Notes on video’s per week – Management Methods for Research 1
Week 1
Video 1 – introduction
Introduction of the term data, how to use data, plotting and how to analyze.
Video 2 – data
Data are pieces of information collected for analysis or reference. Different techniques are needed to
analyze different kinds of data. Data has a fixed structure, so you need to distinguish between
variables and units.
- Variables; it consists of a number of properties. For example, each column represents one
variable. What exactly is measured.
- Units; measured from a set of things/people/ect. For example, each row represents one unit.
You can have a name as variable, and compare this amongst people (units). You can measure
several variables for one unit.
Levels of measurement:
- Categorial measurements
o Binary variable; you have 2 outcomes (dead
OR alive). Circa 300 observations are needed
to make plausible estimates.
o Nominal variable; you have multiple given
outcomes, no order (omnivore OR vegetarian
OR vegan). Also, hair color, names, etc.
o Ordinal variable; ranking the outcomes (bad
OR intermediate OR good). Clothing size, etc.
You need circa 30-300 observations to make
good estimates for the outcomes.
- Numerical measurements
o Discrete data; a real number that can’t be fractional, so a round number with nothing
in between (13 people passed the course)
o Continuous data; entities get a distinct score, there can be something in between
(temperature, length)
*the more you go from categorical to numerical continuous data, you obtain more precise
information about the unit. You need less data to estimate outcomes then. Therefore, its important
,to measure with a higher level of measurement, when possible while data cannot be retrieved or
reversed. The lower the amount of information in your data, the larger your sample needs to be.
You need to collect the right data for the right conclusions. In quantitative research, you need to
motivate and document the way you collected the data. Is the sample representative? Is the data
valid? Is there measurement error?
1. Representativeness of the sample; statistics only gives conclusions about the population you
have samples from.
o What is the population?
o How to make my sample representative for that population?
Ideally, we use random sampling. Assign numbers to all units in the population and let a computer
draw 30 random numbers. You need to include these observations in your sample.
2. Validity of the data; do the data reflect what is should reflect? And can they be used to
answer the research question?
You need to take a ‘face validity check’ for errors and mistakes. Definitions should be clear to
interpret the data. Also, check if people involved know the measurement procedure. You need to
make clear what kind of errors occurred, what are the definitions of the data, etc. to avoid mistakes.
You should motivate your argument, do not delete your original input. Internal validity is about if you
measured the element you said you will measure. External validity is about whether or not you can
generalize the outcome to other cases.
3. Measurement error; discrepance between the actual value we are trying to measure, and the
number we use to represent that value. There are 2 types of errors:
a. Systematic measurement error; difference between the average result and the true
value. The solution is calibration of the outcome if you have your reference point
(true value)
b. Random measurement error; unsystematic deviations due to imprecision of the
measurement system. This happens for example when you use different values,
equipment, people, systems, etc.
You need to ask yourself the following questions:
Which method has the largest bias? Which one has the largest measurement spread? Which method
do you prefer and why?
When collecting data, the measurement error cannot be large.
Video 3 – data analysis
You have to describe data in a few numbers, these are locations (center of data):
- Median; the middle score when the data is ordered. 50% of the numbers is below the
median and 50% above the median. It’s not influenced by outliers, and it doesn’t say
anything about minimum or maximum
- Mean; the sum of the data divided by the total amount of data. This is noted as S = sum.
Outliers have a great effect on the mean/average.
- Mode; the most frequent number. You are dealing with different groups. This is usually a
problem, so you have to split the population into homogeneous groups.
, o Bimodal; having 2 modes
o Multimodal; having several modes
The data can be diversly spread, this is the
dispersion. How far apart are the
measurements?
- Range; the smallest value
subtracted from the largest value,
maximum – minimum. Therefore,
this is very sensitive to outliers.
This takes into account 100% of the data.
- Interquartile range; the range of the
middle 50% of the data. This is less
sensitive to outliers. You take the
median, which you again divide by 2.
You now have 4 quarters, with each a
median (lower quartile, median, upper
quartile). The interquartile range is the difference between upper and lower quartile. This
takes into account the middle 50% of the data.
- Variance; the average squared distance between each point and the mean of the data. This
number is hard to interpret, therefore
you look at the standard deviation of
it.
- Standard deviation; the square root of the variance. This number is in the same language as
the original data. A high standard deviation means that there is a lot of spread of the data, so
no spikes
- Confidence interval; an indication of how certain you are of your averages with different
samples. What kind of range is expected? You have the measured average (X) and an
expected population average (u). The more spread you have (high sd), the less certain you
can be about the confidence interval. Then the 95% confidence interval will be lower.
- Skewness; asymmetry of the distribution
o Positive skew (maximum is located left); score
bunched at low values with the tail pointing to
high values
o Negative skew (maximum is located right);
scores bunched at high values with the tail
pointing to low values
You have to plot your data to tell the story. This depends on whether you have: 1/2 variables to
display and categorical/numerical data.
