H.

Gladius Jennifer

Associate Professor

School of Public Health

SRM IST
 Introduction
 Medical research: Aim to make general statement based on
observations about a wider set of subjects.

 Large Sample: cost, time, person

 Small Sample: unable to detect clinically important

results

 Not very large and not small

 Population Sample
 The set of all aggregate of, all  A representative part of
individuals with the specified population .
characteristics.  A finite subset of
 A finite / infinite subjects statistical individuals in
(Humans / Animals) a population.

 Some proportion of the

 Example: - total population.
Total no of students / households,
 Ex. Select 10 houses in
population at risk.
a village.
 Sampling Technique
 The process of selecting unit from a population of
interest, so that by studying the sample, result can be
generalized back to the population, from which they were
chosen.
 Sample size: - The total no of individuals in a sample.

 Sampling unit: -A member of the sample.

 Sampling frame:- The list of people from which the

sample is taken. It should be comprehensive, complete &
up to data.
Probability Sampling
Techniques
 Simple Random Sampling
 Population has an equal chance of being

selected.

Procedure: -

 Obtain a complete sampling frame.

 Give each case a unique no starting at one

 Decide on the required sample size.

 Select from table of random numbers.

 Systematic Random Sampling
Selecting one unit on a random basis and
choosing additional elementary units at every
spaced interval until the desired number of units
is obtained.

Procedure: -

 Number the units in the population 1 to N

 Decide on the n that is needed

 Take every kth element = N/n

 Stratified Random Sampling
Objective: -

To give better cross section of the population, So as to gain

higher degree of relative precision.

Principles of stratification: -

 The strata should be non-overlapping, comprise the whole

population.

 Strata are homogeneous within themselves with respect to the

characteristics under study

 Administrative convenience may be considered as the basis for

stratification
Procedure: -

 All people in sampling frame are divided in to strata.

 Within each stratum a SRS or Systematic Random

Sampling is selected.
 Cluster Random Sampling
 When entire population is divided in to groups or clusters
are selected. All the observations in the selected clusters
are included in the sample.

Procedure: -

 Population is first divided into clusters

of homogeneous units.

 A sample of such clusters are selected by random sampling

 All the units in the selected clusters are then studied

Probability
Sampling Merits Demerits
Techniques
Simple  Simple to accomplish  SF large - difficult &
Random  personal bias are less expensive
Sampling  being representative  sub groups of interest ..
Systematic  Simple and convenient  May not represent
Random  Time and work is reduced much whole population
Sampling  The sample is evenly spread over  Investigator bias
the population
 Adequate representation of Stratified population may
Stratified minority subgroups of interest overlap, such a sample will
Random can be ensured by stratification. not be a representative
Sampling  For non homogeneous one.
population it may yield more
reliable results

Cluster  Neighboring elements is easier, less efficient than SRS.

Random cheaper and faster
Sampling  It is less costly than SRS
 Research Question…. Sampling Technique

Prevalence of depression and its associated risk Simple Random

factors among medical students of Private Sampling
medical collage, TN.
Study of mobile phone usage on sleep Systematic Random
disturbance, stress and academic performance Sampling
among Medical Students, Tamil Nadu

Pre-Hypertension among Young Adults in Stratified Random

Coastal Villages of Udupi District in Southern Sampling
India
Association between socio demographic Cluster Random
characteristics of female community health Sampling
volunteers and their knowledge and
performance on maternal and child health
services
Prevalence study on HIV/AIDS in India Multistage Sampling
 Other Probability sampling
 Probability Proportional to size

 Two stage / Phase sampling

 Multistage / Phase Sampling

 Design based / Model Based Sampling….

Non Probability Sampling
Techniques
 Convenient Sampling
 By stopping anybody, who is prepared answer the question
asked.

 Useful in getting general ideas

of phenomenon of interest.
 Purposive sampling
 Judgmental sampling

 One in which is selected by the researcher subjectively

 The researcher attempts to obtain the sample, that

appears to him, to be representative of the population.

 Selection is made by human choice than at random

 Quota sampling
 Based on defining the distribution of

characteristics required in the sample

and selecting respondents until a quota

has been filled

 Similar like stratified sampling. The

selection of respondents with in strata

is non random.

 Decide on the characteristics of which

sample is to be representative
 Snow Ball Sampling
 It involves starting process with one individual or group
and using their contacts to develop the sample.

 Explorative, qualitative and descriptive

 Population of individuals who have been difficult to

enumerate through

 Net work analysis

 Non
Probability Merits Demerits
Sampling
Techniques
Convenient  Save time  No randomness
Sampling  Money
 Effort  Bias is high
 Not generalizable
Purposive  Easy to undertake results  Not generalizable
Sampling

Snow Ball  Can access people who are  Relied on personal

Sampling difficult to access. recommendation.
 Can access groups of like  Will not be able to
minded people. make generalization.

