Empirical Research Project – Week 1 – Data and Measurement
What does the concept of measurement reflect?
The relationship between numbers and what is being measured.
Which measurement levels exist?
- Nominal (i.e. categorical; e.g. colors).
- Ordinal (i.e. logically ordered categories; e.g. ‘1st’, ‘2nd’, and ‘3rd’).
- Interval (i.e. information about differences between points on a scale; e.g. Celsius scale).
- Ratio (i.e. information about differences between points on a scale, but with an absolute zero;
e.g. weight).
What does the concept of validity reflect?
The extent to which a measure correctly represents the concept of study (i.e. ‘can you trust the results
of the study?’).
What does the concept of external validity reflect?
‘Generalizability of results to other situations’ or ‘do the results of the study reflect what would
happen in the real world?’. Naturalistic observations—in which participants don’t know that they’re
being observed—have the highest external validity: i.e. observing people in the bathroom to
determine whether they wash their hands without them knowing that they’re being observed.
What does the concept of internal validity reflect?
‘How well the study was done’ or ‘the extent to which researchers are sure about what caused the
results of the study’. Experiments—with control and randomization—have the highest internal
validity: i.e. whether people with more knowledge and awareness about germs causes people to wash
their hands more.
What is the trade-off between internal and external validity?
Increasing internal validity requires controlling of the environment—i.e. in a lab setting—but this
causes lower external validity. Increasing external validity requires a natural environment, but this
lack of control causes low internal validity.
What does the concept of accuracy reflect?
Accuracy reflects the extend to which the measured value is close to the actual value.
What does the concept of reliability reflect?
Reliability reflects the extend to which a variable is consistent in what it is intended to measure.
When researching a concept, which dimensions should a researcher take into account?
- The type of data source
o Primary data – data collected/generated by the researcher; e.g. survey data. Primary
data has lower external validity, is subject to personal prejudice, and is more time
consuming and costly than secondary data.
o Secondary data – data collected/generated by other agencies; e.g. financial statement
data. Secondary data has higher external validity, less personal prejudice, and is a
cheaper alternative to primary data.
, - The type of measure – there are different ways to measure the same thing, but which one will
be measured in the study; e.g. financial performance, operational performance, or reputational
performance?
- The level of analysis – e.g. firm-level and country-level.
Differences in data sources, types of measure, and levels of analysis result in what?
They result in differences in results and conclusions. They also affect one another. That is, the data
source (partially) determines the types of measure and the level of analysis, and vice versa.
On which factors depend choices regarding data source, types of measure, and level of analysis?
- The theoretical framework
- The research question
- The hypotheses
- Data availability
What is selection bias?
Selection bias occurs when a sample is restricted to observations that reflect a particular
characteristic, which implies that the sample is not representative for the population.
What types of selection bias exist?
- Resource bias (i.e. choosing the cheapest option)
- Convenience bias (i.e. choosing the easiest accessible option)
- Popularity bias
- Data source bias
- Knowledge about the topic
When selection bias occurs, what should a researcher do?
The researcher should (1) be careful when interpreting results and (2) be careful when drawing
conclusions.
Which statistical tests can be used to check the representativity of a sample?
- The t-test – A rough approach that tests ‘sample v. population including sample’ (e.g. 100 v.
1000)
- The chi-square test – A detailed approach that tests ‘in-sample v. out-sample’ (e.g. 100 v. 900)
, Empirical Research Project – Week 2 – Data and Descriptive Statistics
What are the main objectives of providing descriptive or summary statistics?
It is basically the first step of the analysis and it functions to introduce the players of the game (i.e.
variables) and get a feel of the data.
What are some examples of summary statistics?
- The number of observations
- The measures of central tendency
- Skewness
- Kurtosis
- Minimum and maximum
- Variance and standard deviation
Which measures of central tendency exist?
- The mean – Influenced by large observations
- The median – The middle point when values are ranked in order of magnitude; is relatively
unaffected by large observations
- The mode – The most frequent value
Which measure of central tendency should be used for which measurement level?
- Nominal – Mode
- Ordinal – Median
- Interval or ratio (not skewed) – Mean
- Interval or ratio (skewed) – Median
What does the concept of skewness reflect?
The shape of the distribution compared to a normal distribution.
When the mean is larger than the median, the data is?
Positively skewed and the tail is displayed on the right side of the bell curve.
What does the concept of kurtosis reflect?
The shape of the distribution compared to a normal distribution.
Leptokurtic implies what?
Kurtosis is larger than 3. Heavy tails are involved, and the figure could be pointy.
Platykurtic implies what?
Kurtosis is smaller than 3. Light tails are involved, and the figure is rather flat.
The minimum and maximum are measures of what?
The variability of the data.
How can we calculate the range of data?
By subtracting the minimum value from the maximum value.
The variance and standard deviation are measures of what?
