42
BIR6014 QUANTITATIVE RESEARCH METHODS IN LANGUAGE TEACHING TOPIC: VARIABLES, VALIDITY AND RELIABILITY PRESENTER: SREENIVASA RAO A/L HANUMANTHA RAO MATRIX NUMBER: M20141000152 LECTURER: DR. GOH HOCK SENG

Variables, Validity & Reliability

Embed Size (px)

DESCRIPTION

Variables, Validity & Reliability

Citation preview

BIR6014

QUANTITATIVE RESEARCH METHODS IN LANGUAGE TEACHING

TOPIC:

VARIABLES, VALIDITY AND RELIABILITYPRESENTER:

SREENIVASA RAO A/L HANUMANTHA RAOMATRIX NUMBER:

M20141000152LECTURER:

DR. GOH HOCK SENG

VARIABLES

RESEARCH OBJECTS

PROPERTIES OF THE OBJECT

VALUE OF PROPERTIES

RESEARCH OBJECT

Definition

OBJECT:

•Persons - Students

•Things – Curriculum programs, handout materials

•Places – School, day care centres

PROPERTIES OF THE OBJECT

DEFINITIO

N

PROPETIES :

•Characteristics or attribute of an object

EXAMPLE

S :

•Students - Genders

•School – Length of school day

VALUE OF PROPERTIES

DEF 1:

•Is a number that represent the magnitude of the variables

•Exp: Weight = 18kg-20kg

DEF 2:

•Or the category of the variable

•Exp: Gender = male/female

Exp: Measuring the height of a group of students

Exercise

Q1: What are the variables?Q2: What is the magnitude and

category involved?

VARIABLES

HEIGHT (CM) MAGNITUDE

GENDER (MALE/FEMALE) CATEGORY

MALE FEMALE0

20

40

60

80

100

120

Exp: Measuring the height of a group of boys and girls

(Category of variable)

(Magnitude) cm

GENDER (V)

HEIGHT (V)

CONCLUSIONS:The larger the magnitude, the greater the height

Variables must have at least 2 categories of measure,

if only one, then they are constants, not variables

Values of variables must have:

- Exhaustive – Means each object can be assigned a value

- Mutually exclusive – Means that each object have one

& only one value

Fraenkel, J.R. and Wallen,N.E. (2009). How to design and evaluate research in education(7th ed. New York, NY:McGraw-Hill.

Types of variables

Dependent & Independent

variables

Categorical Variables

Continuous Variables

Attribute Variables

Extraneous Variables

Confounding Variables

Dependent & Independent variables

Independent variables

AffectsDependent variables

(Presumed or possiblecause)

(Presumed results)

NOTE:

- Reciprocal causation – Causal relation that flows in both directions, each variables causes the other.

Fraenkel, J.R. and Wallen,N.E. (2009). How to design and evaluate research in education(7th ed., pp.42-43). New York, NY:McGraw-Hill.

Research Topic: Effects of Think Pair Share Technique on descriptive writing

Exercise

•Continuous Variable

•Measured on a scale. Exp: Test scores range from 0-100

•Categorical Variable

•Measured and assigned to groups of specific characteristics. Exp: Gender

•Attribute Variable

•A measured or classified pre-existing quality of research. Exp: Anxiety level of students effects test performance, Independent Variable: Anxiety level, Dependent Variable: Test performance, Attribute variable: Test taking experience

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

•Extraneous Variable

•Alternate factors which unintentionally influence the dependent variables. Exp:size of the class, age of teacher, length of class hours.

•Can be control by : - Hold them constant

•Confounding Variable

•When any other extraneous variables changes along with the deliberate change in the independent variable is confounded with extraneous variable. Exp: If 2 methods of teaching were studied by comparing one method in the fall and the other method in spring, then the teaching method confounded with time of the year (extraneous variable)

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

Validity

•How well the instrument used measures what it purported to measure

•Ability of a scale or measuring instrument to measure what is intended to be measured

•Validation-process of collecting and analyzing data or evidence to support inference

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

Why validity is important?

