Scribd
Upload
18 views28 pages

Opt407 5

Uploaded by

teamtruth50
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
18 views28 pages

Opt407 5

Uploaded by

teamtruth50
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 28

SAMPLE SIZE

DETERMINATION AND
SAMPLING PROCEDURE IN
PUBLIC HEALTH
RESEARCH
Ekpenyong, Bernadine Nsa
OD, MPH,Ph.D,FNCO,FAAO
[email protected]
LEARNING
OBJECTIVES
▪ At the end of the presentation, participants should be able to:

❑ Calculate sample size for different study designs


❑ Enumerate the determinates of sample size
❑ List and describe different sampling procedure/technique
INTRODUCTION
❑ Sample size determination and the selection of sample from the
study population are key to quality research in Public Health
❑ Study populations are usually too large to study
❑ How large should our sample be?
❑ A study with small sample size has little chances of providing
reliable results and as such cannot detect a significant
difference.
❑ When a sample size is too large, it is a waste of resources and
quality can be compromised’
❑ How do we select them from the population, to be
representative and reduce systematic error
STEPS IN SELECTING A SAMPLE
1.
SIZE
STUDY POPULATION -- if population is small 200 or less , survey
all and earn 0% sampling error otherwise sample ; if population is less
than N100,000 apply correction factor nf

2. STUDY DESIGN - Formula differs according to study design ( Cross


sectional, case control, cohort or experimental

3. SAMPLING TECHNIQUE – Cluster sampling (design effect)

4. RESOURCES AVAILABLE – this determines your level of precision,


the stronger it is the larger your sample size
STEP CONT..
▪6 . Non Response Rate - To make allowance for non-response rate,
divide the sample size by proportion of response = n/1 – non
response in proportion =n/(1-NR)
▪ or
▪ Add the resultant percentage of non response to the calculated
sample size eg 10%, 20% etc
▪ 7. Confidence level– 90%, 95%, 99% . Sample size increases as Cl
increases
▪ 8. Power – The probability of correctly rejecting the null
hypothesis when it is false; commonly denoted by "1 - β". ; 80% -
95% the higher, the higher the sample size
▪ 9. Estimated prevalence P – if not available 50% is used
SAMPLE SIZE CALCULATION FOR CROSS SECTIONAL
STUDIES FORMULA 1
▪
CASE SENERIO 1
▪
FINITE POPULATION FACTOR
▪

Where n is the calculated sample size and N is the study population


TABLE 1: SAMPLE SIZE FOR CSS
▪
SAMPLE SIZE FOR CSS FORMULA BY
TARO YAMANE 1967
▪
SAMPLE SIZE FORMULA FOR ANALYTIC AND
INTERVENTION STUDIES
▪
WHEN THERE ARE NO
CONTROL GROUP
Just to remove the two in front

▪
EXERCISE
▪ A researcher wish to know what proportion of adults living in Owerri Imo
State have Conjunctivitis. If the estimated occurrence is 40%, determine the
sample size that would be required to estimate the true proportion to within a
margin of error of 3% and 95% confidence.
▪ If the study population N was 8000. what would be the corrected sample size.
▪ Considering a non-response rate of 20%, what would be the final sample size?
SOLUTION
▪
SAMPLE SIZE CALCULATION FOR PREVALENCE OR
CROSS SECTIONAL STUDIES - SUMMARY
▪
SAMPLING
Target
population

Study
population

sample
TERMS
▪ Target population (reference population); the population in which we are
interested in studying

▪ Study population (survey population); the population of sampling units from


which we draw our sample

▪ A sample is a small collection of observations from a population about which


we wish to have information. A sample is part of the population
▪ Sampling frame; the comprehensive list of units from which the sample is to be
selected. Eg list of patients registered at health clinics.
▪ Sampling unit:- This may be an individual, person, a household or a school etc
that make up the sampling frame.
PRE-SAMPLING
Target population
Study population
Sampling unit
Sampling frame
Sample
TYPES OF SAMPLING
PROBABILIT NON-PROBA
▪. Y BILITY

