Statistical tests

• Factors affecting choice of statistical test

  1. Difference or correlation
  2. Experimental design: matched or unmatched
  3. Type of data
• Types of statistical tests

  Carrots should come mashed with swede under roast potatoes

  1. Chi-squared, Sign test, Chi-squared
  2. Mann-Whitney, Wilcoxon, Spearmann’s Rho
  3. Unrelated T test, Related T test, Pearson’s R
• What statistical tests uses nominal data

  Chi-squared (test of correlation or test of difference:unrelated design) and sign test (test of difference: related design)
• What statistical tests use ordinal data

  Spearmann’s Rho (test of correlation), Wilcoxon (test of difference: related design) and Mann-Whitney (test of difference: unrelated design)
• What statistical tests use interval data

  Pearson’s R (test of correlation), Related T test (test of difference: related design) and Unrelated T test (test of difference: unrelated design)
• Sign test - S

  1. Convert data into nominal data by working out the pps with the higher amount, then the lower by subtracting the control from the independent
  2. use the table to add up the number of pluses and minuses
  3. Take the less frequent sign, and call that s
  4. Compare the calculated value with the critical value using the table (n.o. of pps and .x = level of significance, check for one tailed or two tailed with hypothesis
  5. Then check for significance
• Unless stated otherwise, the level of significance is always 0.05
• Wilcoxon Test - T

  1. Work out the type of hypothesis
  2. Find probability level that study is using
  3. Work out N value - number of pps who scored differently in the condition
  4. Critical = where N value and probability meet
  5. Read sentence beneath table to work out the requiremen for the critical value to be significant
• Mann Whitney Test - U

  1. Work out Na (number of pps condition 1)
  2. Work out Nb (number of pps condition 2)
  3. Where the two Ns meet, this is the critical value
  4. Read sentence underneath for significance
• Spearman Rho’s Test - rho

  1. Work out hypothesis type
  2. Find the probability level
  3. N value
  4. N value and probability meet for critical
  5. Read statement underneath for significance
• Pearson’s R - r

  1. Work out hypothesis type
  2. Probability
  3. N value
  4. Critical value
  5. Significance
• Unrelated t-test - t

  1. Work out hypothesis type
  2. Find probability level
  3. Df level = NA +NB -2 ( no of score in group A plus no of group B minus 2(minus 2 because the experiment uses 2 groups))
  4. Find critical value
  5. significance
• Related t-test - t

  1. Hypothesis type
  2. Probability level
  3. Df value = N - 1 (no of scores in each condition minus 1 since pps did both both conditions)
  4. Critical value
  5. Significance
• Chi-squared test - x^2

  1. Hypothesis type
  2. probability level
  3. Put data in contingency table if not completed
  4. Df value = (no of rows - 1) x (no of columns - 1) of contingenc, excluding totals and headings
  5. Significance
• The results are/not significant since __ = x which is higher/lower/equal to the critical value of x for a one/two-tailed hypothesis test at p<0.05 (unless stated otherwise) for x pps. Therefore the null hypothesis should be rejected/accepted and the alternative hypothesis should be rejected/accepted