•refers to the level to which data or evidence support any inferences a researcher makes

•The collected data is able to be concluded into meaningful specific inference intended

What is validity and why it is important in research?. Retrieved September 25,2014 from http://psucd8.wordpress.com/2011/11/20/why-is-validity-important-in-research

Types of evidence

Content-related

evidence

Criterion-related

evidence

Construct-related

evidence

Content-related evidence

•The instrument includes an adequate and representative items that hit the concept

•Key elements :•the

adequacy of the questions

•format of the instrument (clarity of prints, clarity of directions, appropriate language etc.)

•The validity of the evidence is usually determine by the content experts

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

Criterion-related evidence

•The test results (the one being validated) is compared with other test results (criterion)

•Criterion – other assessment to measure the same variable

•2 types of criterion-related validity•Predictive

validity – instrument data and criterion data obtained over a period of time

•Concurrent validity – instrument data and criterion data obtained at the same time

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

•The key index for both types is the correlation coefficient (r)

•Positive relationship – the scores of both instrument and criterion are at the same degree

•Negative relationship – the scores of the instrument and the criterion are at different degree

•Validity coefficient- the relationship between the scores obtained by the same individuals on particular instrument and their scores on the criterion

r = correlation coefficient between two halves (reliability for ½ test)

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

The choice of the criterion is crucial and will determine the validity

•The criterion must be relevant

•The criterion must be reliable

•The criterion should be free from bias

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

Construct-related evidence•A

variety of different types of evidence are collected

•To obtain construct-related evidence of validity;•The variable being

measured is clearly defined

•Hypotheses are formed (based on theory underlying the variable)

•The hypotheses are tested both logically and empirically

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

construct-related evidence

Convergent validity

multiple measures of the same construct

operate in similar ways

Divergent validity

the measures that should not be related are not related in

reality

Construct validation involves various type of procedures and evidence, including content-related and criterion-related evidence

Note:

Colin,P.,& Julie,W. (2005-06). Exploring Reliability In Academic Assessment. Retrieved September 20,2014 from http://www.uni.edu/chfasoa/reliabilityandvalidity.htm

•Any relationship observed between two or more variables should be clear as to what it means rather than due to other thing

•The differences observation on the dependent variable is directly to the independent variable and not to unintended variableInternal Validity

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

1

•Subject characteristic

2

•Loss of subjects (mortality)

3

•Location

4

•Instrumentation

5

•Testing

6

•History

7

•Maturation

8

•Attitude of subjects

9

•Regression

10

•Implementation

Threats to Internal Validity

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

1

•Standardize the conditions under which the study happen

2

•Obtain more information on the subjects of the study

3

•Obtain more information on the details of the study

4

•Choose an appropriate design

How researcher can minimize the threats?

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

Def:

•To which extent the results of a study can be generalized to wider population, cases or situation

1

•Population generalizability - the degree which a sample represents the population of interest

2

•Ecological generalizability – the degree which the results of a study can be expanded to other settings or conditions

External Validity

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

1

•Selection effects

2

•Setting effects

3

•History effects

4

•Construct effectsThreats to external Validity

Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in education (6th ed.).New York, NY: Routledge

•Reliability indicates how consistently a test measures whatever it does measure

•A test or instrument is considered reliable if it can give same result over and over again

•The consistency and stability of the scores/data obtained

•Reliability coefficient – a relationship between scores obtained by the same individuals on the same instrument at two different times, or on two parts of the same instrument

Reliability

Trochim, William M. The Research Methods Knowledge Base, 2nd Edition. Retrieved September 21,2014 from http://www.socialresearchmethods.net/kb/reliable.php

Why reliability is important?

