Please enable JavaScript.

Coggle requires JavaScript to display documents.

FRCA Stats (Statistical terms and calculations (Null hypothesis: no…

- - - - Power: probability of rejecting a false null hypothesis. Probability of avoiding a T2 error. Probability of concluding a result is statistically significant
        Power = 1 - β
        If no difference between 2 groups is detected AND power is adequate, then it can be concluded that there is no clinically important difference.
        Determined by calculations or normograms to pre-determine sample size of population to avoid
- - - - Frequency of observed results vs Frequency expected if no difference between groups
      - Aim to accept or reject null hypothesis.
        Pick DoF and see if chi-square result is above (reject null) or below (accept null) critical values on a set table
  - - - Parametric:
        Mean, SD, Variance
        Kolmogorov-Smirnov test is a formal test to see if a distribution is parametric
        
        2 groups (Student T test)
        Relies on knowledge of difference of means of samples and SE of difference.
        T = difference between means / SE of the difference
        p-value is read from statistical tables
        
        Paired Student t test
        (Paired: 2 variables being tested from same patients)
        Useful for small samples (<60).
        
        Unpaired Student t test
        (2 different groups of patients studied)
        
        Pearson correlation: used for parametric distribution to measure linear relation between variable.
        
        more than 2 groups
        
        Paired ANOVA (analysis of variance)
        
        Unpaired ANOVA (analysis of variance)
      - Non-parametric:
        Median, range, interquartile range
        
        2 groups
        
        Paired:
        Wilcoxsin signed-rank test
        
        Unpaired:
        Mann-Whitney U test
        
        Spearman's correlation.
        
        more than 2 groups
        
        Paired:
        Friedman test
        
        Unpaired:
        Kruskal-Wallis 1-way analysis of variance