1

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Never Ever Give Up !

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Data Analysis Basic with SPSS (PART 1)

Prof. Dr Chua Yan Piaw Institute of Educational Leadership (IEL,UM)

Unit for the Enhancement of Academic Performance (ULPA,UM)

University of Malaya

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Targets: 1. Concept of Data analysis

2. Prepare Data for Analysis

3. Checking the normality of a data

4. Establishing reliability of a questionnaire

5. Descriptive analysis

6. Inferential analysis

7. Analyse data and report the data

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Research Methods and Statistics Reference Books

2

Volume 1: Research Methods

Chapter 1: Introduction to Research

Chapter 2: Research Ethics

Chapter 3: Literature Review

Chapter 4: Research Design

Chapter 5: Experimental Study

Chapter 6: Quasi-experimental Study

Chapter 7: Survey Study

Chapter 8: Field Study

Chapter 9: Case Study

Chapter 10: Action Study

Chapter 11: Historical Study

Chapter 12: Probability Sampling Procedures

Chapter 13: Non-probability Sampling Procedures

Chapter 14: Measurement in Research

Chapter 15: Index, Scales and Specific Measurement

Procedures

Chapter 16: Pilot Study

Chapter 17: Research Instrumentation

Chapter 18: Format of Writing Research Report

2006 2011

2015

Volume 2: Fundamental Research Statistics

Chapter 1: Descriptive Statistics

Chapter 2: Inferential Statistics and

Significance Test

Chapter 3: Qualitative Data Analysis

Chapter 4: Data Preparation for SPSS

Program

Chapter 5: Reliability of Research

Instrument

Chapter 6: Chi-Square Tests

Chapter 7: T Tests

Chapter 8: ANOVA Tests

Chapter 9: Correlation Tests

Chapter 10: Multiple Regressions

Chapter 11: Reporting the Results of Data

Analysis Based on the APA Format

2012

2006

2015

Volume 3 (2nd edition):

Fundamental Research Statistics: Data Analysis for Likert Scale

Chapter 1: Measurement Scales and

Statistical Test

Chapter 2: Data Preparation for SPSS

Program

Chapter 3: Data Transformation

Chapter 4: Mann-Whitney U Test

Chapter 5: Wilcoxon T Test

Chapter 6: Kruskal-Wallis H Test

Chapter 7: Friedman Test

Chapter 8: Spearman Correlation Test

Chapter 9: Contingency Table Data Analysis

Chapter 10: Cramer V Correlation Test Chapter 11: Reporting the Results of Data Analysis

Based on the APA Format

2013

2008

Volume 4: Advanced Research Statistics:

Univariate and Multivariate Tests

Chapter 1: Research Statistics Concept and

Data Preparation for SPSS Program

Chapter 2: One-Way ANOVA Test

Chapter 3: Two-Way ANOVA Test

Chapter 4: SPANOVA Test

Chapter 5: ANCOVA Test

Chapter 6: Independent Samples

MANOVA Test

Chapter 7: Repeated Measures

MANOVA Test

Chapter 8: MANCOVA Test

Chapter 9: Trend Analysis

Chapter 10: Method of Writing

High Impact

Journal Paper

2009

2014

3

Volume 5: Advanced Research Statistics:

