1
Applications for management
Lesson 1
What is applied research?
• Systematic enquiry into real world problems.
• “Original investigation undertaken in order to acquire new knowledge but is directed primarily
towards a specific, practical aim or objective.” Australian Research Council
• “Research directed toward a current need. The purpose of the research is to discover results
that can be applied to the need.”
Ex:
Design of modern hotels and guest services (based on savers of guests needs, preferences and
attitudes).
What is multivariate statistics?
• Univariate statistics – summaries and inferences about a single [uni-] variable.
• Bivariate statistics – summaries and inferences about the relationships between two [bi-]
variables.
• Multivariate statistics – summaries and inferences about the relationships between three or
more [multiple] variables
Model based approach
The approach is model based: models are used to understand how phenomena interrelate to
each other.
They provide a simplified and meaningful representation of the reality so to disentangle the reality
and the essential interrelationships.
Statistical Modelling
1. to understand the phenomena and improve theory development
2. to explain the effect of different characteristics
3. to predict
4. to reduce the complexity (dimensionality) of data
1. To understand causality and theory development
Models and theory are mutually related.
On one side models are specified so to address research questions that in several cases are
stated as results of a theory.
On the other side, results of the fit of a model from empirical data inform the development of a
new theory.
2. To explain the effect of different characteristics
Which are the main determinants of the hourly pay?
Which factors affect the choice of a given product or service?
3. To make predictions
• Managers use models to predict the potential outcome in terms of sales or market shares of a
given change in the production process or a given marketing campaign.
• Policy makers can anticipate the changes in economic variables, inflation rates, unemployment
but also in social variables as educational rates in connection with a given policy.
, 2
4. To reduce the complexity of data
How can we measure the level of satisfaction of customers for a product?
Can we identify a few dimensions of the learning process of a child?
Two basic kinds of multivariate statistical models
• Dependency (or “causal”) models
• concerned with causality (the effects of independent variables, interventions, or time on
dependent variables)
• Interdependency or interrelationship models
• concerned with structure: relationships among variables or among cases
Dependency models
• Dependent variable (denoted Y) : a variable that is to be predicted or that changes in response
to an intervention; the value of this variable depends on the values of other variables
• Independent variable (denoted X) :a variable that is used to predict values of another variable,
that records experimental categories such as participant/non-participant
• Confounding variable: a variable that might interfere with or ’confound’ the relationship between
the independent and the dependent variable
A simple dependency model with three independent variables
Interventions
Much management research seeks to discover the effect of an intervention on an outcome.
An intervention is an action taken in order to achieve change, e.g.,
• training is an intervention designed to improve knowledge and skills
• governments often use policy change as interventions to improve social conditions
The intervention is an independent variable (X).
The outcome to be changed (e.g., skills, social conditions) is a dependent variable (Y).
Interdependency model: Correlation
What is the relationship between a country’s
population and the number of Internet users in
the country?
I hypothesise that there is a correlation between
these two variables.
, 3
An overview of the research process
1. Define the problem : the research question.
2. Prepare the research design
3. Identify the sources of data and, if necessary, a sampling plan
4. Collect the data
5. Process and analyse the data
6. Formulate conclusions and prepare the report
Problem definition
The first step consists in the specification of the issue to be addressed:
“What, exactly, do we need or want to know?”
“Why is it important or interesting to know this?”
The issue is then stated in terms of question or hypothesis: the research question.
Common elements of a problem definition
• Research question
• Selection of a theory or other basis for definition of the specific study: a in-depth literature
search has to be run on “recognized sources”.
• Clear identification of the unit of analysis (individuals, couples, countries ecc)
• Clear identification of the fundamental concepts to be studied
• Expected or proposed relationships between the concepts
Research design
The research design describes how you will:
• Model the research problem
• Collect and analyse the data in order to answer the research question
The quality of the model we build depends not only on the research design but also on the data
we gather (or use) to conduct the research.
Model and data collection are intertwined.
