The Statistical Data and Sampling

Framework of Statistical Analysis

Planning

Collection of Data

Organization and Presentation of Data

Analysis of Collected Data

Interpretation and Conclusions

Recommendations based on Conclusion

Nature of Data

Classification of Data

Primary -- gathered directly from an original source by the

analyst or researchers.

Ex. Interview, Questionnaires, diaries, etc.

Secondary -- gathered from published or unpublished materials

that have been previously obtained by other individuals or
agencies.

Ex. From books, journals, newpapers, registration of

births, deaths, marriages. etc.

Advantage of primary data is accuracy, reliability and relevance.

Qualitative Data measures quality, an attribute or a characteristics on

each experimental unit

produced data that can be categorized according to similarity

and differences in kind

Ex. house type (single, detached or duplex); eye color;

sex/gender; civil status; level of management; educational
attainment; etc.

Quantitative Data -- measures a numerical quantity or amount in

each experimental unit.

Ex. Interest Rate; height; weight; temperature, scores in a test;

etc.

Classification of Quantitative Data:

Discrete Variable - a finite or countable number value

Continuous Variable can assume infinitely many values

Ex. numbers or orders per day; number of

telephone calls

Ex. Height; weight; time; volume

Methods of Data Collection

Interview Method -- a direct method of investigation

Ways of conducting interviews:

Mail

Telephone

Computer (on-line)

Personal

provides more consistent and more precise information

it is inexpensive in terms of questionsl prepared

necessary for in particular situations, such patients suffering

from strokes.

Disadvantages:

time consuming

uncomfortable for some persons

limited field coverage

Analyzing information gathered difficult to quantify

requires training

Questionnaire

Registration

Observation

Experimentation

Classification of Measurable Data

Nominal Scale

lowest level of measurements

consist of labels or names to classify observed elements

Ex. Gender Distribution of 100 Adults; Marital Status

Ordinal or Ranking Scale

elements can be arranged in some meaningful kind of natural

orders

Ex. Academic performance of Students (poor, fair, good, very

good and outstanding); size of 100 shirts (small; medium, large,
XL)

is QUALITATIVE VARIABLE

Interval Scale

elements can be differentiated according to characteristics; can

be ranked or order -- with zero point origin

Ex. Temperatures in degrees celcius; mental ability scores;

blood pressures

A quantitative variable

Ratio Scale

Considered as the highest level of measurement

have order property; unit of measurement; meaningful

differences (interval); but have a FIX ZERO Point ORIGIN.

Ex. Temperature in degree Kelvin; volume; time; length

A quantitative variable

Sampling

Sampling is the act, process, or technique of selecting suitable sample,

or representative part of a population.

Sample is a set of n objects to be selected from the population

Slovins formula is often used to calculate the sample size:

Margin of error (e) is the probability of committing an error.

While (100-e)% is the probability of getting the correct result.

Note: the lower the margin of error, the larger is the required
sample size.

Example1: In a population of 22,000 students enrolled at Saint Louis

University in a particular semester, what sample size is needed to get
an accurate result for a study using a margin of error:

1%

n = 22000/(1+22000(0.01)^2)

n = 6875

n = 22000/(1+22000(0.05)^2)

n = 393

5%

2.5%

n = 22000/(1+22000(0.025)^2)

n = 1492

Another Formula:

Example: Using 95% confidence level, what is the sample size needed
to estimate the mean serum cholesterol level of a population within
9.8mg/dl of the true mean and it is given that = 20?

n =((1.96)*(20)/9.8)^2

n = 16

Sampling Technique:

Probability Sampling -- selection using randomized mechanism.

Simple random sampling -- individual in the population has

equal and independent chance of being selected.

also called lottery or raffle sampling

used to eliminate biased

Systematic Random Sampling

a good substitute of simple random sampling

Divide the population size by the sample size and round

off the result to the nearest whole number, use the
resulting number to start with and as interval in the
selection.

Example1: In a population of 200 items in a production

process.

N = 200 ; e=5% ; n = 133

Number = 200/133 = 2

2nd , 4th , 6th , 8th . . . . . . .

N = 200 ; e=5% ; n = 40

Number = 200/40 = 5

5th , 10th , 15th , 20th . . . . . . .

Stratified Sampling

Selects simple random samples from each of the subpopulations

or strata of the population.

Example1: Given the following data, determine the share of

each stratum if the desired sample size, n is 1500

Stratum (Age
Range)
30-39
20-29
10-19

Populatio
n
1500
4500
9000
15000

% Share
0.1
0.3
0.6

Sample Size (n),

% share x Total Sample
Size
0.10 x 1500 = 150
0.30 x 1500 = 450
0.60 x 1500 = 900
n = 1500

Example2: A researcher wishes to conduct a survey among the

students from the different departments. He wishes to find out
how many from each department is needed to represent the
population. Suppose the distribution of the population is as
follows:

Department

Number of Students

Business

1500

%
Share
0.32

Administration
Management
Finance
Entrepreneurs
hip
Culinary Arts

1200
850
850
200

0.25
0.18
0.18
0.04

Sample
Share
0.32 x 369 =
117
93
66
66
16

150
N = 4750, e=0.05
Slovins Formula, n =
369

0.03
1.00

12
369

Cluster Sampling (Area Sampling)

Cluster in any group of persons or experimental units having

similar characteristics.

Useful when population under consideration are scattered

geographically -- a simple random samples of groups is
selected.

Example: getting samples from socio-economic groups in a

province like a community clustered to high-income, middleincome and low-income classes.

Multi-stage Sampling:

Uses several stages or phases in getting random samples from

the general population.

Very useful in nation-wide surveys: Example National Survey

Random Sampling of Regions, then

Random Sampling of provinces, then

Random Sampling on municipality level

Non-probability Sampling -- selection of samples is solely determined

by rules or guidelines set by the researcher/investigator. Very prone to
bias.

Purposive Sampling pick sample based on certain criteria and

subjects that satisfy the criteria.

Quota Sampling -- example, choice of the actual persons is left

to the interviewers own convenience and preference (not
random). Example a QUOTA of 50 samples.

Convenience Sampling used when really impossible to select a

random sample (selection is at the researchers convenience).
Example: finding out demand for a new product in a store.

Judgement Sampling -- samples are picked based on the opinion

of someone who is familiar with the relevant characteristics of
the population. Example: small sample from a highly
heterogeneous population.

Snowball Sampling useful when population are inaccessible or

hard to find.

Example: investigator begins by identifying someone who

meets the criteria for inclusion in the study.

Summation Notation: X = 2, 5, -3, 4, 1, 1

x = 2 + 5 3 +4 + 1 + 1 = 10

x2 = 4 + 25 + (-3)2 + 16 +1 + 1 = 56

( x )2 = (2+5-3+4+1+1)2 = 100

3x

= 3 (2+5-3+4+1+1) = 3 (10) = 30