Regression Test, Factor Analysis and

Structural Equation Modeling Analysis

Chapter 1: Data Preparation for SPSS

Program

Chapter 2: Partial Correlation Test

Chapter 3: Hierarchical Multiple

Regressions Analysis

Chapter 4: Hierarchical Binary

Logistics Analysis

Chapter 5: Log-Linear Analysis

Chapter 6: Factor Analysis

Chapter 7: Discriminant Analysis

Chapter 8: Cluster Analysis and Chapter 9: Structural Equation Modeling

Analysis Using AMOS

2014

2009

Mastering Research Methods

Chapter 1: Introduction to Research

Chapter 2: Research Ethics

Chapter 3: Literature Review

Chapter 4: Research Design

Chapter 5: Experimental Study

Chapter 6: Quasi-experimental Study

Chapter 7: Survey Study

Chapter 8: Field Study

Chapter 9: Case Study

Chapter 10: Action Study

Chapter 11: Historical Study

Chapter 12: Probability Sampling Procedures

Chapter 13: Non-probability Sampling Procedures

Chapter 14: Measurement in Research

Chapter 15: Index, Scales and Specific

Measurement Procedures

Chapter 16: Pilot Study

Chapter 17: Research Instrumentation

Chapter 18: Format of Writing Research Report

2011

Mastering Research Statistics

Chapter 1: Descriptive Statistics

Chapter 2: Inferential Statistics and

Significance Test

Chapter 3: Qualitative Data Analysis

Chapter 4: Data Preparation for SPSS

Program

Chapter 5: Reliability of Research

Instrument

Chapter 6: Chi-Square Tests

Chapter 7: T Tests

Chapter 8: ANOVA Tests

Chapter 9: Correlation Tests

Chapter 10: Multiple Regressions

Chapter 11: Multiple Responses Analysis

Chapter 12: Reporting the Results of Data

Analysis Based on the

APA Format

Amazon.com or MPH Online

Concept of Data analysis

4

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Research design and statistics

A quantitative researcher should understand statisics before planning his research design.

Planning a design without the

knowledge of statistics, the

researcher will find difficult to

analyse the data after collecting it. R

ES

EA

RC

H S

TA

TIS

TIC

S

DR

CH

UA

YA

N P

IAW

(UM

)

Research Statistics

Determined by several factors:

Factor 1. Type of study

Factor 2. Type of measurement

scales in an instrument

Factor 3. Sample size

Factor 3. Type of Statistical tests

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Factor 1

1. Types of study:

a. Descriptive study

b. Inferential study

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

How to recognize whether the study is a

descriptive study or an inferential study?

In a descriptive study,

No sample is drawn from the

population.

Respondents are the whole

population of the study.

5

A descriptive study

•No sample is drawn from the population, so the

results is not generalised. So no inferential

statistical test is needed.

X X

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

In a descriptive study,

No sample is drawn from the

population.

Respondents are the whole

population of the study.

The results/ findings are not

generalised to any subject outside the

population .

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Frequency

Percentage

Mean

Median

Mode

Descriptive statistics describes the

characteristics (variables) in the

population.

Standard deviation

Distribution of Scores

The results are nearly 100% correct for

that population.

For example:

Motivation level of the students.

The result is nearly 100% correct (if the measurement is reliable and is correctly done).

1. Collect data (motivation score) from

the students.

2. Calculate the mean score.

3. It represents the motivation level of

the students

Did I select a sample?

Can I generalize the result to other group? Students from other universities?

6

B. For an inferential study:

Respondents are subjects selected randomly from a population, to form a sample.

Statistics tests is used to analyse data collected from the sample.

The results / findings are generalised back to the population from where the sample was selected.

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W

An inferential study

A sample A population

Study is conducted on the sample.

Results from the sample is

generalised to the population.

Statistical test is

needed to generalised

the results.

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Krejcie &

Morgan

(1970)

N = Population size

S = Sample size

Determine the

sample size

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

T tests

Chi-square tests

ANOVA tests

Pearson r

Inferential statistics

Tests of differences

Spearman rho

Kruskal-Wallis test

Mann-Whitney U

Tests of relationship

Cramer V

7

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Summary:

For a descriptive study, no inferential

test / test of significance is needed.