The variability of the data and how spread out the data are from the mean.
What does the concept of measurement reflect?
The relationship between numbers and what is being measured.
Which measurement levels exist?
- Nominal (i.e. categorical; e.g. colors).
- Ordinal (i.e. logically ordered categories; e.g. ‘1st’, ‘2nd’, and ‘3rd’).
- Interval (i.e. information about differences between points on a scale; e.g. Celsius scale).
- Ratio (i.e. information about differences between points on a scale, but with an absolute zero;
e.g. weight).
What does the concept of validity reflect?
The extent to which a measure correctly represents the concept of study (i.e. ‘can you trust the results
of the study?’).
What does the concept of external validity reflect?
‘Generalizability of results to other situations’ or ‘do the results of the study reflect what would
happen in the real world?’. Naturalistic observations—in which participants don’t know that they’re
being observed—have the highest external validity: i.e. observing people in the bathroom to
determine whether they wash their hands without them knowing that they’re being observed.
What does the concept of internal validity reflect?
‘How well the study was done’ or ‘the extent to which researchers are sure about what caused the
results of the study’. Experiments—with control and randomization—have the highest internal
validity: i.e. whether people with more knowledge and awareness about germs causes people to wash
their hands more.
What is the trade-off between internal and external validity?
Increasing internal validity requires controlling of the environment—i.e. in a lab setting—but this
causes lower external validity. Increasing external validity requires a natural environment, but this
lack of control causes low internal validity.
What does the concept of accuracy reflect?
Accuracy reflects the extend to which the measured value is close to the actual value.
What does the concept of reliability reflect?
Reliability reflects the extend to which a variable is consistent in what it is intended to measure.
When researching a concept, which dimensions should a researcher take into account?
- The type of data source
o Primary data – data collected/generated by the researcher; e.g. survey data. Primary
data has lower external validity, is subject to personal prejudice, and is more time
consuming and costly than secondary data.
o Secondary data – data collected/generated by other agencies; e.g. financial statement
data. Secondary data has higher external validity, less personal prejudice, and is a
cheaper alternative to primary data.
, - The type of measure – there are different ways to measure the same thing, but which one will
be measured in the study; e.g. financial performance, operational performance, or reputational
performance?
- The level of analysis – e.g. firm-level and country-level.
Differences in data sources, types of measure, and levels of analysis result in what?
They result in differences in results and conclusions. They also affect one another. That is, the data
source (partially) determines the types of measure and the level of analysis, and vice versa.
On which factors depend choices regarding data source, types of measure, and level of analysis?
- The theoretical framework
- The research question
- The hypotheses
- Data availability
What is selection bias?
Selection bias occurs when a sample is restricted to observations that reflect a particular
characteristic, which implies that the sample is not representative for the population.
What types of selection bias exist?
- Resource bias (i.e. choosing the cheapest option)
- Convenience bias (i.e. choosing the easiest accessible option)
- Popularity bias
- Data source bias
- Knowledge about the topic
When selection bias occurs, what should a researcher do?
The researcher should (1) be careful when interpreting results and (2) be careful when drawing
conclusions.
Which statistical tests can be used to check the representativity of a sample?
- The t-test – A rough approach that tests ‘sample v. population including sample’ (e.g. 100 v.
1000)
- The chi-square test – A detailed approach that tests ‘in-sample v. out-sample’ (e.g. 100 v. 900)
, Empirical Research Project – Week 2 – Data and Descriptive Statistics
What are the main objectives of providing descriptive or summary statistics?
It is basically the first step of the analysis and it functions to introduce the players of the game (i.e.
variables) and get a feel of the data.
What are some examples of summary statistics?
- The number of observations
- The measures of central tendency
- Skewness
- Kurtosis
- Minimum and maximum
- Variance and standard deviation
Which measures of central tendency exist?
- The mean – Influenced by large observations
- The median – The middle point when values are ranked in order of magnitude; is relatively
unaffected by large observations
- The mode – The most frequent value
Which measure of central tendency should be used for which measurement level?
- Nominal – Mode
- Ordinal – Median
- Interval or ratio (not skewed) – Mean
- Interval or ratio (skewed) – Median
What does the concept of skewness reflect?
The shape of the distribution compared to a normal distribution.
When the mean is larger than the median, the data is?
Positively skewed and the tail is displayed on the right side of the bell curve.
What does the concept of kurtosis reflect?
The shape of the distribution compared to a normal distribution.
Leptokurtic implies what?
Kurtosis is larger than 3. Heavy tails are involved, and the figure could be pointy.
Platykurtic implies what?
Kurtosis is smaller than 3. Light tails are involved, and the figure is rather flat.
The minimum and maximum are measures of what?
The variability of the data.
How can we calculate the range of data?
By subtracting the minimum value from the maximum value.
The variance and standard deviation are measures of what?
The variability of the data and how spread out the data are from the mean.