For dependency (“causal”) models
• Multiple regression
• Analysis of variance (ANOVA)
For interdependency models
• Factor analysis – relationships among variables
• Cluster analysis – relationships among cases
We often combine interdependency and dependency models, e.g., 1
Step 1:
Use factor analysis to reduce 6 observed variables to 2 factors, attitudes to having a child (X1)
and social influences on childbearing (X2)
Step 2:
Use multiple regression to explain intention to have a child (Y) as a function of X1 and X2:
Y=βο +β1X1 +β2X2 +ε
, 4
Lesson 2
Variables types
Two broad types of variable:
1. Categorical (nonmetric) – takes on non-numerical values, so to describe membership of a
group or category
2. Numerical (metric) – takes on numerical values, providing a numerical measure of size or
extent to which a case has a characteristic:
• Discrete – takes a finite or infinite but countable number of values
• Continuous – can take any value between two given numbers.
Levels of measurements
Categorical variables
• Nominal – non numerical values, cannot be ordered, e.g., sex (male or
female)
national health model
• Ordinal – non numerical values that can be ordered, e.g., age group
(<21, 21-30, >30)
bank size (sml, med, lge)
Numerical (metric) or “scale” variables
• Interval – numerical variables measured on a scale that does not have a
true ‘zero’ point, e.g.
temperature in degrees Celsius scores on a scale of 1 (strongly
disagree) to 7 (strongly agree)
• Ratio – variables measured on a scale that does have a true ‘zero’ point
(a value for which there is absence of the quantity being measured,
e.g., actual age, health expenditure.
Anchored scales
• Particularly in survey research, we often use “anchored scales”, scales in which selected values
are named or “anchored” as a guide to the respondent.
• A fully anchored scale that ranges from strongly disagree to strongly agree is known as a “Likert
scale” ...
• Scales that have anchors other than strongly agree to strongly disagree are just know as n-point
scales
Metric-Interval
Does not have a ‘true 0’ & cannot be used to estimate ratios.
Metric-Ratio
Has a ‘true 0’ and can be used to calculate ratios, e.g.,
Composite scales (some time called summated scales)
Instead of using a single variable to represent a complex concept, responses to a number of
indicator variables are combined into a composite measure.
Composite measures increase validity (we are really measuring what we want to measure!) a
combination of measures provides a more accurate estimate of the true score than a single
measure.
Data quality is measured by:
• Validity
• Reliability
• Accuracy
• Precision
Applications for management
Lesson 1
What is applied research?
• Systematic enquiry into real world problems.
• “Original investigation undertaken in order to acquire new knowledge but is directed primarily
towards a specific, practical aim or objective.” Australian Research Council
• “Research directed toward a current need. The purpose of the research is to discover results
that can be applied to the need.”
Ex:
Design of modern hotels and guest services (based on savers of guests needs, preferences and
attitudes).
What is multivariate statistics?
• Univariate statistics – summaries and inferences about a single [uni-] variable.
• Bivariate statistics – summaries and inferences about the relationships between two [bi-]
variables.
• Multivariate statistics – summaries and inferences about the relationships between three or
more [multiple] variables
Model based approach
The approach is model based: models are used to understand how phenomena interrelate to
each other.
They provide a simplified and meaningful representation of the reality so to disentangle the reality
and the essential interrelationships.
Statistical Modelling
1. to understand the phenomena and improve theory development
2. to explain the effect of different characteristics
3. to predict
4. to reduce the complexity (dimensionality) of data
1. To understand causality and theory development
Models and theory are mutually related.
On one side models are specified so to address research questions that in several cases are
stated as results of a theory.
On the other side, results of the fit of a model from empirical data inform the development of a
new theory.
2. To explain the effect of different characteristics
Which are the main determinants of the hourly pay?
Which factors affect the choice of a given product or service?
3. To make predictions
• Managers use models to predict the potential outcome in terms of sales or market shares of a
given change in the production process or a given marketing campaign.
• Policy makers can anticipate the changes in economic variables, inflation rates, unemployment
but also in social variables as educational rates in connection with a given policy.
, 2
4. To reduce the complexity of data
How can we measure the level of satisfaction of customers for a product?
Can we identify a few dimensions of the learning process of a child?
