Business Research Methods

ORGANISATIONAL RESEARCH-determining sample size BY CHAITANYA(08MBA043) (08MBA102)

K. N. K. SUSHANT JOLLY

TYPES OF SAMPLES • Small sample When n < 30

• Large sample When n > 30

Factors affecting the SAMPLE SIZE • • • • • •

Nature of universe Number of classes proposed Nature of study Types of sampling Availability of finance Other considerations

SIGNIFICANCE LEVEL 5% SIGNIFICANT LEVEL Its value is 1.96 It is the most acceptable significant level. It is mainly used in categorical data. 1%signicant level Its value is 2.58 It is used in continuous data.

• COMPUTATION OF STANDARD ERROR • ESTIMATION OF VARIANCE • DETERMINING SAMPLE SIZE Continuous data-eg 10,20,30 Categorical data-eg 10,67,92 Population

ESTIMATION OF ERROR • STANDARD ERROR It is a standard deviation of the sampling distribution. It measures the sampling variability due to random forces.

TEST FOR ERROR test for number of sucesses se = √ n p q where, n = size of sample p = probability of success q=1–p

• TEST FOR PROPORTION OF SUCCESS se = √ pq √ n where, p = probability of success q = 1-p n = size of sample

• TEST FOR SIGNIFICANCE PROPOSITION

BETWEEN

se = √ pq (1/n1+ 1/n2) where, p = n1p1+ n2p2/ n1+n2 • STANDARD ERROR OF MEAN

se = √ sd/ no. of observations

• STANDARD ERROR BETWEEN MEANS OF TWO SAMPLES FROM SAME POPULATION

se = √ sd2 (1/n1+1/n2) • STANDARD ERROR BETWEEN MEANS OF TWO SAMPLES FROM DIFFERENT POPULATION

se = √ sd12/n1+sd22/n2 • WHEN

TWO

STANDARD

DEVIATION

VARIANCE ESTIMATION • There are 4 ways population variance

of estimating

Determining appropriate additional responses Pilot study result Data from previous studies Estimate the structure of the organization

• V = n/s where, v = variance estimation n = number of points of scale s = number of standard deviation The appropriate variable estimation is 0.50

FORMULA FOR CALCULATING SAMPLE SIZE FOR CONTINUES • N = (zs/e)2 • S = range/6 where, Z = significance level S = standard deviation level E = tolerate error

• Samples size will increase if standard deviation is higher • significance level is higher • Tolerance level is lower

• FORMULA FOR SAMPLE SIZE when variable is categorical n = pq (z/e)2 Where, p = probability of success q = 1-p e = tolerance level Z = significance level

• TOTAL SAMPLE SIZE FOR PROPORTIONATE Statified SAMPLING n =(z/e)2 ∑ W i Si2 Where, W = weights of age group s = standard deviation

• Total sample size for disproportionate stratified sampling n = (z/e)2 (∑ WiSi)2 ni = (NiSi)

OR n/ ∑ NiSi

REGRESSION ANALYSIS • In regression analysis there is a dependent and independent variables. • We compute multiple R, multiple R2, adjusted R2, standard error of estimate, intercept and number of cases. • There are two types of regression analysis forward stepwise regression. backward stepwise regression.

COMPUTATION OF SAMPLE SIZE IN REGRESSION • CONTINUOUS DATA n=v r where, v = number of independent variable r = number of ratio

CONCLUSION • It can be concluded that for a research process a person should have a reliable and valid data and the sample size should be appropriate and easy to study.