Quantitative research summary
Lecture 1 (in class lecture)
Quality criteria quantitative research
Internal validity The extent of certainty about the causal relationship
Important:
- eliminating confounding variable
- Cause should precede the e ect
External validity Can we generalize the results? Does it apply to
other similar cases?
Reproducability/ replicability Can we reproduce the study? Is it transparent what
the researcher did?
Hypothesis qualitative research
Testing hypothesis about the possible relationship between two or more variables is often a key
part of quantitative research.
Quantitative research has often research question where 2 or more answers are possible. In
hypothesis testing, these answers are identi ed béfore the data collection.
You can have:
NULL HYPOTHESIS You predict that the independent variable will have
no e ect on the dependent variable.
ALTERNATIVE HYPOTEHSIS You predicts the independent variable will have a
demonstrable e ect on the dependent variable.
The research will enable you to test the question of e ect or no e ect and show if the hypothesis
is true or false.
Positive and negative relationship between variables
Postitive relation If one increases the other increases as well
Negative relation If one increases the other one decreases
Moderator an in uence or variable that a ects the relationship
between two or more factors or variables.
a moderating variable (or moderator) a ects the
strength and direction of the relationship between 2
variables.
example of this
- gender is the moderator (Factor/variable) it in uences the other 2 variables.
- Positive relation & negative relation
fffl ff ffff ff fi fl ff ff
,
The expectation of positive or negative relation are about the direction of the relationship, not
about the starting point or the size of the di erence! So could be both:
Mediating vs confounding variable
Mediating variable A mediating variable (or mediator) explains the
process through which two variables are related.
(It basically explains cause and e ect, see image
below)
Confounding variable Alternative explanation (those that a ect other
variables in a way that produces spurious or
distorted associations between two variables)
di erence mediator vs moderator:
A mediating variable (or mediator) explains the process through which two variables are related,
while a moderating variable (or moderator) a ects the strength and direction of that relationship
Example mediating:
Implicit gender Work-family
Job satisfaction
stereotypes guilt
Mediator variable Dependent variable
Independent variable
ff ff ff ffff
,
Eliminating confounding variable=
important for internal validity
Common designs qualitative research
Cross-sectional You only have 1 point in time➤ you only have a
snapshot of a certain moment.
- often associated with survey study
Longitudinal: Design that takes a longer period of time to actually
get familiar with the situation e.g. ethnography.
2 forms:
1. Observational ➤ researcher does not
manipulate the situation
2. Experimental ➤ researcher does manipulate
the research
Conceptualizing and operationalizing
Conceptualising: Describing/ de ning a concept
fi
,
Operationalization: Making abstract concepts quanti able.
First ask:
Is a concept directly measurable/countable➤ e.g.
- number of students in lecture
- Age
- Etc.
If not use indicators (signals)
- attitudes
- Motivation
- Guilt
Why use multiple indicators:
1. Lower chance of misclassi cation
2. You can measure the entire concept
3. You can make a more detailed distinction
between respondents.
Reliability & validity
- Reliability= is your measurement stable and internally consistent?
- Cronbach’s alpha-> measuring internal consistency, if there’s no consistency in the
outcomes= low cronbach’s alpha. Cronbach's alpha is a measure of the reliability or
consistency of a set of items or questions in a survey or test. It quanti es how closely related
the items are to each other and provides a single numerical value, ranging from 0 to 1, where
higher values indicate higher internal consistency.
- Validity= do you measure what you want to measure?
Cronbach’s alpha
Cronbach’s alpha-> you look at correlation between indicators. High value on cronbach’s alpha
means that indicators are very closely related to each other -> so it measures the same thing.
So e.g. in that case three indicators measure the same thing: guilt.
This increases the reliability as it is consistent. Cronbach’s alpha shows that indicators are
consistent and therefore reliable, however it still doesn’t tell us what we measure
(validity).
Multi-dimensionality
Every dimension has to be measured for a valid measurement.
if there is no agreement on what the concept is how can we have agreement on how to measure
it, what are we even measuring? How does it a ect our research? How can we compare the
ndings of the di erent studies if they use di erent dimensions.
Another problem is multi-dimensionality➤in the dart boards we see, e.g. 4 di erent dimensions, to
get a valid measurement we have to measure all of the dimensions of a concept. Lets say it had 4
dimensions you need to measure all of them to get a valid measurement of the concept.
How to determine validity of a new concept?
Face validity The measure re ects the content of the concept
explored.
Face validity refers to the extent to which an
assessment or measurement appears to be valid
and relevant at rst glance. It is a subjective
judgment made by individuals by simply looking at
the content of the assessment or measurement. In
essence, it is a measure of common sense and
whether the tool "looks like" it measures what it
claims to measure.
fi
fifl ff fi fi ff ff fi ff
Lecture 1 (in class lecture)
Quality criteria quantitative research
Internal validity The extent of certainty about the causal relationship
Important:
- eliminating confounding variable
- Cause should precede the e ect
External validity Can we generalize the results? Does it apply to
other similar cases?
