sampling method in which the population is divided into subgroups called strata so that each population item belongs to only one strata.

This sampling procedure separates the population/Universe into mutually exclusive homogeneous sets (strata), and then draw samples from each stratum.

Since the sample is allocated to each stratum, we are able to get more precise estimates for each stratum also a better estimate of the whole. In brief, Stratified sampling results in more reliable and detailed in formations. Following questions are relevant in this context. context How to form Strata ?. How to allocate the Sample size to each stratum ? How the items can be selected from each stratum ?. Stratified Random Sampling

Proportional Sampling

Non proportional Sampling

Probability Methods(3.a) 3.a. Stratified Proportionate Random Sampling Proportional allocation is considered most efficient and optimal design when there is no difference with in the stratum variance. A sample of size 1,000 is to be drawn S tratu m 1 2 3 4

O u tlettyp e C onv enience G roceries O n-P rem ise P harm acies

U n ivers e P ro p o rtio n

S tratu m S ize 25% 40% 30% 5%

T o ta l :

250 400 300 50

1000

Probability Methods(3.b) 3.b. Stratified Disproportionate Random Sampling According to method, sample items from the stratum are selected in a non-proportional way. In some cases, strata differ not only in size but also in variability, in this situations it is considered reasonable to take larger samples from the more variable strata and smaller sample from less variable data.

Probability Methods(4) 4. Cluster Sampling(Area Sampling) If the total area of interest(Universe) happens to be a big one, a convenient way in which a sample can be taken is to divide the area into a number of smaller nonoverlapping area and then randomly select a number of these areas(clusters). This method reduces the cost by concentrating only on the selected clusters but certainly it’s less precise than the other random sampling methods.

Sample Error Estimate of the likelihood that the sample deviates from the Universe on the criterion measures.

How can we reduce the Sample Error ? Increase sample size(Because of the square root formula, the error is reduced by half if the sample size is quadrupled.) Thus, if sample of 100 produce a error of 5%, the sample size must be 400 for for 2.5% standard error. Sample Error = Conf.Level x

(p x q /n)

Non - Sampling Errors In case of Market Research, common non-Sampling errors are : Choosing the sample items conveniently Bias from the interviewee Bias from the Data collectors Coding errors Data Entry Errors Data Processing errors Wrong Interpretation

Confidence Level  Confidence level or reliability is the expected percentage of times that the actual value will fall with in the stated precision levels (Likely hood that the result will fall in the range).  For a 95% confidence level, if we do a same test 20 times then it is statistically probable that the results will fall between the stated level (for ex.61-69 %), at least 19 times.  In other words, 95 chances in 100 (0.95 in 1) that the sample result represent the true condition of the population within a specified precision range against 5 chances in 100(0.05 in 1) that it does not.