For an inferential study (a sample is

used), test of significance is needed to

generalise the result to the population. R

ES

EA

RC

H S

TA

TIS

TIC

S

DR

CH

UA

YA

N P

IAW

(UM

)

Research Instruments : Questionnaire / other quantitative

data measurement devices/ tests

Factor 2:

Types of measurement in an instrument

Types of analysis is determined by

the Scales of measurement in an

instrument

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

SCALES OF MEASUREMENT

1.*Nominal scale

- categories/groups of data

2.*Ordinal scale

- Rank

- distances among scales are different

3.*Interval / Ratio scale

- distances among scales are identical

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Scales of measurement 1.*Nominal scale

- categories/groups of data

Gender: 1= Male

2= Female

Ethnicity: 1= M 2= C 3= I

Climate change phenomenon:

1= Pollution 2= Fires

3= Drought 4= Flood

2.*Ordinal scale

- Rank

- distances among

scales are different

Income: 1= < $1000 2= $1001-3000 3 = >$3000

Attitude towards sport management 1= Negative 2= Neutral 3= Positive

3.*Interval / Ratio scale

- distances among scales are identical Math Score: 1, 7, 13, 26…

Age: 5, 12, 13, 42 yo…

Overweight: 1= Yes 2= No 3= Unsure

Level of agreement 1= Strongly disagree, 2= Disagree, 3= Undecided, 4= Agree, 5= Strongly agree

Gender: 2= Male

1= Female

1= M 2= I 3= C

Score: 0, 1, 2, 3, 4…100

Temperature : 50C, 120C,

130C, 420C… Income: 5,000; 12,000; …

8

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Identifying the scale of a measurement

Continuous

data?

Yes No

Are the distances among

scales identical?

Can the data be

categoried?

Example:

male/female

Yes No Yes

Interval /

ratio scale

Ordinal

scale

Nominal scale R

ES

EA

RC

H S

TA

TIS

TIC

S

DR

CH

UA

YA

N P

IAW

(UM

)

What kind of measurement for this data?

The most appropriate price for an original DVD is:

Less than RM 5 …………………..…1

RM 5 to RM 12 ………….……………2

RM 13 to RM 20 ………….……..……3

RM 21 to RM 30 ……………….….....4

More than RM 30 ……………..……..5

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Statement

1. The main factor of air pollution at your place is:

A. Heavy metals

B. Oil

C. Chemicals

D. Fertilizers

2. Your attitude towards the performance of our sport school is

positive.

Strongly agree 1 2 3 4 5 Strongly disagree

3. Do you agree that the main cause of obesity is too much

intake of oily foods.

1. Yes

2. No

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Yearly Incomes (X$1,000)

24 76 35 58 94 93 77 44 66 88 85 55 55

Company type

1. Big

2. Small

1 = <$20 2 = $20-$40 3=$41-$80 4>$80

Human Activities

1.Urbanisation

2.Tourism

3.Agriculture

4.Industry

5.Fisheries

Sport team

1. Under 16

2. Under 18

3. Under 21

9

RE

SE

AR

CH

ST

AT

IST

ICS

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Scales of measurement

Interval and Ratio data are parametric

data (the data are assumed normal

distributed) Example:

Maths scores of a class

Parametric tests are used: T tests, ANOVA tests Pearson Correlation test, etc

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Scales of measurement

Nominal and Ordinal data are non-

parametric data (the data are assumed

not normal distributed)

Level of agreement Strongly disagree - disagree -------- undecided -------- agree - strongly agree 1 2 3 4 5

Non-parametric tests are used: Chi-square tests (Nominal/ordinal data) Mann-Whitney test (Ordinal data) Kruskal-Wallis test (Ordinal data) Spearman rho Correlation test (Ordinal data) Cramer V correlation test (Nominal data) etc

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Example: Stress level of the residents during air

pollution period

data: Ordinal scale (1=Very low, 2=Low, 3=Average, 4=High, 5=Very high)

The Difference between male and female residents: use the non-parametric test - Mann-Whitney test

data: Ratio scale (stress score, exp: 34, 45,

8, 33, 56, 78, 34, 23, 67, 55)

The Difference between male and female residents: use the parametric test - T test

10

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Factor 3: Sample size

In inferential statistics, Two

assumption are:

1. The data for continouos data

(interval / ratio) is normal

distributed when n => 30.

Each sub-sample: n =>15

2. Data for categorical data are not-

normally distribued.

Each sub-sample => 5

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Any question?

RE

SE

AR

CH

ST

AT

IST

ICS

D

R C

HU

A Y

AN

PIA

W (U

M)

Take 5?