Transcript So What Are Research Methods?

Quantitative Research 2
Dr N L Reynolds
Lecture Objectives
• To define what causal research is and
provide examples of its use in business
• To understand how to identify appropriate
statistical tests
• To understand the requirements of statistical
analysis techniques
Importance of…
Causal research
• Exploratory research can create hypotheses to test
• Descriptive research can show that variables are associated
• Only causal research can provide support, or not, for whether
one variable is the cause of another
Data analysis
• Analysis obtains meaning from the collected data. All previous
steps in the research process have been undertaken to support
the search for meaning. The specific analysis that can be used is
closely related to the preceding steps, and the careful analyst will
remember this when designing the other steps
• Statistical testing allows you to state how likely it is that the results
you have found are also found in your population
Key Issues
1. What is the difference between description and
demonstrating causality
2. How is causal research done and what are its
uses in Business/Management research
3. Why use statistics?
4. What are the stages of data analysis
5. How do you choose the right data analysis
method?
Key Issue 1:
Demonstrating causality
• Concomitant variation
– A predictable statistical relationship between
two variables
• Time order of occurrence
– A change in the independent variable must
occur before a change in the dependent
variable
• Elimination of other possible causal factors
Key Issue 2:
Conducting experiments
• Assigning subjects to groups
– Control groups
•
•
•
•
Making interventions
Making observations
Internal and external validity
Types of experiment
– Field or laboratory
– True and quasi-experimental designs
Malhotra & Birks (2000)
Factor
Laboratory
Field
Type of environment
Artificial
Realistic
Error caused by type of environment
High
Low
Control
High
Low
Responses guided
High
Low
Internal validity
High
Low
External validity
Low
High
Time
Short
Long
Number of units
Small
Large
Ease of implementation
High
Low
Cost
Low
High
Key Issue 2:
Causal questions in B/M research
• Which of three advertisements works best at increasing
awareness of the MA Management at Bradford
University with Brazilian undergraduates?
• Which of two methods of training is more effective at
eliminating bullying behaviour among factory workers?
• Is written communication of new auditing standards
better at ensuring their understanding and correct
application than verbal communication of standards?
• Do different sampling regimes affect the speed at which
faults on a production line are detected?
• Which of three methods of preparing salespeople for
overseas assignments produces the fastest acceptance
of the new working environment?
Key Issue 3:
What do statistics do?
The world
before analysis
The world
after analysis
Data
interpretation
Data collection
Data organisation
and manipulation
Kachigan (1991)
Key Issue 3:
Why do statistics?
• To reduce a data set to a manageable summary
(e.g., measures of central tendency, measures
of spread)
• To determine the degree of confidence in the
accuracy of the measurements we take
(including whether two measurements differ)
• To identify associations or relationships between
and among sets of observations
Kachigan (1991)
Key Issue 3:
Statistics are tools
• Methods of determining the answer to the
research question.
• By knowing the analysis technique, its strengths
and weaknesses, you can use it to solve the
problem: apply the tool to the problem, do not
manipulated the problem to fit a tool you are
comfortable using.
Key Issue 4:
Stages of data analysis
1. Designing the measurement methods
and collecting the data
2. Preparation of the data
•
Coding, data entry, etc.,
3. Describing the data
4. Answering your research question
Key Issue 4:
Variables, scale types and samples
•
•
Dependent and independent variables
Levels of measurement
–
–
–
–
–
•
Nominal (assignment)
Ordinal (assignment and order)
Ordinal interval or assumed interval
Interval (assignment, order and distance)
Ratio (assignment, order, distance and origin)
Impact of sample size
Key Issue 4:
Variables, scales and relationships
•
Variables
–
–
•
Scales
–
–
•
Measures of central tendency and spread
Distribution and frequency tables
Multi-item scales and reliability (and validity)
Multi-dimensional scales
Relationships
–
–
–
Cross-tabulation and 2 tests
Correlations
Between groups
Key Issue 5:
Parametric or non-parametric?
Start
Use
non-parametric
statistics
Nominal or
ordinal
Level of
measurement
Interval or ratio
Population
distribution?
Normal
Use
parametric
statistics*
Other
* Assuming additional
assumptions are satisfied
Small
Source: Diamantopoulos, 2000
Sample size?
Large
Key Issue 5:
Type I and Type II Error
Type I error
Type II error
• Occurs when the sample • Occurs when the sample
results lead to the
results lead to acceptance
rejection of a null
of the null hypothesis when
hypothesis when it is in
it is in fact false (also called
fact true (also called
beta error)
alpha error)
• Power is the probability of
• Significance is the
rejecting the null hypothesis
probability of making a
when it should be rejected
type I error
Key Issue 5:
But which test?
1. What is the question you want
answered?
2. How many variables are you using?
– Independent? Dependent?
– What is the level of measurement?
3. How many samples are you dealing
with?
– What is the relationship between the
samples?
Causal Research and Data Analysis:
Their contribution to your dissertation
Causal research
• Most will read papers that claim causality when the methods used
do not support those claims
• Most will read papers that use causal research
• A few will develop research questions that are best answered by
causal research
Data analysis
• All will read papers that use quantitative data analysis
• Almost all will need to analyse (either qualitative or quantitative)
data
• Many will need to summarise large data sets that cannot easily be
interpreted
• Many research objectives should be tested statistically