Sampling

WHY TAKE A SAMPLE?

REASON FOR SAMPLING

• PRACTICAL CONSIDERATION

• CANNOT ANALYZE HUGE AMOUNT OF DATA GENERATED BY A CENSUS.

• POSSIBLE TO DRAW VALID INFERENCES ON THE BASIS OF CAREFUL OBSERVATION OF VARIABLES WITHIN A RELATIVELY SMALL PROPORTION OF THE POPULATION.

REASON FOR SAMPLING

• EFFICIENT

• ECONOMICAL

• PRACTICAL

• ACCURATE

• TIMELINESS

PURPOSE OF SAMPLE

SAMPLING

• IS THE PROCESS OF SELECTING A FRACTION OF UNITS OF INTEREST OF THE POPULATION OF THE RESERACHER TO BE ABLE TO DRAW GENERAL CONCLUSION ABOUT THE ENTIRE POPULATION

POPULATION

• AS THE ENTIRE GROUP UNDER STUDY AS SPECIFIED BY THE OBJECTIVES OF THE RESEARCH PROJECT

SAMPLE

• A SUBSET OF THE POPULATION AND THE SAMPLE UNIT PERTAINS TO THE BASIC LEVEL OF INVESTIGATION

SAMPLING ERROR

• ANY ERROR IN A SURVEY THAT OCCURS BECAUSE A SAMPLE IS USED

• IT CAN BE CAUSED BY TWO FACTORS– METHOD OF SAMPLING

SELECTION– SIZE OF THE SAMPLE

SAMPLE FRAME

• A MASTER LIST OF ALL THE SAMPLE UNITS IN THE POPULATION.

SAMPLE FRAME ERROR

• THE DEGREE TO WHICH IT FAILS TO ACCOUNT FOR ALL THE POPULATION.

SAMPLING PROCESS

SAMPLING METHOD• PROBABILITY

– AN OBJECTIVE PROCEDURE IN WHICH THE PROBABILITY OF SELECTION IS KNOWN IN ADVANCE FOR EACH POPULATION UNIT.

• NON-PROBABILITY– IS A SUBJECTIVE PROCEDURE

IN WHICH THE PROBABILITY OF SELECTION FOR EACH POPULATION UNIT IS UNKNOWN BEFOREHAND

SAMPLING METHOD

• PROBABILITY – SIMPLE RANDOM

• THE RESEARCHER USES A TABLE RANDOM NUMBERS, RANDOM DIALING,OR SOME OTHER RANDOM SELECTION PROCEDURE THAT GUARANTEES THAT EACH MEMBER OF THE POPULATION HAS AN IDENTICAL CHANCE OF BEING SELECTED INTO THE SAMPLE

– BLIND DRAW

– TABLE OF RANDOM NUMBER

SAMPLING METHOD

• PROBABILITY – RANDOM WALK SAMPLING

THIS FORM OF SAMPLING IS USED EXTENSIVELY IN MARKET RESEARCH AS A CHEAP APPROXIMATION TO TRUE RANDOM SAMPLING. THE SAMPLE INVOLVES CONDUCTING RANDOM WALKS IN SMALL AREAS.

SAMPLING METHOD• PROBABILITY

– SYSTEMATIC SAMPLING• USING A LIST OF THE MEMBERS OF THE

POPULATION, THE RESEARCHER SELECTS A RANDOM STARTING POINT FOR THE FIRST SAMPLE MEMBER. A CONSTANT SKIP INTERVAL IS THEN USED TO SELECT EVERY OTHER SAMPLE MEMBER.

Step 1. Number the subjects serially up to 100

Step 2. Divide 100 by 10 ex. N/n=100/10=10

Step 3. Randomly select your starting point say 10 on the list

Step 4 Then select every 10th subject after the first

SAMPLING METHOD• PROBABILITY

– CLUSTER SAMPLING• THE POPULTION IS DIVIDED INTO

GEOGRAPHIC AREAS, EACH OF WHICH MUST BE CONSIDERED TO BE VERY SIMILAR TO THE OTHERS. THE RESEARCHER CAN THEN RANDOMLY SELECT A FEW AREAS AND PERFORMS A CENSUS OF EACH ONE.

– Identify population to be sampled

– Identify the salient characteristics

– Locate the areas where subjects with the characteristics cluster

– Use random selection procedure to select your sample subject from each cluster

SAMPLING METHOD• PROBABILILITY

– STRATIFIED SAMPLING• THE RESEARCHER IDENTIFIES

SUBPOPULATIONS CALLED STRATA THEN, A SIMPLE RANDOM SAMPLE IS THEN TAKEN OF EACH STRATUM. WEIGHTING PROCEDURES MAY BE APPLIED TO ESTIMATE POPULATION VALUES SUCH AS THE MEAN.

– If 10% of the population is made up of adults between 60-75, then 10% of the total sample should be subjects in this age category. After this categorization the simple random sampling procedure is used to determine those who make the selected sample.

