Lecture 1
Model of argumentation (Stephen Toulmin) with three steps:
1. Ground: information, statistical output
2. Claim: statistical decision, choice for a technique
3. Warrant: general rules, statistical principles
Four steps in statistics:
1. Research question: descriptive/explanatory, static/dynamic
2. Variables: nominal/ordinal/interval/ratio
3. Relationship: non (univariate)/correlation/causal relation
4. Criteria analysis: measurement level/shape distribution
Three central tendency measures:
1. Mode (N/O/I/R)
2. Median (O/I/R)
3. Mean (I/R)
Three dispersion measures:
1. Interquartile range (O/I/R)
2. Variance (I/R) (sum of all differences squared)
3. Standard Deviation (I/R)
Normal distribution: mean = median
Positively skewed: mean > median
Negatively skewed: mean < median
Z-score: relative position (how far away from the mean, how many SD)
Empirical rule (normal distribution)
68 % lies between one standard deviation from the mean
95 % lies between two standard deviations from the mean
99,7 % lies between three standard deviations from the mean
Chebyshev’s rule (skewed distribution)
75 % lies between two standard deviations from the mean
88,9 % lies between three standard deviations from the mean
Lecture 2:
Probability: the chance that a situation occurs in the real world
Sampling distribution: all possible samples in the total population
Sample size: number of respondents in your sample (related to Central Limited Theorem)
Sample mean: means of every sample you draw, you can also have a mean of those means or a
standard deviation of those means (standard error of the mean)
Confidence interval: the probability that the random selected variable encloses the unknown parameter
with 90, 95 or 99 percent confidence (alpha is 10, 5 or 1 percent)
Null hypothesis: status quo, no effect or difference (=, ≤ or ≥)
Alternative hypothesis: difference, correlation, change (≠, < or >)
One tailed: increase or decrease
Two tailed: you don’t have a clue which direction (divide alpha by two)
Type I error: H0 is true but we rejected it
Type II error: HA is true, but we didn’t reject H0
Model of argumentation (Stephen Toulmin) with three steps:
1. Ground: information, statistical output
2. Claim: statistical decision, choice for a technique
3. Warrant: general rules, statistical principles
Four steps in statistics:
1. Research question: descriptive/explanatory, static/dynamic
2. Variables: nominal/ordinal/interval/ratio
3. Relationship: non (univariate)/correlation/causal relation
4. Criteria analysis: measurement level/shape distribution
Three central tendency measures:
1. Mode (N/O/I/R)
2. Median (O/I/R)
3. Mean (I/R)
Three dispersion measures:
1. Interquartile range (O/I/R)
2. Variance (I/R) (sum of all differences squared)
3. Standard Deviation (I/R)
Normal distribution: mean = median
Positively skewed: mean > median
Negatively skewed: mean < median
Z-score: relative position (how far away from the mean, how many SD)
Empirical rule (normal distribution)
68 % lies between one standard deviation from the mean
95 % lies between two standard deviations from the mean
99,7 % lies between three standard deviations from the mean
Chebyshev’s rule (skewed distribution)
75 % lies between two standard deviations from the mean
88,9 % lies between three standard deviations from the mean
Lecture 2:
Probability: the chance that a situation occurs in the real world
Sampling distribution: all possible samples in the total population
Sample size: number of respondents in your sample (related to Central Limited Theorem)
Sample mean: means of every sample you draw, you can also have a mean of those means or a
standard deviation of those means (standard error of the mean)
Confidence interval: the probability that the random selected variable encloses the unknown parameter
with 90, 95 or 99 percent confidence (alpha is 10, 5 or 1 percent)
Null hypothesis: status quo, no effect or difference (=, ≤ or ≥)
Alternative hypothesis: difference, correlation, change (≠, < or >)
One tailed: increase or decrease
Two tailed: you don’t have a clue which direction (divide alpha by two)
Type I error: H0 is true but we rejected it
Type II error: HA is true, but we didn’t reject H0
