Statistical/inferential testing

• Why is statistical testing used in psychological

  research?

  To assess the likelihood that a difference/relationship occurred due to chance.
• Explain what is meant by a p-value.


  P-values represent the probability that the difference/relationship occurred due to chance. E.g. a p-value of 0.05 means there was a 5% probability that the result occurred due to chance and so the researcher can be 95% confident that there

  was a significant difference/relationship.
• Explain why researchers

  typically use p=0.05.


  To balance the risk of making a type I and a type II error.
• Distinguish between a type I and a type II error.


  A type I error refers to when the researcher states the result is significant when it is not. This means they accept the alternative hypothesis and reject the null hypothesis in error.

  A type II error refers to when the researcher states the result is not significant when it is. This means they accept the null hypothesis and reject the alternative hypothesis in error.
• Explain how you can assess the likelihood that a researcher has made a type I error.


  The p-value can be used to identify the probability that the result occurred due to chance and so the probability of making a type I error

  Using the critical table, assess whether the result would still be significant when using a stricter p-value e.g. after checking significance using 0.05, use 0.01. This would reduce the risk of making a type I error from 5% to 1%. If the result is still significant, the researcher can be more confident they had not made a type I error.
• When is it appropriate to use Chi squared?


  independent groups design and nominal level of data
• when is it appropriate to use the Sign test?


  repeated measures designs and nominal level of data
• when is appropriate to use Chi squared?


  test of association and nominal level of data
• when is it appropriate to use Mann Whitney?


  independent groups design and ordinal level of data
• when is it appropriate to use Wilcoxon?


  repeated measures design and ordinal level of data
• When is it appropriate to use Spearman's rho?


  test of correlation and ordinal level of data
• When is it appropriate to use unrelated T-test?


  independent groups design and interval level of data
• when is it appropriate to use related t-test?


  repeated measures designs and interval level of data
• when is it appropriate to use Pearson's R?


  test of correlation and interval level of data
• Explain how you would use a statistical test to establish concurrent validity.


  Concurrent validity would involve correlating participants’ results from the researcher’s new test to the same participants’ results on an established, valid test measuring the same thing.

  A stats test appropriate for correlations should therefore be used.

  This means, depending on the level of measurement, that the Spearman’s rho or Pearson’s r test should be used.

  If the new test is also valid, there should be a significant positive correlation between the participants’ two sets of results.
• Explain how you would use a statistical test to establish inter-observer

  reliability.

  Inter-observer reliability would involve correlating the two observer’s independent recordings.

  A stats test appropriate for correlations should therefore be used.

  This means, depending on the level of measurement, that the Spearman’s rho or Pearson’s r test should be used.

  If the observers are consistent in their recordings, there should be a significant positive correlation.
• Explain how you would use a statistical test to establish test-retest reliability.


  Test-retest reliability would involve correlating the participants’ result with the same participants’ result when completing the test again.

  A stats test appropriate for correlations should therefore be used.

  This means, depending on the level of measurement, that the Spearman’s rho or Pearson’s r test should be used.

  If the participants’ two sets of results are consistent, there should be a significant positive correlation.
• What is the value of S and how do you calculate it?


  The value of S is the frequency of the least frequently occurring sign (ignoring any 0s).

  To calculate it, all participants will need to be assigned a sign (+ if their result increased, - if it decreased, 0 if it stayed the

  same).

  Then identify the least frequently occurring sign (+ or -) and count how many times it occurs.