SAMPLING METHOD• NON-PROBABILILITY

– CONVENIENCE SAMPLING• THE RESEARCHER OR

INTERVIEWER USES A HIGH TRAFFIC LOCATION SUCH AS A BUSY PEDESTRIAN AREA OR A SHOPPING MALL TO INTERCEPT POTENCIAL RESPONDENTS. ERROR OCCURS IN THE FORM OF MEMBERS OF THE POPULATION WHO ARE INFREQUENT OR NON-USERS OF THAT LOCATION.

SAMPLING METHOD

• NON-PROBABILILITY– PURPOSIVE/JUDGMENT

SAMPLING• THE RESEARCHER USES HIS OR

HER JUDGMENT OR THAT SOME OTHER KNOWLEDGEABLE PERSON TO IDENTIFY WHO WILL BE IN THE SAMPLE .

SAMPLING METHOD

• NON-PROBABILILITY– REFERRAL SAMPLING

• RESPONDENTS ARE ASKED FOR THE NAMES OR IDENTITIES OF OTHERS LIKE THEMSELVES WHO MIGHT QUALIFY TO TAKE PART IN THE SURVEY.

SAMPLING METHOD

• NON-PROBABILILITY– QUOTA SAMPLING

• THE RESERACHER IDENTIFIES QUOTA CHARACTERISTICS SUCH AS DEMOGRAPHIC OR PRODUCT USE FACTORS AND USES THESE TO SET UP QUOTAS FOR EACH CLASS OF RESPONDENTS.

OTHER SAMPLING TECHNIQUES

• MULTISTAGE SAMPLING– IS MOST COMMON WHEN THE

POPULATION OF ELEMENTS IS VERY LARGE OR WHEN THE ELEMENTS ARE NOT INDIVIDUALLY IDENTIFIED

– SAMPLING CONTINUES FROM WITHIN LARGER UNIT UNTIL THE SMALLEST UNITS ARE SELECTED

OTHER SAMPLING TECHNIQUES

• DOUBLE SAMPLING– This is a modified version of

multistage sampling procedure, which aims at a high level of precision through sampling intensity. This procedure is sometimes employed by researchers who have time and money at their disposal.

FACTORS THAT DETERMINES THE SELECTION OF A SAMPLE

SIZE• Type of project

• Purpose of project

• Complexity of project

• Amount of error that may be tolerated

• Time constraint

• Financial resources available

• Previous research in the area

SAMPLE SIZE

• THE LARGER THE SAMPLE ,THE SMALLER IS THE MAGNITUDE OF THE SAMPLING ERROR AND A GREATER REPRESENTATION OF THE POPULATION

SAMPLE SIZE

• SURVEY STUDIES SHOULD HAVE LARGER SAMPLES COMPARED WITH EXPERIMENTAL STUDIES

SAMPLE SIZE

• WHEN SAMPLES ARE TO BE DIVIDED, THERE SHOULD BE ENOUGH SAMPLE TO HAVE AN ADEQUATE SIZE PER SUBGROUP.

SAMPLE SIZE

• FOR MAILED QUESTIONNAIRE, A LARGE INITIAL SAMPLE IS NEEDED BECAUSE THERE WOULD BE A POSSIBLITY OF HAVING A LOW TURNOVER OF RESPONSE.

SAMPLE SIZE

• SUBJECT AVAILABILITY AND COST FACTORS ALSO DETERMINE APPROPRIATE SAMPLE SIZE

SAMPLE SIZE

• RESEARCHERS OFTEN USE SAMPLES OF 50,75 OR 100 SUBJECTS PER GROUP(SUCH AS ADULT 18-42 YEARS OLD).

COMPUTATION OF SAMPLE SIZE

• SLOVIN FORMULA

n=(1+Ne2)

where:

n= sample size

N= population size

e= desired margin of error

* Ignorance of population or behaviour

COMPUTATION OF SAMPLE SIZE• MILDRED PARTEN’S FORMULA

n=NZ2pq

Nd2+2pq

where:n= sample sizeZ= critical value at a given confidence leveld=maximum tolerable errorp=proportion of the respondent that has a large sample size.N=population size

*knowledgeable of the population’s behaviour

• Cost and time consideration always control sample size.

• Although researchers may wish to use a sample of 1000 for a survey, the economics of such a sample are usually prohibitive.

• If a smaller sample is forced on a researcher by some one else or circumstance beyond him or her, the result must be interpreted accordingly with caution.

• Multivariate studies always require larger samples than do univariate studies because they involve analyzing multiple response data.

• 50 =very poor; 100 = poor; 200 = fair; 300 = good; 500 = very good;1000 = excellent. (Comrey and Lee, 1992).

• Other researchers suggest using a sample of 100 plus 1 subject for each dependent variable in the analysis (Gorsuch, 1983).

•The larger sample compensates for those subjects who drop out of research studies for one reason or another, and allowances must be made for this in planning the sample selection. .•Researchers can expect 10% - 25% dropout within the sample in panel studies.

•The larger the sample the better it is for the researcher and the readers or other future researchers. •A large unrepresentative sample is as meaningless as a small unrepresentative sample, so researchers should not consider number alone•Quality is always more important in sample selection than mere size.

END