Describing Data Flashcards
Inferential statistics
Use descriptive statistics to make assumptions about the populations.
Categorical variables
IV- categorical.
Usually group people to compare.
Have to be distinct groups.
Types of data
Categorical variable- nominal data. (Measured in quality)
Measured variable- interval or ordinal data. (Measured in quantity)
Nominal data
Discrete categories.
Can’t be in more than one.
No particular order.
Arbitrary levels.
Ordinal data
Ordered by rank.
Differences between values aren’t important.
Interval data
Intervals between each point are equal.
Constant scale.
Shared understanding of scale.
No natural zero.
Measurements
Numerical scores don’t give insight into why scores are different/similar.
Qualitative/contextual data explains more than just numbers.
Mean
Total divided by number.
Use when scores are grouped around central value.
Not if scores are unevenly distributed; cluster around outliers.
Median
Central value in order.
For even number, mean of two central scores.
Mode
Most commonly occurring.
May be more than one.
May be none.
Two modes- bimodal data set.
Which to use?
Nominal- mode.
Ordinal- mode/median.
Interval- mode/median/mean.
Variability or dispersion
Knowledge of spread/dispersion + measure of central tendency would give useful description of data set.
Range- highest - lowest.
Variance- distance of scores from the mean.
Standard deviation- distance of each score on average from the mean.
Descriptive statistics
Measure aspects of target group.
Summarise measurements.
Identify patterns.
Measured variables
DV- continuous/measured.
Commonly recognised scale.
Some continuum which people will score against (eg distance, time).
Range
Smallest from largest, plus 1.
Only use when scores are clustered together.
Most useful measure with nominal or ordinal data.
Extreme values have a disproportionate effect.
Interquartile range
Look at central grouping by excluding too and bottom 25%.
Variance
Indicates how scores vary from the mean. Subtract each score from the mean. Square each deviations and add them. This gives total variance. Smaller number shows more consistent.
Standard deviation
Calculate square root of variance.
Indicated whether distributions is narrow or wide.
Small number suggests variance is more standardised.
