1. Kursus ‘Research
Methodology’ KKKT
By: Azwadi Ali
Department of Accounting and Finance,
Faculty of Management and Economics,
Universiti Malaysia Terengganu.
Studio SPS, KKKT. 29 November 2011.
2. Data collection
Observation – assembly line in factories, arrival of aircrafts,
activities of rural community.
Interview – open-ended with some guided questions. Suitable
for exploratory research especially qualitative.
Questionnaire – widely used, easy and objective.
Focus on questionnaire .
Quantitative – research model exists, items have been developed
from existing literature.
May need to perform pilot study.
3. Constructing Instrument
Begin with literature search.
Use google scholar, then use provided databases at your
institution to locate the articles.
Certain articles provide actual questionnaire – use/adopt wisely.
If the aim is not to merely replicating existing research, find some
ways to improve/extend/modify research models – a new
variable is already sufficient.
Ensure that variables forming your research model make sense
such as the appropriateness of mediator/moderator/antecedent.
4. Latent variables
Also known as unobserved variables.
Items in questionnaire are observed variables.
Latent variables can either reflect or formed by items.
If reflective mode is suitable, ensure that items of the same
variables make sense (i.e., co-variance can be expected).
Items of three or more are good enough to represent a latent
variable.
Too many items are not advised.
5. Reflective vs Formative
Example: Computer Self-Efficacy
Reflective – I am capable at performing tasks on my computer.
I feel confident in my ability to perform computer-
related tasks.
Formative – I am confident at my ability to perform tasks in MS
Word.
I am skillful at using Excel.
Example: System Quality
Reflective – Overall, I would rate the system quality of the system
highly.
The quality of the system is appropriate for my needs.
Formative – Reliability, Ease of Use, Complexity, Accessibility,
Responsiveness
6. Devising questionnaire/item scales
Normally questionnaire includes several sections – e.g.
demographic, cases/experiment/quiz, and items making up the
research model. The order of the sections depends on the
researcher.
Only ask relevant demographic questions, especially if they are
useful to answer a research question – e.g. Do individuals differ
in opinions between male and female? Do large companies more
transparent than small companies?
Questions of items in research model may be asked via Likert
scale, equal-appearing, semantic differential or cumulative.
11. Research model
Questionnaire is easily constructed when research model has
been identified.
In social science, many well established theories or concepts can
be adopted/extended/modified.
Theory of Reasoned Action – Theory of Planned Behavior
Social Cognitive Theory
Technology Acceptance Model
Diffusion of Innovation
Elaboration Likelihood Model
Attitude Mediation Hypothesis
12. Examples
Attitude
Behavioural
towards
Beliefs
Behaviour
Normative Subjective
Intention Behaviour
Beliefs Norm
Perceived
Control
Behavioural
Beliefs
Control
16. Validity & Reliability of Instrument
What is validity?
- A study is valid if its measures actually measure what they
claim to, and if there are no logical errors in drawing conclusions
from the data.
- Face and content validity (expert/pilot study)
- Construct validity (≈ reliability)
- Internal validity (defend against source of biases)
- Statistical validity (proper use of statistics)
Reliability?
- Reliability is the correlation of an item, scale, or instrument
with a hypothetical one which truly measures what it is
supposed to.
Focus on construct validity and internal consistency.
17. Construct validity
Convergent validity and Discriminant validity
Convergent -> internal consistency (cronbach alpha, simple
factor structure), concurrent (correlation between scale),
predictive validity (criterion in the future) and external validity
(possible biases?).
Discriminant validity -> correlational method (rule of thumbs),
factor methods (principal component), average variance
extracted (AVE) and nested model in structural equation
modelling (SEM).
18. Hands-on factor method and cronbach alpha
Use SPSS.
These are ‘pre-requisite’ to test a research model.
19. Research hypotheses
H1: ‘Information usefulness’ is positively related to ‘attitude
towards IR Websites’
H2: ‘Usability’ is positively related to ‘attitude towards IR
Websites’
H3: ‘Attractiveness’ is positively related to ‘attitude towards IR
Websites’
H4: ‘Attitude towards IR Websites’ is positively related to
‘intention to re-use IR Website’
20. Research Model with Indicators
COG3 COG4
IQ1
COG2 COG5
IQ3
IQ4 COG1 COG6
IQ7 IQ
IQ8 AFT1
IQ9 COG AFT AFT2
AFT3
IU
CRD2
CRD4
CRD5
CRD
CRD6
USB1
USB2 INT1
USB3 INT2
USB5
USB AT_ST INT
INT3
USB6 INT4
USB7
ATR1
ATR2
ATR3
ATR
ATR5
ATR6
21. Sample Results
Information
Usefulness
γ=
0
t = .341
2.8
65
Attitude
Intention to
γ = 0.297 towards IR β = 0.640
Usability Re-use
t = 2.425 Websites t = 8.873
(σ2 = .409)
(σ2 = .784)
11
0.3 30
γ= 5.0
t=
Attractiveness