SIMPLE RANDOM
SAMPLING CONVENIENCE
SYSTEMATIC PURPOSIVE
RANDOM SAMPLING

STRATIFIED RANDOM
SAMPLING
SNOW BALL
CLUSTER SAMPLING
MULTI-STAGE RANDOM QUOTA
SAMPLIING
TYPES OF SAMPLING
▪ Probability sampling methods- samples are chosen by using a method based on
the theory of probability. It gives everyone equal chance of being selected

▪ Non-Probability Sampling methods – samples are selected based on subjective


judgement.

▪ Probability sampling methods are preferred over non-probability sampling


methods[ its more accurate and rigorous, gives everyone equal chance of being
selected.
CASE SCENARIO 1
▪ PREVALENCE OF PRESBYOPIA AMONG LECTURERS IN
COLLEGE OF MEDICINE UNICAL
▪ STUDY POPULATION = 300
▪ SAMPLE SIZE = 150
▪ SAMPLING PROCEDURE
▪ Simple Random sampling – manual or software, to select 150 units from a
sampling frame of 1 – 300 lecturers by balloting.[ we start with the
complete list of the entire population (sampling frame), assign a unique
identifying number, then a set of numbers are selected at random by
balloting]
SYSTEMATIC RANDOM
SAMPLING
Systematic sampling- random
sampling done by using a fixed
interval and a random starting
point to select samples from a
population
eg k = pop size/ sample size =
300/150 =2 . If 1 was randomly
selected= 1st unit, 2nd unit =1 +2
=3th, 3rd unit = 3 +2=5th
STRATIFIED RANDOM SAMPLING-/
PROPORTIONATE SAMPLING
▪Stratified random sampling- traits or characteristics important to the
researcher are represented by dividing the population into sub
sections of homogenous groups called Strata

▪Before random sampling is done. Using the pervious example .

▪ if you are interested in knowing the variability in prevalence of


presbyopia by faculties and gender for instance.
STRATIFIED SAMPLING CONT…
Gender pop Proportionate Facilities POP pps
sampling

male 180 180/300 x 150 med 140 140/300x150


= 90 = 70

female 120 120/300 x 150 ALLIED 100 100/300


= 60 x150= 50

300 150 BASIC 60 60/300 x150


=30
CLUSTER SAMPLING
Usually applied when the size of the population is very large and difficult
to sign individuals as sampling units

▪ Population is divided into already existing smaller groups, which now


serves as sampling units

▪ Random sampling is done to select some clusters

▪ Data is then obtained on every sampling unit in each selected cluster


MULTI-STAGE RANDOM SAMPLING
▪ It is a sampling procedure in which all the methods above are combined in
stages
▪ Eg In a study involving the whole of Cross River State. ( sample size = 784)
▪ First stage – selection of LGA -select one LGA per senatorial or health zones.
(784/3 = 261)
▪ Second stage- selection of wards – one ward per LGA (261 per ward)
▪ Third stage – selection of villages/kindred’s/streets - 3 streets per ward- total 9
streets per LGA (261/9 = 29 per street)
▪ Fourth stage – selection of house holds ( 29 households per street)
▪ Fifth stage – selection of respondents ( one adults 35 year above x 29 houses=
29 adults x 9 x 3 = 784
FURTHER READING
Bowen, C.(2015).Straight forward Statistics. 1st Edition. SAGE Publishers USA
PP 1 -285
Mazhhindu, D and Scott, I.(2014).Statistics for healthcare professionals. An
introduction 2nd Edition. SAGE Publishers, India. Pp 1 -231
Ogbeibu, A.E.(2014).Biostatistics; A practical Approach to Research and data
handling. 2nd Edition. Mindex Publishing co. Ltd Nigeria. Pp 1 - 285
Rao k. V(2007).Biostatistics; A manual of statistical methods for use in health,
nutrition and anthropology. 2nd Edition Pp 1- 727
THANK YOU FOR LISTENING

You might also like