probability and significance

• significance


  if the result of a statistical test is significant it is highly unlikely to have occurred by chance
• Statistics tests employ a level of significance – this means that at this point the researcher can state that the relationship between the variables is due to more than just chance. They can accept the alternative hypothesis and reject the null hypothesis.

  The level of significance tells you whether to accept or reject the alternative hypothesis.
• what is the usual level of significance

  The usual level of significance in psychology is <0.05 or 5%. This means that the probability of results of the test occurred by chance equal to or less than 5%
• statistical tables

  1. Calculated value – the number from the statistical test
  2. Critical value – comes from the statistical test table
  3. N or DF– how many participants
  4. P value – significance level
•  The level of probability doubles when two-tailed hypotheses are being used – p=0.10
• There are lower levels of significance when there is a human cost to participants – for example, drug trials or one off studies where the conditions cannot be replicated again. The lower level of significance is 0.01 or 1%
• It is up to 5% possible that the wrong hypothesis can be accepted.
• type I errors

  a false positive. Where the null hypothesis is rejected and the alternative is accepted. Where the researchers claim to have found a significant difference when there is not one.
• type II errors

  a false negative. Null is accepted and alternative is rejected. Claim of no significant difference between variables when there is actually one.
• Researchers are more likely to make a type I error if they are too lenient (0.1) and more likely to make a type II error if they’re too strict (0.01)
• nominal data


  data that can be put into specific categories.
• ordinal data

  continuous data (0-10, 11-20, 21-30)
• interval data

  continuous data but represents things the researcher cannot manipulate.
• If a repeated measures or matched pairs experimental design is used, the design is a related one. If an independent measures design is used, it is an unrelated design.
• what are parametric tests

  those that make assumptions about the parameters of the population distribution from which the sample is drawn. This is often the assumption that the population data are normally distributed.

  Non-parametric tests are distribution-free.
• choosing a statistical test