Quota  Marketing Research  Not generalizable

Sampling
 Research Question…. Sampling Technique

Influence of Religious Beliefs on the Convenient

Health of Cancer Patients Sampling

Study on satisfaction of patients Purposive Sampling

admitted in a tertiary care hospital

KAP study about condom Usage Snowball Sampling

among Sexual workers
Other Non probability Sampling
 Volunteer Sampling

 Opportunity Sampling

 Accidental Sampling

 Universal sampling
Sample Size Calculation
 Types of Error
Type I Error (α): - the probability that difference shown occurred due to chance

 It is known as level of significance / critical region / region of rejection

 P [reject H0 when it is true]

Type II Error (β): - if a true difference of a stated magnitude existed but the study
would not have picked up as statistically significant.

 P [accept HA when it is false]

Power of Study (1 – β): - the probability that if a true difference of stated magnitude
existed then the study would have picked up as statistically significant.
Reality True Reality false
Researcher True Correct Type II error (β)

Researcher False Type I error (α) Power (1 – β)

 Sample Size Calculation
Estimation: (Single Group)

 Prevalence studies (Prop / PR)

 Descriptive studies (Mean & SD)

Test Hypothesis: (Two Groups)

 Case control studies (PR)

 Cohort studies (IR)

 Clinical trial studies (Mean & SD)

I. Qualitative data: - (proportion given)

Sample size n = Z2 * p *q / d2

P – Proportion / percentage

q=1–p

d = clinical difference

Z = normal variate
 Example – Single Proportion
 From pilot study it was reported that among headache
patients 28% had vascular headache. It was decided to
have 95% CI and 10% variability in the estimated 28%
how many patients are necessary to conduct the study?

Ans: - P = 28%, q = 72%, d = 2.8% Z = 1.96

n = (1.96)2 * 28 *72 / (2.8)2

n = 987.8
 Example – Proportion (True Value)
A local health department wishes to estimate the prevalence
of TB among children under 5 years of age in its locality.
How many children should be included in the sample so that
the prevalence may be estimated to within 5% points of the
true value with 95% confidence, if it is known that the
true rate is unlikely to exceed 20%?

Ans: - P = 20%, q = 80%, d = 5% Z = 1.96

n = (1.96)2 * 20 *80 / (5)2

n = 246
II. Quantitative Data: - (Mean & SD given)

Sample size n = Z2 *S2 / d2

S = standard deviation,

n = sample size,

d = clinical difference

Z = normal variate
 Example – Single Group
 In a health survey of school children it is found that the
mean HB of 55 boys is 10.2/ 100ml with SD of 2.1
consider the precision as 0.8

 Mean = 10.2, SD = 2.1, Z = 1.96, d = 0.8

 n = (1.96)2 *(2.1)2 / (0.8)2

n = 26
 Test of Hypothesis
III. Qualitative data: - (proportion given)

Sample size n = (Zα + Zβ )2 * 2* p *q / d2

Zα = Z value for α level

Zβ = Z value for β level

P = Proportion / percentage

q=1–p

d = clinical difference
Example – Two Proportion
Dead Alive Total
High Dose 4 (16.7) 20 24
Low Dose 1 (4.2) 23 24
Total 5 (10.4) 43 48

P = 16.7+4.2 /2 = 10.4%, q =1-p= 89.6%,

d = 12.5, Z = 1.96 for  @ 0.05
Zβ = 1.28 for β @ 0.10
n = (1.96 + 1.28 )2 * 2* 10.4 *89.6 / (12.5)2
n = 125 in each arm
IV. Quantitative data: - (Mean & SD given)

Sample size n = (Zα + Zβ )2 * 2* S2 / d2

Zα = Z value for α level

Zβ = Z value for β level

S2 = pooled variance

d = clinical difference
Example – 2 Groups Mean & SD
Low Dose High Dose
Mean 61.7 33.2
SD 91.8 89.1
n 23 20

S2 = 90 d = 28.5 zα = 1.96 for  @0.05

zβ= 1.28 for β @0.10
n = (1.96 + 1.28 )2 * 2* 902 / (28.5)2
n =210 in each arm
 Errors In Sampling
1. Sampling error 2. Non sampling error.

Sampling Error: -

 Origin in sampling

Reasons: -

 Faculty selection of sample

 Substitution

 Error due to improper choice of the statistics for

estimating the population parameters

 Non Sampling Error
Arises: - stage of observation, ascertainment &
processing data.

Reasons:-

 Faculty planning or definitions

 Response error, Compiling error, Publication error

 prestige bias

 self interest

 Bias due to interviewer

 Failure of respondent memory

 Non response bias Error in coverage

Take Home Message…
 study design & Setting

 Consult statistician before starting your

research

 calculate sample size and power at the

beginning

 No rule of thumb…30 (or) reference articles sample

size
THANK YOU