SAMPLING DESIGN

PROBABILITY SAMPLING & APPLICATION OF HOW

TO USE PROBABILITY SAMPLING
 SAMPLING DESIGN

• is a means of selecting a subset of units from a target

population for the purpose of collecting information. This
information is used to draw inferences about the population
as a whole.
SAMPLING
• What is your population of interest?
>> To whom do you want to generalize your results?
*All doctors
*School children
*Indians
*Women aged 15-45 years
*Other
• Can you sample the entire population?
PROBABILITY SAMPLING TECHNIQUES

• Use a variety of sources for the collection of data, both

primary and secondary.
• Describe and justify the survey methodology and frame used.
EXAMPLE:
EXAMPLE:
EXAMPLE:
 COMMON PITFALLS IN CONDUCTING
A SURVEY
• TIME & COST REQUIREMENTS One of the more
common pitfalls is to underestimate the amount of time required to
properly conduct a survey. As a general rule, surveys are an expensive
meth-od of obtaining information. Sponsors of surveys should be wary
of firms offering to conduct surveys at prices markedly below those of
other firms.
• INITIAL PLANNING. Not clearly defining the purpose of a survey
is a common pitfall. Before beginning a survey, it is strongly recommended
that the following questions be answered:

• (1) What is the purpose of the survey?

• (2) Can the information be obtained by other means?
• (3) Can the intended audience be reached?
• (4) If a survey is needed, who will use the information and how will it be used?
• (5) Are specific cross-tabulations on certain characteristics desired?
• (6) Given the resources available, is the study feasible?
• QUESTIONNAIRE DESIGN The questionnaire should
be designed with the study objectives in mind. The challenge
for the researcher is to design the questions and word them in
such a way as to obtain unbiased information.
• When developing
questions for a
survey,
abbreviations and
colloquialisms
should be avoided.
• Requiring respondents to answer questions on items for which they have
incomplete or little knowledge is a common pitfall.

• SAMPLING PROCEDURES. It is rarely feasible to survey the entire

population and, thus, a sample of the population is selected to be surveyed.
Probability samples are the only type of samples where the results of the survey
can be generalized to the entire population.
• PRE-TESTING. Following the questionnaire design and sample selection, a
pretest should be conducted on a subset of the target population to find out if the
questions elicit the desired information. Although pretesting is frequently over-
looked, it is the only way to discover possible misinterpretations and
misunderstandings which may arise from the questions.
• NONRESPONDENTS. According to the American Statistical
Association, a low response rate produces more questionable results than a
small sample, since there is no scientifically valid way to infer the
characteristics of the population that the non-respondents represent.
• PROCESSING THE DATA. The process of data entering, editing, and
analyzing the completed questionnaires is subject to a great deal of human
error.
 OTHER SOURCES OF ERROR
• SELECTION BIAS Persons being studied may differ in some significant
ways from the larger population, which they are supposed to represent, or
from the comparison group with whom they are supposedly comparable.
• Selection bias takes many forms.
(1) Persons who seek medical care, or get referred for specialist
opinion, are likely to be those who are the sickest compared to those in the
community.
• (2) Migration Bias.
• (3) High dropout rates. A high refusal rate is very likely to introduce bias.
• (4) Response Bias.
• (5) Observer Error
• (6) Measurement bia due to faulty instrumen
• INFORMATION BIAS. A distortion in the estimate of the outcome can
occur if there are errors in measurement, or misclassification of subjects on
one or more variables. Important sources include invalid measurements,
incorrect diagnosis, omissions, and imprecision or other mistakes in data
entry. Unequal diagnostic surveillance of exposure groups in follow-up
studies is another significant cause of
bias in longitudinal studies.