Reproducability/ replicability Can we reproduce the study? Is it transparent what
the researcher did?
Hypothesis qualitative research
Testing hypothesis about the possible relationship between two or more variables is often a key
part of quantitative research.
Quantitative research has often research question where 2 or more answers are possible. In
hypothesis testing, these answers are identi ed béfore the data collection.
You can have:
NULL HYPOTHESIS You predict that the independent variable will have
no e ect on the dependent variable.
ALTERNATIVE HYPOTEHSIS You predicts the independent variable will have a
demonstrable e ect on the dependent variable.
The research will enable you to test the question of e ect or no e ect and show if the hypothesis
is true or false.
Positive and negative relationship between variables
Postitive relation If one increases the other increases as well
Negative relation If one increases the other one decreases
Moderator an in uence or variable that a ects the relationship
between two or more factors or variables.
a moderating variable (or moderator) a ects the
strength and direction of the relationship between 2
variables.
example of this
- gender is the moderator (Factor/variable) it in uences the other 2 variables.
- Positive relation & negative relation
fffl ff ffff ff fi fl ff ff
,
The expectation of positive or negative relation are about the direction of the relationship, not
about the starting point or the size of the di erence! So could be both:
Mediating vs confounding variable
Mediating variable A mediating variable (or mediator) explains the
process through which two variables are related.
(It basically explains cause and e ect, see image
below)
Confounding variable Alternative explanation (those that a ect other
variables in a way that produces spurious or
distorted associations between two variables)
di erence mediator vs moderator:
A mediating variable (or mediator) explains the process through which two variables are related,
while a moderating variable (or moderator) a ects the strength and direction of that relationship
Example mediating:
Implicit gender Work-family
Job satisfaction
stereotypes guilt
Mediator variable Dependent variable
Independent variable
ff ff ff ffff
,
Eliminating confounding variable=
important for internal validity
Common designs qualitative research
Cross-sectional You only have 1 point in time➤ you only have a
snapshot of a certain moment.
- often associated with survey study
Longitudinal: Design that takes a longer period of time to actually
get familiar with the situation e.g. ethnography.
2 forms:
1. Observational ➤ researcher does not
manipulate the situation
2. Experimental ➤ researcher does manipulate
the research
Conceptualizing and operationalizing
Conceptualising: Describing/ de ning a concept
fi
,
Operationalization: Making abstract concepts quanti able.
First ask:
Is a concept directly measurable/countable➤ e.g.
- number of students in lecture
- Age
- Etc.
If not use indicators (signals)
- attitudes
- Motivation
- Guilt
Why use multiple indicators:
1. Lower chance of misclassi cation
2. You can measure the entire concept
3. You can make a more detailed distinction
between respondents.
Reliability & validity
- Reliability= is your measurement stable and internally consistent?
- Cronbach’s alpha-> measuring internal consistency, if there’s no consistency in the
outcomes= low cronbach’s alpha. Cronbach's alpha is a measure of the reliability or
consistency of a set of items or questions in a survey or test. It quanti es how closely related
the items are to each other and provides a single numerical value, ranging from 0 to 1, where
higher values indicate higher internal consistency.
- Validity= do you measure what you want to measure?
Cronbach’s alpha
Cronbach’s alpha-> you look at correlation between indicators. High value on cronbach’s alpha
means that indicators are very closely related to each other -> so it measures the same thing.
So e.g. in that case three indicators measure the same thing: guilt.
This increases the reliability as it is consistent. Cronbach’s alpha shows that indicators are
consistent and therefore reliable, however it still doesn’t tell us what we measure
(validity).
Multi-dimensionality
Every dimension has to be measured for a valid measurement.
if there is no agreement on what the concept is how can we have agreement on how to measure
it, what are we even measuring? How does it a ect our research? How can we compare the
ndings of the di erent studies if they use di erent dimensions.
Another problem is multi-dimensionality➤in the dart boards we see, e.g. 4 di erent dimensions, to
get a valid measurement we have to measure all of the dimensions of a concept. Lets say it had 4
dimensions you need to measure all of them to get a valid measurement of the concept.
How to determine validity of a new concept?
Face validity The measure re ects the content of the concept
explored.
Face validity refers to the extent to which an
assessment or measurement appears to be valid
and relevant at rst glance. It is a subjective
judgment made by individuals by simply looking at
the content of the assessment or measurement. In
essence, it is a measure of common sense and
whether the tool "looks like" it measures what it
claims to measure.
fi
fifl ff fi fi ff ff fi ff
