Stats tests

• Descriptive statistics - summary statistics that identify trends and analyse sets of data.

  Examples - mean, mode, median range
• Inferential statistics - refers to the use of statistical tests which tell whether the difference/relationship is significant or not which helps decide which hypothesis to accept and reject
• Probability - likelihood of results being due to chance
• Level of significance - in psychology 0.05 level is used as this is between being stringent and not stringent - in the middle
• Null hypothesis - states that any difference between to conditions is due to chance
• If a result is statistically significant we reject the null hypothesis and accept the alternative hypothesis (either directional or non-directional)
• Type one error - occurs when null hypothesis is rejected when it is true. This is more likely to occur when a less stringent level of significance is applied such as 0.1 or 0.5.

  Known as false positive
• Type two error - occurs when null hypothesis is accepted when it is false. More likely to occur when a more stringent level of significance is applied such as 0.01 or 0.005.

  Known as false negative
• Example of type 1 error - Man being told he is pregnant when he is in fact definitely not
• Example of type 2 error - woman who is pregnant being told she is not pregnant
• Criteria needed to work out if something if significant:

  • Number of participants (N)
  • Level of significance = 0.05
  • Observed/calculated value
  • One-tailed or two-tailed test?
• Levels of measurement - types of data:

  • Nominal
  • Ordinal
  • Interval
• Nominal data - data that is in separate categories, data can only fall into one category.

  Example - grouping people in your class who are tall or short
• Ordinal data - data that is ordered in some way

  Example - gathering test scores up and ordering them from highest to lowest
• Interval data - data is measured using units of equal measurement

  Example - temperature scales and reaction times
• Parametric tests - more powerful than non parametric tests as they are better at detecting significant differences.

  They use means and standard deviations rather than using ordinal or nominal data
• Criteria for parametric test:

  • Level of measurement - Interval
  • Data comes from a population that has a normal distribution
  • Variances of the 2 samples are similar
• Criteria to work out what stats test to use:

  • Difference or relationship/correlation in data
  • Level of measurement - nominal, ordinal or interval
  • Related or unrelated data - if correlational study or repeated measures then data is related
• Chi Squared:

  • Test of difference
  • Nominal data
  • Unrelated data (independent measures)
• Sign test:

  • Test of difference
  • Nominal data
  • Related data (repeated measures/matched pairs)
• Mann-Whitney:

  • Test of difference
  • Ordinal data
  • Unrelated data (independent measures)
• Wilcoxon:

  • Test of difference
  • Ordinal data
  • Related data (repeated measures/matched pairs)
• Unrelated t-test:

  • Test of difference
  • Interval data
  • Unrelated data (independent measures)
• Related t-test:

  • Test of difference
  • Interval data
  • Related data (repeated measures/matched pairs)
• Spearman's Rho:

  • Test of correlation
  • Ordinal data
  • Related data
• Pearsons r test:

  • Test of correlation
  • Interval data
  • Related data