Stat tests

• Why do we use statistical tests?

  Tells us if results are significant or a due to chance and if the null hypothesis can be accepted or rejected
• What does probability mean? How does this apply to scientific research?

  The likelihood an event will occur. We analyse our results at a certain level of probability (At a p<0.05 level, we can say with 95% confidence that our results were or were not significant. At p<0.01, our confidence will increase to 99%)
• When would standard deviations be an appropriate statistical test to select?

  To investigate the spread around the mean of an individual data set, of which the data has been measured
• What do we mean by standard deviations? What type of measure is this?

  The spread of data around the mean - if the value is large, there is a large spread. It is a measure of dispersion and produces a value like ±x
• How can standard deviation be calculated?

  Find the mean of a data set. For each point in the data set, subtract the mean and square it. Add up this value for each point in the data set and divide by number of values. Finally, square this answer
• How can standard deviations show the significance of results?

  If the standard deviations of two data sets are overlapping, the difference between the two means is not statistically significant (any difference is just due to chance) If the standard deviation are not overlapping, the differences in the two means are statistically significant (the differences are not due to chance, but some other factor)

  The standard deviations can be drawn as error bars on graphs, which make the overlapping easier to identify
• Why is the standard deviation a more useful measure of dispersion than the range?

  Less influence by outliers and anomalies
• When would Student T-test be an appropriate statistical test to select?

  To compare the means of two unrelated sets of data, of which the data has been measured
• How is the Student T-test result calculated?

  For each data set, square the standard deviation and divide by number of values. Then add these values for each data set together and find the square root. Subtract the mean of data set B from the mean of data set A and divide by the squared value found earlier.
• How can we use the Student T-Test result to see if results are significant?

  Find the critical value at the set probability level and degree of freedom ((n₁-1)+(n₂-1)) For results to be significant, the observed value must be less than the critical value
• When would Chi Squared be an appropriate statistical test to select?

  When looking at data that has been counted (not measured) and seeing if difference in expected and observed frequencies are significant
• How is Chi-Squared calculated?

  For each point in a data set, subtract the expected frequency from the observed frequency and square it. Then divide by the expected frequency. Add up the values for each point in the data set
• How can Chi-Squared show us if results are significant?

  The observed value must be greater than the critical value
• When is a Correlation Coefficient used?

  Value between +1 and -1 shows the correlational relationship between two variables
• What does a Correlation Coefficient of 0 indicate?


  No correlational relationship
• What do Correlation Coefficient values close to +1 or close to -1 indicate?

  +1 = strong positive correlation

  -1 = strong negative correlation
• When do You use spearman’s rank?

  Correlation analysis