You are on page 1of 31

# Research Methodology

QUANTITATIVE RESEARCH DESIGN

1

Correlational Research
 To study to what extent variation in one factor is related to variation on one or more factors based on a coefficient index.  PURPOSE – to establish there is relationship, and strength of relationship between two or more quantitative variables based on a coefficient value.  Purpose – to ascertain there exist a relationship or to use relationships to predict.

Correlational Research
Example: Relationship between achievement in English and achievement in mathematics

Variables that have strong relationships vs variable that is the cause and it affects another variable.
Example: Relationship between self concept and achievement)

Coefficient of Determination
Assesses the proportion of variability in one variable that can be determined or explained by A second variable.  e.g.
– R = .7 – R2 = .49 = 49% – I.E. 49% of the variability in Y can be determined or explained by X – E.G. Parents' education level explains 49% of students satisfaction

Examples of Correlational Research
Relationships between grades obtained through criterion reference for several subjects.  A study to predict success at the graduate level based on intercorrelational patterns of variables at the undergraduate level.

SCATTERGRAM OF ‘r’
Y
HIGH

LOW

LOW

HIGH

X

PERFECT POSITIVE CORRELATION (+1.00)

SCATTERGRAM OF ‘r’
Y
HIGH

LOW LOW HIGH X

PERFECT NEGATIVE CORRELATION (-1.00)

SCATTERGRAM OF ‘r’
Y
HIGH

LOW LOW HIGH X

HIGH POSITIVE CORRELATION ( .83)

SCATTERGRAM OF ‘r’
Y
HIGH

LOW LOW HIGH X

SUBSTANTIAL NEGATIVE CORRELATION (- .76)

SCATTERGRAM OF ‘r’
Y
HIGH

LOW LOW HIGH X

MODERATE POSITIVE CORRELATION (.57)

SCATTERGRAM OF ‘r’
Y HIGH

LOW

LOW

HIGH

X

ZERO CORRELATION (0.00)

Coefficient of Correlation (bivariate)
Pearson’s product moment coefficient of correlation  Point biserial correlation  Biserial correlation  Phi (φ) coefficient  Contingency coefficient  Spearman’s rho coefficient of correlation

Multivariate Correlation

MULTIPLE REGRESSION
DISCRIMINANT ANALYSIS CANONICAL CORRELATION PATH ANALYSIS

Ex-post Facto Research
 Also known as causal comparative
research.  Purpose = to study cause-effect relationships by observing ‘effect’ that exist and relooking again the data to understand the causal factor.  Differ from experimental research in which data collection is done in a controlled situation.

Ex-post Facto Research
Some Examples:
 To identify the attributes of an effective teacher which has been defined as performance in the annual assessment and comparing data on students’ performance for the last 10 years from his/her personal file.  To see behavioural patterns and student achievement of year one students who had undergone or not undergone kindergarten before entering school.  To identify the attributes of an effective teacher which has been defined as performance in the annual assessment and comparing data on students’ performance for the last 10 years from his/her personal file.

Ex-post Facto Research
Researched after a situation or an event has happened. The researcher will look at one or more ‘effects’ and study the data by reflecting on previous evidences to seek the answers to the cause, relationships and give explanations.

Ex-post Facto Research
Strengths:
1. Very suitable for situations where experimental research which is stronger, is unable to be done:
a.If in a situation where factors such as selection, control and manipulation of variables are unable to be done to see direct cause-effect relationships. b.If laboratory control for the purpose of research is not practical, too expensive, or not ethical if conducted.

Ex-post Facto Research
Strengths:
2. This method provides useful information on phenomena which had happened : a. Which variable is related to which variable? b. Under which situation? c. How is the sequence, how is the pattern like?

Weaknesses of Ex-post Facto’ Research
 Main – lack of control over independent variable – not able to manipulate the variable which affect the ‘effect’– therefore, the need to study as many possible causes which may provide the results – problem = cannot be sure of the cause-effect relationships.  Problem of ascertaining that all causal factors have been included in the research.  Complications due to the fact that may be not one factor is the cause – maybe combinations of several factors or interactions of several factors are the cause.

Steps to Do Ex-post Facto Research
1. Identify the problem 2. Survey related literature 3. State the hypothesis 4. List all possible assumptions that become the basis of the hypothesis 5. Approach: a) Choose the subjects
b) Select data collection technique c) Specify the category to classify the data which may not be directly related, but may have relationships.

6. Check for the validity of data collection 7. Provide explanations, analysis, and interpretation of results.

Experimental Research
 Unlike ex post facto and correlational studies which stress on relationships / associations, experimental research could explain causeeffect relationships between variables.  A design that involves the manipulation of one variable (treatment) followed by observations made on the effect of the manipulation on one or more dependent variables

Experimental Research
The variables that are manipulated are the experimental variable or treatment variable or independent variable.
Most experimental research in education uses a group that is compared to the experimental group and do not receive the treatment = control group.

Experimental Research
Example:
pre-experiment (one-group pretest and posttest) O1 x O2 Problems in experiment – to ascertain the suitable control group so that any change in the posttest should be identified due to the treatment which the researcher has manipulated (true) r O1 x O2 R O1 O2

Internal Validity Threats Experimental Research
To what extent extraneous variables are being controlled by researcher. If extraneous variables are not controlled, the researcher is not certain that the affected change on the experimental group is caused by the treatment or the extraneous variable.

in

Experimental Research
CONTROL:

nonrandomized control group pretest-posttest
(Quasi) E O1 X O2 C o1 o2

Extraneous Variables Experimental Research
a. b. c. d. e. f. g. h.

Threatening

History Maturation processes Pretesting procedures Measuring instruments Statistical regression Differential selection of subjects Experimental mortality Interaction of selection and maturation

Extraneous Variables Experimental Research

Threatening

 Interaction of selection & x  Interaction of pretesting & x

 Reactive experimental procedures
 Multiple-treatment interference

Extraneous Variables Experimental Research

Threatening

 History – Situation/event other than treatment may occur between the first and second measurement.  Maturation Processes – Changes (biological/psychological) which may occur to the subjects at the right time.  Pretesting Procedures – Answering the pretest before treatment may interfere with subject’s performance in the posttest.

Extraneous Variables Experimental Research

Threatening

Measuring Instruments – May be due to instruments that are not reliable; changes in instruments; changes in level of difficulty Statistical Regression – May occur if students with extreme scores in the pretest regress towards the mean of the posttest even if without treatment

Extraneous Variables Experimental Research

Threatening

 Differential Selection of Subjects
– Important differences had existed between the e group and the c group even before the treatment was administered

 Experimental Mortality
– Reduction in number due to drop outs

 Selection-Maturation Interaction
– Preexisting maturation undetected during selection especially in intact groups

Extraneous Variables Experimental Research
 Experimenter Bias  Hawthorne Effect
Subject effects

Threatening

Unintentional effects that the researcher has on the study

 Henry Effect
Control group knowing they are being experimented on exerting extra effort and performing above expected average