analysing data

• qualitative data - content analysis

  content analysis:

  • taking words, turning into number data after putting into categories, taking info from Ps, diaries, pictures, questions etc.
  • repeat it to ensure reliability, do another analysis. try get someone else to repeat
• qualitative data - thematic analysis

  thematic analysis:

  • usually for interview, got a transcript of the interview, go through interviews and look for certain themes that come up. how many people talked about -?
• levels of measurement (types of data)

  nominal data:

  • categories (names)(bar chart)

  ordinal data:

  • rankings, things placed in order (rating scale - made up by researcher)

  interval data:

  • ordered numbers with equal divisions e.g. test scores (proper measurements e.g. time, distance, weight, temp)(gets put with ratio data)

  ratio data:

  • similar to interval but with a fixed zero point e.g. weight (gets put with interval data)
• inferential statistical tests


  • you can infer a conclusion from them
  • they can never show 100% certainty that the IV is causing the DV (we can work on different probability levels)
  • in psychology, they work to a 95% probability level, that our results are 'significant' (5%, P<0.05)
  • avoid using 'prove', it supports not proves
  • P<0.05: probability results are due to chance is <5%.
  • 5% chance of making an error
• type 1 and type 2 errors

  type 1:

  • refers to situations when the researcher has been overly optimistic and has rejected the null hypothesis when they shouldn't have, the results were actually due to chance

  type 2:

  • refers to situations when the researcher has been overly pessimistic, and has accepted the null hypothesis when they shouldn't have, and should actually have accepted their experimental/alternative hypothesis
  • if results are significant have to reject the null, if results are not significant you accept the null
• example paragraph

  as the observed value of ... is higher/lower/equal to the critical value of ..., we can conclude that the results are significant/not significant at the 0.05 level for one tailed/two tailed hypothesis. this therefore means that we can reject/accept the null hypothesis and reject/accept the experimental hypothesis
• how to decide which test to use - research design

  research design:

  • is it related (repeated measures) or unrelated (independent groups)
• how to decide which test to use - research aim

  research aim:

  • significant difference - 2 groups of Ps in 2 conditions (IV and control group) - aims to test for a difference
  • significant association - correlation between 2 variables - association
• how to decide which test to use - level or measurement

  level of measurement:

  • nominal - categories/frequency count
  • ordinal - list of data that can be ranked in order
  • interval - data that is in equal increments
• how to decide which test to use - statistical tests (spearmans and wilcoxon)

  statistical tests:

  • spearmans rho:
  • test for association - produces a correlation coefficient
  • ordinal or interval data and the research design can be either related or unrelated
  • wilcoxon signed ranks test:
  • test of difference for related data
  • repeated measures or matched pairs
  • ordinal or interval data
• how to decide which test to use - statistical tests (mann-whitney U and chi-squared)

  statistical tests:

  • mann-whitney U test:
  • test of difference for unrelated data
  • independent groups design
  • ordinal or interval
  • chi-squared:
  • test of difference or association
  • nominal, unrelated
  • independent groups
• unrelated t-test


  test of difference

  • interval data
  • independent groups
• related t-test


  test of difference

  • interval data
  • repeated measures/matched pairs
• choosing a statistical test

  difference/correlation?:

  • is it a 'test difference' or is it a 'test of association'

  design:

  • is it repeated (related) measures or is it independent (unrelated) groups?

  data:

  • what type of measurement (type of data) is being measured?
• steps to identify the correct test

  is it correlation?:

  • yes - spearmans
  • no - chi-squared, wilcoxon, mann-whitney (test of difference)

  type of data, is it nominal (categories)?:

  • yes - chi-squared (c for categories)
  • no - wilcoxon, mann-whitney (must be ordinal or interval)

  is it repeated mesures?:

  • yes - wilcoxon (one name)
  • no - mann whitney (two names)(therefore 2 groups of 2 Ps)
• in order to justify which test it should be

  • state whether it is a correlation or a test of difference
  • state whether categories or ordinal/interval data is used
  • state whether repeated measure or independent groups