Statistical Concepts 1: Variables
Think of variables as questions
and values as answers.
e.g. Two groups of
participants (man and women). CCTV cameras should be used
on public transport (Strongly agree, Agree, Neutral,
Disagree, Strongly Disagree)
If all the values are the
same then that variable must be discarded. (e.g. Are you
willing to participate in this experiment?)
Independent Variables aka Factors
- Fixed factors - manipulated (varied) by the experimenter.
The different values the factor can take are known as values,
levels or conditions.
- Subject variables - recorded by the experimenter, e.g.
- Random factors - levels can't be repeated exactly, e.g.
animal litter, household aka Randomised Blocks.
- What you measure, e.g. score, reaction time, number of
errors, answer to a question.
Levels of Measurement
- Nominal aka Categorical, e.g. colour, political
- Ordinal, e.g. Likert scales, ice dancing
scores (subjective scales). [Nominal variables with 2 values]
- Interval aka Scale, e.g. Temperature Celcius,
- Ratio, e.g. reaction time, number correct.
A subset of Interval.
- Flat, e.g. die
- Binomial, e.g. multiple coins
- Bell curve, e.g. sum of several dice
- Poisson, e.g waiting for a rare event to happen
- Central tendancy - mean, median, mode
- Variation - range, interquartile range, standard deviation
- Skewness - how lop-sided is it? Normal distribution has a skewness
of 0. Significantly skewed if |skewness| > 2 * SE skewness
- Kurtosis - how pointy is it? Normal distribution has a kurtosis of
0. Significantly skewed if |kurtosis| > 2 * SE kurtosis