Pre-Master
Business Administration
2019-2020
Management Research Methods I
, Levels of measurement
- Categorical variables
à Binary (two outcomes) (Representative at 300 observations)
à Nominal (multiple variables, no order)
à Ordinal (multiple variables that can be ordered) (Representative at 30-300
observations)
- Numerical variables (Representative at 30 observations)
à Discrete (counts, number of defects)
à Continuous (entities get a distinct store)
3 questions when collecting data
- Is the sample representative? (Generalizable from a sample to a population)
- Is the data valid? (Does the data reflect what is should?)
- Is there measurement error?
à Systematic measurement error; difference between average
measurement result and true value. (Non-digital bathroom scale or bias)
à Random measurement error: unsystematic deviations due to
measurement imprecision. (Two observers)
- When defining the difference between random and systematic measurement
error think of a dartboard. If somebody is constantly throwing in the same area
apart from the bull, that would be systematic measurement error (same with
constantly giving yourself a higher grade in comparison to the true value,
however the differences between those grades so let’s say the average is 7, and
somebody gives himself a 6, another one an 8, would be random.). If the darts
would be all around that’s random.
Describing data
- Location
à Median: the middle score when data is ordered.
à Mean: the sum of data divided by the amount of data.
- Dispersion
à Range: the smallest value subtracted by the largest – very sensitive to outliers.
à Interquartile range: the range of the middle 50% of the data.
à Variance: the average squared distance between each point and the mean of
the data. We square the value in order to get rid of the minus. So when we talk
about -5 as difference compared to the mean it is squared to 25 to get only
positive numbers.
à Standard deviation: the square root of the variance.
- Other properties
à Confidence interval: 95% of the observations will have a result that is
included in a small range (the confidence interval). Within this range, you can
say that with 95% certainty the population mean will fall. When increasing the
certainty from 95% to 99% the interval will be wider. If you want more certainty
you need to allow for more options. If you increase sample size the interval will
be more narrow, as you are decreasing the amount of variance. Make sure to
remember that the CI does not consider all data, but just the parameter Mean.
Business Administration
2019-2020
Management Research Methods I
, Levels of measurement
- Categorical variables
à Binary (two outcomes) (Representative at 300 observations)
à Nominal (multiple variables, no order)
à Ordinal (multiple variables that can be ordered) (Representative at 30-300
observations)
- Numerical variables (Representative at 30 observations)
à Discrete (counts, number of defects)
à Continuous (entities get a distinct store)
3 questions when collecting data
- Is the sample representative? (Generalizable from a sample to a population)
- Is the data valid? (Does the data reflect what is should?)
- Is there measurement error?
à Systematic measurement error; difference between average
measurement result and true value. (Non-digital bathroom scale or bias)
à Random measurement error: unsystematic deviations due to
measurement imprecision. (Two observers)
- When defining the difference between random and systematic measurement
error think of a dartboard. If somebody is constantly throwing in the same area
apart from the bull, that would be systematic measurement error (same with
constantly giving yourself a higher grade in comparison to the true value,
however the differences between those grades so let’s say the average is 7, and
somebody gives himself a 6, another one an 8, would be random.). If the darts
would be all around that’s random.
Describing data
- Location
à Median: the middle score when data is ordered.
à Mean: the sum of data divided by the amount of data.
- Dispersion
à Range: the smallest value subtracted by the largest – very sensitive to outliers.
à Interquartile range: the range of the middle 50% of the data.
à Variance: the average squared distance between each point and the mean of
the data. We square the value in order to get rid of the minus. So when we talk
about -5 as difference compared to the mean it is squared to 25 to get only
positive numbers.
à Standard deviation: the square root of the variance.
- Other properties
à Confidence interval: 95% of the observations will have a result that is
included in a small range (the confidence interval). Within this range, you can
say that with 95% certainty the population mean will fall. When increasing the
certainty from 95% to 99% the interval will be wider. If you want more certainty
you need to allow for more options. If you increase sample size the interval will
be more narrow, as you are decreasing the amount of variance. Make sure to
remember that the CI does not consider all data, but just the parameter Mean.