Week 1
Video 1 – introduction
Introduction of the term data, how to use data, plotting and how to analyze.
Video 2 – data
Data are pieces of information collected for analysis or reference. Different techniques are needed to
analyze different kinds of data. Data has a fixed structure, so you need to distinguish between
variables and units.
- Variables; it consists of a number of properties. For example, each column represents one
variable. What exactly is measured.
- Units; measured from a set of things/people/ect. For example, each row represents one unit.
You can have a name as variable, and compare this amongst people (units). You can measure
several variables for one unit.
Levels of measurement:
- Categorial measurements
o Binary variable; you have 2 outcomes (dead
OR alive). Circa 300 observations are needed
to make plausible estimates.
o Nominal variable; you have multiple given
outcomes, no order (omnivore OR vegetarian
OR vegan). Also, hair color, names, etc.
o Ordinal variable; ranking the outcomes (bad
OR intermediate OR good). Clothing size, etc.
You need circa 30-300 observations to make
good estimates for the outcomes.
- Numerical measurements
o Discrete data; a real number that can’t be fractional, so a round number with nothing
in between (13 people passed the course)
o Continuous data; entities get a distinct score, there can be something in between
(temperature, length)
*the more you go from categorical to numerical continuous data, you obtain more precise
information about the unit. You need less data to estimate outcomes then. Therefore, its important
,to measure with a higher level of measurement, when possible while data cannot be retrieved or
reversed. The lower the amount of information in your data, the larger your sample needs to be.
You need to collect the right data for the right conclusions. In quantitative research, you need to
motivate and document the way you collected the data. Is the sample representative? Is the data
valid? Is there measurement error?
1. Representativeness of the sample; statistics only gives conclusions about the population you
have samples from.
o What is the population?
o How to make my sample representative for that population?
Ideally, we use random sampling. Assign numbers to all units in the population and let a computer
draw 30 random numbers. You need to include these observations in your sample.
2. Validity of the data; do the data reflect what is should reflect? And can they be used to
answer the research question?
You need to take a ‘face validity check’ for errors and mistakes. Definitions should be clear to
interpret the data. Also, check if people involved know the measurement procedure. You need to
make clear what kind of errors occurred, what are the definitions of the data, etc. to avoid mistakes.
You should motivate your argument, do not delete your original input. Internal validity is about if you
measured the element you said you will measure. External validity is about whether or not you can
generalize the outcome to other cases.
3. Measurement error; discrepance between the actual value we are trying to measure, and the
number we use to represent that value. There are 2 types of errors:
a. Systematic measurement error; difference between the average result and the true
value. The solution is calibration of the outcome if you have your reference point
(true value)
b. Random measurement error; unsystematic deviations due to imprecision of the
measurement system. This happens for example when you use different values,
equipment, people, systems, etc.
You need to ask yourself the following questions:
Which method has the largest bias? Which one has the largest measurement spread? Which method
do you prefer and why?
When collecting data, the measurement error cannot be large.
Video 3 – data analysis
You have to describe data in a few numbers, these are locations (center of data):
- Median; the middle score when the data is ordered. 50% of the numbers is below the
median and 50% above the median. It’s not influenced by outliers, and it doesn’t say
anything about minimum or maximum
- Mean; the sum of the data divided by the total amount of data. This is noted as S = sum.
Outliers have a great effect on the mean/average.
- Mode; the most frequent number. You are dealing with different groups. This is usually a
problem, so you have to split the population into homogeneous groups.
, o Bimodal; having 2 modes
o Multimodal; having several modes
The data can be diversly spread, this is the
dispersion. How far apart are the
measurements?
- Range; the smallest value
subtracted from the largest value,
maximum – minimum. Therefore,
this is very sensitive to outliers.
This takes into account 100% of the data.
- Interquartile range; the range of the
middle 50% of the data. This is less
sensitive to outliers. You take the
median, which you again divide by 2.
You now have 4 quarters, with each a
median (lower quartile, median, upper
quartile). The interquartile range is the difference between upper and lower quartile. This
takes into account the middle 50% of the data.
- Variance; the average squared distance between each point and the mean of the data. This
number is hard to interpret, therefore
you look at the standard deviation of
it.
- Standard deviation; the square root of the variance. This number is in the same language as
the original data. A high standard deviation means that there is a lot of spread of the data, so
no spikes
- Confidence interval; an indication of how certain you are of your averages with different
samples. What kind of range is expected? You have the measured average (X) and an
expected population average (u). The more spread you have (high sd), the less certain you
can be about the confidence interval. Then the 95% confidence interval will be lower.
- Skewness; asymmetry of the distribution
o Positive skew (maximum is located left); score
bunched at low values with the tail pointing to
high values
o Negative skew (maximum is located right);
scores bunched at high values with the tail
pointing to low values
You have to plot your data to tell the story. This depends on whether you have: 1/2 variables to
display and categorical/numerical data.