Two basic kinds of multivariate statistical models
• Dependency (or “causal”) models
• concerned with causality (the effects of independent variables, interventions, or time on
dependent variables)
• Interdependency or interrelationship models
• concerned with structure: relationships among variables or among cases
Dependency models
• Dependent variable (denoted Y) : a variable that is to be predicted or that changes in response
to an intervention; the value of this variable depends on the values of other variables
• Independent variable (denoted X) :a variable that is used to predict values of another variable,
that records experimental categories such as participant/non-participant
• Confounding variable: a variable that might interfere with or ’confound’ the relationship between
the independent and the dependent variable
A simple dependency model with three independent variables
Interventions
Much management research seeks to discover the effect of an intervention on an outcome.
An intervention is an action taken in order to achieve change, e.g.,
• training is an intervention designed to improve knowledge and skills
• governments often use policy change as interventions to improve social conditions
The intervention is an independent variable (X).
The outcome to be changed (e.g., skills, social conditions) is a dependent variable (Y).
Interdependency model: Correlation
What is the relationship between a country’s
population and the number of Internet users in
the country?
I hypothesise that there is a correlation between
these two variables.
, 3
An overview of the research process
1. Define the problem : the research question.
2. Prepare the research design
3. Identify the sources of data and, if necessary, a sampling plan
4. Collect the data
5. Process and analyse the data
6. Formulate conclusions and prepare the report
Problem definition
The first step consists in the specification of the issue to be addressed:
“What, exactly, do we need or want to know?”
“Why is it important or interesting to know this?”
The issue is then stated in terms of question or hypothesis: the research question.
Common elements of a problem definition
• Research question
• Selection of a theory or other basis for definition of the specific study: a in-depth literature
search has to be run on “recognized sources”.
• Clear identification of the unit of analysis (individuals, couples, countries ecc)
• Clear identification of the fundamental concepts to be studied
• Expected or proposed relationships between the concepts
Research design
The research design describes how you will:
• Model the research problem
• Collect and analyse the data in order to answer the research question
The quality of the model we build depends not only on the research design but also on the data
we gather (or use) to conduct the research.
Model and data collection are intertwined.
For dependency (“causal”) models
• Multiple regression
• Analysis of variance (ANOVA)
For interdependency models
• Factor analysis – relationships among variables
• Cluster analysis – relationships among cases
We often combine interdependency and dependency models, e.g., 1
Step 1:
Use factor analysis to reduce 6 observed variables to 2 factors, attitudes to having a child (X1)
and social influences on childbearing (X2)
Step 2:
Use multiple regression to explain intention to have a child (Y) as a function of X1 and X2:
Y=βο +β1X1 +β2X2 +ε
, 4
Lesson 2
Variables types
Two broad types of variable:
1. Categorical (nonmetric) – takes on non-numerical values, so to describe membership of a
group or category
2. Numerical (metric) – takes on numerical values, providing a numerical measure of size or
extent to which a case has a characteristic:
• Discrete – takes a finite or infinite but countable number of values
• Continuous – can take any value between two given numbers.
Levels of measurements
Categorical variables
• Nominal – non numerical values, cannot be ordered, e.g., sex (male or
female)
national health model
• Ordinal – non numerical values that can be ordered, e.g., age group
(<21, 21-30, >30)
bank size (sml, med, lge)
Numerical (metric) or “scale” variables
• Interval – numerical variables measured on a scale that does not have a
true ‘zero’ point, e.g.
temperature in degrees Celsius scores on a scale of 1 (strongly
disagree) to 7 (strongly agree)
• Ratio – variables measured on a scale that does have a true ‘zero’ point
(a value for which there is absence of the quantity being measured,
e.g., actual age, health expenditure.
Anchored scales
• Particularly in survey research, we often use “anchored scales”, scales in which selected values
are named or “anchored” as a guide to the respondent.
• A fully anchored scale that ranges from strongly disagree to strongly agree is known as a “Likert
scale” ...
• Scales that have anchors other than strongly agree to strongly disagree are just know as n-point
scales
Metric-Interval
Does not have a ‘true 0’ & cannot be used to estimate ratios.
Metric-Ratio
Has a ‘true 0’ and can be used to calculate ratios, e.g.,
Composite scales (some time called summated scales)
Instead of using a single variable to represent a complex concept, responses to a number of
indicator variables are combined into a composite measure.
Composite measures increase validity (we are really measuring what we want to measure!) a
combination of measures provides a more accurate estimate of the true score than a single
measure.
Data quality is measured by:
• Validity
• Reliability
• Accuracy
• Precision
