DatAna Flashcards
What is the null hypothesis?
→no statistically significant difference in the
relevant endpoint exists (or will be observed) between the groups
What is the alternate hypothesis?
→statistically significant difference
does exist between the groups
What should the hypothesis detail?
→key endpoint, dependent variable, or parameter is being measured
and compared between groups
What is the p-value?
→estimates the probability of the
difference observed between the groups occurring due to random variation in
the data if there was not any genuine difference between the groups
What is p<a></a>
→the difference is statistically significant → H0 rejected
What is p>a?
→the difference is not statistically significant → H0 retained
What is a smaller alpha value more likely to result in?
→result in a false positive as
very strong evidence of a statistically significant difference is required to
reject the null hypothesis
What is a greater alpha value more likely to result in?
→a false positive as
only weak evidence of a statistically significant difference is required to
reject the null hypothesis
What is the unpaired t-test?
→analyse the statistical significance of differences
between the means of two groups, where the data is parametric (normally
distributed) and samples are independent (unpaired)
What variables can impact the p-value by t-test?
→the magnitude of the difference between the mean values of the two groups,
→the standard deviation,
→ the number of samples in each group
What is a two-tailed test?
→assesses whether there is a statistically significant difference
between one group and another, regardless of whether it is an increase or a
decrease
→X being
different to Y
What is a one-tailed test?
→only assesses whether there is a statistically significant
increase/decrease between one group and another
→X being
greater than/less than Y
When are non-parametric methods used?
→evidence that the variable/data
in question is not normally distributed
What is the Mann-Whitney U test?
→non-parametric equivalent of the unpaired t-test used to assess whether
differences between two groups of independent samples are statistically
significant
How is Mann-Whitney U test carried out?
→assigning ranks to the data values
observed in the study
→estimating the probability of the ranks being
distributed between the two groups in the manner observed due to random
variation
What is used to measure the central tendency for non-parametric data?
→median rather than the mean which is used in parametric
How can we determine whether data is parametric or non-parametric without a test?
→the data (when expressed as a histogram) can be visually inspected to
subjectively judge whether the shape is approximately normal
→whether there is substantial kurtosis (flattening/peaking) or skewing
(asymmetry)
How can we determine whether data is parametric or non-parametric with a test?
→Shapiro-Wilks test
evaluates the null hypothesis that the data is not normally distributed
How do we interpret the Shapiro-Wilks test?
→If
the p-value calculated is less than the chosen alpha (threshold value,
e.g. 0.05) there is sufficient evidence to indicate that the data is non-
parametric, otherwise the data is assumed to be parametric due to there
being insufficient evidence to reject the null hypothesis
Define paired data
→Samples are dependent if each group contains the same samples; each
experimental condition involves the same set of participants
→eg. the same group of participants have their blood pressure
measured before and after taking a drug
Define unpaired data
→Samples are independent if each group contains different samples; each
experimental condition involves a different set of participants, animals, cells,
→E.g. one group of participants is treated with a drug, a different group of
participants is treated with placebo
What type of test for paired and non-parametric data?
→Wilcoxon Signed-Rank
What type of test for unpaired and non-parametric data?
→Mann-Whitney U test
What can result if multiple t-test are conducted for multiple comparisons?
→increased likelihood of type-1
errors (false positives)