Data Analysis

• Descriptive Statistics:

  Describe and summarise collected data/statistical information; central tendency and dispersion measures
• Measures of Central Tendency

  • One number to represent general trend or pattern set
  • Often used to describe score group from psychological study
  • Mean: average
  • Mode: most common answer
  • Median: middle value
  • Often used in with dispersion measures to indicate how representativeness
• Central Tendancy Types

  • Mean:
  • Most sensitive
  • Considers all scores
  • Easily distorted by extremes
  • Unrepresentative
  • Arithmetic average: Give unrealistic results (decimals)
  • Used with interval/ratio data
  • Mode:
  • Most frequent number
  • Useful for large data sets
  • Unaffected by extremes
  • Unreliable for small ones as they can be bi/multimodal, not useful as central tendency
  • Median:
  • Middle number
  • More representative than mean in smal sets
  • Unaffected by extremes
  • Less representative in polarised ones
  • Often used with ordinal data
• Measures of Dispersion

  • Useful to know score set spread (variability/dispersion)
  • Shows how representative central tendency measure is
  • It is not distorted/skewed by extremes
  • Data set with lowest dispersion (more numbers resemble mean) is more representative
  • Large spread would suggest lots of variation from mean
  • Range
  • Standard Deviation
• Range

  • Normally used with median
  • Subtract lowest score from highest
  • Ignoring other numbers
  • For those with extremes it’s better to calculate interquartile range (remove top and bottom 25%)
  • Easy to calculate
  • Indicates extent of individual difference
  • Easily distorted
  • Only uses 2 numbers no matter set size
  • Basic dispersion indicator
  • Doesn’t indicate how representative mean is
• Standard Deviation

  • More sensitive and representative
  • Uses whole data set
  • Mean score distance from score set mean
  • Larger SD, and more dispersed score = mean less representative
  • If data is from random representative sample, SD can make inferences of target population
  • If sample is not random, instead of SD formula divide by just n
  • When considering SD and normal distribution curve, behaviours studied must be normally distributed in sample
  • Tells you how precise mean is of true mean, small SD suggests precise
  • Less affected (still impacted) by extremes
  • More difficult to calculate