Errors of measurement

Reliability estimates gives researchers

idea how much variation to expect

Trochim, William M. The Research Methods Knowledge Base, 2nd Edition. Retrieved September 21,2014 from http://www.socialresearchmethods.net/kb/reliable.php

Ways to obtain a reliability coefficient

Test-retest method

Equivalent-forms method

Internal-consistency

methods

Test-retest reliability•The

same test is given twice to the same individuals after a period of time

•The reliability coefficient will be affected by the duration of the time interval between the test

•For most educational study, stability of scores over a two-to-three month period is viewed as sufficient evidence of test-retest reliability

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

Equivalent-forms reliability

1

•Using two different but equivalent forms of test/instrument are given to the same individuals at the same period

2

•Inter-rater reliability-all researchers must come to the same agreement by ensuring each researcher enter the data in the same way•The

inter-rater agreement can be calculated as a percentage:

Reliability can be achieved through;

Number of actual agreementsx 100

Number of possible agreementsFraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

Internal-consistency measures of reliability- The instrument/tests run only once

•Test items are divided into two halves (each half is matched in terms of item difficulty and content) and correlates the individuals’ scores on the two halves

•The marks obtained from one half of a test should match the marks on the other half

•The reliability coefficient is calculated using Spearman-Brown prophecy formula

Reliability of scores on total test =

2r

1 + r r = correlation coefficient between two halves (reliability for ½ test)

i. Split-half method

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

Internal-consistency measures of reliability- The instrument/tests run only once

•r is obtained with Pearson correlation coefficient

•If the r obtained is 0.56 by comparing one half of the test items to the other half, then reliability score for the whole test would be

i. Split-half method

Reliability of scores on total

test=

2 x 0.56

=

1.12

= 0.721 + 0.56 1.56

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

Internal-consistency measures of reliability

•The KR20 and KR21 formulas

•The latter formula required three pieces of information;•K = The

number of items on the test

•M = The mean

•SD = The standard deviation of the set of test scores

* Formula KR21 can be used only if there is assumption that all items in test are of equal difficulty

ii. Kuder-Richardson approaches

Reliability of the whole test =

K( 1 -

M(K-M)

)K - 1K(SD²)

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

Internal-consistency measures of reliability

•A general form of the KR20 formula to be used in calculating the reliability of items that are not scored right versus wrong, such as in essay test, where more than one answer is possible

iii. Alpha coefficient (Cronbach alpha)

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

The summary of methods checking the validity and the reliability of an instrument

Validity (Truthfulness)

Method Procedure

Content-related evidence Obtain expert judgment

Criterion-related evidence Relate to another measure of the same variable

Construct-related evidence Assess evidence on predictions made from theory

Reliability (Consistency)

Method Content Time interval

Procedure

Test-retest Identical Varies Give identical instrument twice

Equivalent forms Different None Give two forms of instrument

Equivalent forms/retest

Different Varies Give two forms of instrument, with time interval

Internal consistency

Different None Divide instrument into halves and score each or use Kuder-Richardson approach

Fraenkel, J.R. and Wallen,N.E. (2009). How to design and evaluate research in education(7th ed., pp.158). New York, NY:McGraw-Hill.

The Standard Error of Measurement (SEMeas)

•Index to show to what extent the measurement would vary under certain circumstances

•The longer the elapse time between measurement, the score is considerably fluctuates more

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

A person score in an IQ test

• The formula for standard error of measurement:

= standard deviation = reliability coefficient

Fig. 1 Fig. 2

Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate research in education(7th ed.). New York, NY:McGraw-Hill

References:• Fraenkel, J.R. and Wallen,N.E (2009). How to design and evaluate

research in education(7th ed.). New York, NY:McGraw-Hill• Cohen, L., Manion, L. and Morrison, K. (2007). Research methods in

education (6th ed.).New York, NY: Routledge• Zamalia Mahmud (2009). Handbook of research methodology: a

simplified version. Shah Alam, Malaysia:UiTM Press• Research designs in education. Retrieved September 21,2014 from

http://adhi301126117.wordpress.com/2010/07/26/the-quantitative-research-and-appreciative-inquiry/

• Trochim, William M. The Research Methods Knowledge Base, 2nd Edition. Retrieved September 21,2014 from http://www.socialresearchmethods.net

• What is validity and why it is important in research?. Retrieved September 25,2014 from http://psucd8.wordpress.com/2011/11/20/why-is-validity-important-in-research/