Formulating a null hypothesis
There is no statistical significant difference between a serum testosterone in male athlete with conc of greater than 50 microgram/ millilitre and male athlete sprinting speed of male athlete serum conc of less than 50 mp/ml
Is the difference observed significant.
p<a>a not significant difference</a>
Comparing mean value, standard deviation , p value(parametric data).
bigger the difference in mean = smaller the. p value (data is significant)
if mean and SD are same between two groups but the p value is different, the group with smaller P value had more samples (more likely to find significant difference in a larger data)
comparing non parametric data by running a mann whitney U test
- the difference in mean doesn’t show significant data because non parametric tests don’t compare data to a mean.
- they rank data and look at difference between two numbers , so you dont look at data with big difference in numbers but look at numbers closer to each other so they will have a smaller difference so smaller p value showing significance.
to identify parametric and non parametric data what graph do you use
- histogram an QQ plot
- QQ plot shows if there is normal distribution there should be spread of data along line of best fit.
- normal bell curve(around histogram) = parametric
p value and histogram that shows significant deviation to be paired or unpaired , parametric or non parametric
- non parametric : doesn’t follow normal distribution
- paired bc the same group of people used before and after
- so you uses wilcoxon signed Rank
What are null and alternative hypothesis tested by ANOVA in the study
- looks at any difference between any different groups
- ANOVA null: there will be no significant difference between any groups
- ANOVA alternative : there will be a significant difference between the groups
Anova test result
- lots of numbers to show the computer system test is working but the only thing you are interested in is the p value
- p value > alpha (0.05), accept null , no significant difference
- p <a></a>
Two way ANOVA
- level of depression of patients that have anxiety or not
- and is there a difference of patients treated with those different drugs
- interaction effect: between patients with anxiety or not and efficacy of drug
Categorical vs continuous variables
categorical : one or the other , male or female
continueos: take any value, height, weight.
fishers exact test:
tests if there are non random associations between two categories.
How does chi-squared test work?
compare one group with the rest of the different group,
so E vs A , E vs B, E vs C, E vs D
work out p value for each one
Why bonferroni correction is limited?
- dont wanna do too many comparison bc you are more likely to get error
- reduce false positive (type 1 error)
- Tukey’s and Dunnett’s reduce error rates mathematically.
