Quantitative research methodology lectures
Lecture 1
Explanatory research: the why or causes of social phenomena
Cause-and-effect relationships, or causal relations, are often represented in a causal model
We examine the relationships in a causal model in a deductive manner:
1. By having derived hypotheses from the theory
2. By statistically testing these on real live data
The relationship between reading list and reading frequency could be caused by a third variable, that
affects both reading list and reading frequency. This is called a confounding variable that could be the
cause of the existing relationship and not so much the reading list
It could be that the relationship disappears. In this case the relationship is sometimes called a
spurious relation
That is when there is a relation, but it is not causal. The size of the reading list was not a cause of how
much reading there was
If the relationships persists, you have stronger arguments for your conclusion that there is a causal
relationship, now that this important confounding variable has been controlled
,The positive relation between reading list and reading frequency disappears and even turns into a
very weak negative relations; it was not causal
It was fully explained by a variable that was in causal order before both reading list and reading
frequency. Because secondary schooling is in causal order before both reading list and reading
frequency, it is cleared to draw the model like this:
Correlation is not the same as causation
Correlation: changes in one variable correlate with changes in another
Causation: change in one variable changes in the other
With causation you make very strong claims, so you have to be able to justify that well
You do that by taking into account confounding variable by statistically controlling them
For some variables it is hard to think of confounding variables. For sex, age, parental education,
personality you have strong arguments that there are no confounders
,Relations can be:
Positive:
- The more books on the list, the more one reads later in life
- The fewer books on the list, the fewer books one reads later in life
Negative:
- The more books on the list, the fewer one reads later in life
- The fewer books on the list, the more one reads later in life
Absent:
- The number of books on the list has nothing to do with the amount of reading later on
With the covariance you can calculate the correlation between x and y
Covariance: the average extent to which deviations of one variable from the mean go hand in hand
with deviations of another variable from the mean: this can be positive or negative
You cannot say whether the covariance indicates a strong relationship – it all depends on the units of
measurements
Disadvantage covariance: if you use other measurement units, the size of the covariance also changes
, We cannot use the covariance as a measure of the strength of the relationship between two
variables. But it is what the correlation is based on
To get a measure of the strength of the correlation between 2 variables, we convers the covariance
into standard units (standardization)
Standard units = standard deviations
If we divide each deviation from the mean by the standard deviation, we get the distance to the
mean in standard deviations
We express the distance to the distance to the average in standard deviations = units of standard
deviation = we use z-scores
Lecture 1
Explanatory research: the why or causes of social phenomena
Cause-and-effect relationships, or causal relations, are often represented in a causal model
We examine the relationships in a causal model in a deductive manner:
1. By having derived hypotheses from the theory
2. By statistically testing these on real live data
The relationship between reading list and reading frequency could be caused by a third variable, that
affects both reading list and reading frequency. This is called a confounding variable that could be the
cause of the existing relationship and not so much the reading list
It could be that the relationship disappears. In this case the relationship is sometimes called a
spurious relation
That is when there is a relation, but it is not causal. The size of the reading list was not a cause of how
much reading there was
If the relationships persists, you have stronger arguments for your conclusion that there is a causal
relationship, now that this important confounding variable has been controlled
,The positive relation between reading list and reading frequency disappears and even turns into a
very weak negative relations; it was not causal
It was fully explained by a variable that was in causal order before both reading list and reading
frequency. Because secondary schooling is in causal order before both reading list and reading
frequency, it is cleared to draw the model like this:
Correlation is not the same as causation
Correlation: changes in one variable correlate with changes in another
Causation: change in one variable changes in the other
With causation you make very strong claims, so you have to be able to justify that well
You do that by taking into account confounding variable by statistically controlling them
For some variables it is hard to think of confounding variables. For sex, age, parental education,
personality you have strong arguments that there are no confounders
,Relations can be:
Positive:
- The more books on the list, the more one reads later in life
- The fewer books on the list, the fewer books one reads later in life
Negative:
- The more books on the list, the fewer one reads later in life
- The fewer books on the list, the more one reads later in life
Absent:
- The number of books on the list has nothing to do with the amount of reading later on
With the covariance you can calculate the correlation between x and y
Covariance: the average extent to which deviations of one variable from the mean go hand in hand
with deviations of another variable from the mean: this can be positive or negative
You cannot say whether the covariance indicates a strong relationship – it all depends on the units of
measurements
Disadvantage covariance: if you use other measurement units, the size of the covariance also changes
, We cannot use the covariance as a measure of the strength of the relationship between two
variables. But it is what the correlation is based on
To get a measure of the strength of the correlation between 2 variables, we convers the covariance
into standard units (standardization)
Standard units = standard deviations
If we divide each deviation from the mean by the standard deviation, we get the distance to the
mean in standard deviations
We express the distance to the distance to the average in standard deviations = units of standard
deviation = we use z-scores
