probability and significance

• The null (H0) and alternative (H1) hypotheses are competing statements. The null hypothesis suggests no causal relationship exists between the independent variable (IV) and the dependent variable (DV), whereas the alternative hypothesis suggests a causal relationship.
• Deciding which hypothesis to accept and which to reject

  Evidence is collected by conducting controlled experiments. The researcher varies levels of the independent variable and measures the result change in the dependent variable. Accepting or rejecting either hypothesis depends on whether or not the results of this test match the null or alternative hypothesis.
• Variability


  The researcher must decide if the data collected is strong enough before rejecting the null and accepting the alternative hypothesis. Even in a well-controlled experiment, there will always be some natural variability in the data collected. Researchers use probability to minimise the chance that they reject the null and accept an alternative hypothesis in error due to natural variability.
• P=<0.05


  Psychologists accept a 1 in 20 chance that their results are due to chance variation. As the fraction, this probability is 1/20, and 5% as a percentage. The probability is usually referred to by the decimal 0.05. P=<0.05 means there is less than a 5% probability the results gained are due to chance.
• P=<0.01


  Is occasionally used by psychologists, this level of significance is usually required of studies attempting to support a particularly controversial theory or if the psychologist is conducting a replication. Using the P=<0.01 level reduces the probability of accepting the alternative hypothesis in error to less than 1 in 100.
• Type I (1) error


  When researchers accept the alternative hypothesis (reject null) in error, the data collected has passed the level of significance, but really the findings were due to chance variation. Using P=<0.05 means this will happen in around 1 in 20 studies.
• Type II (2) error


  When researchers reject the alternative hypothesis (accept null) in error, the data collected has not passed the level of significance, but really the participants not acting as expected is due to chance variation hiding the causal relationship between IV and DV.
• How to reduce Type errors

  To reduce the chance of a type 1 error, a researcher can decide to use a P=<0.01 level of significance, however, this then increases the likelihood of a type 2 error. The same is true in reverse; using P=<0.05 reduces the chance of type 2 error but increases the possibility of a type 1 error compared to using P=<0.01.