1E. Intro to Null Hypothesis Significance Testing AND Choosing the appropriate stats test for differences between two groups Flashcards
in data distributions, what goes on the x axis and what goes on the y axis?
x-axis:bins/categories
y-axis: frequency
in normal distribution what is the relationship between mean median and mode?
mean = median = mode
if you find that the mean and median are very similar, what can you assume about the data you are handling?
that it is normally distributed
what type of tests does normally distributed data allow for?
allows for parametric tests to be used
if the mean and median are different what does this mean?
means that the data is not normally distributed
if the data is skewed to the right in non-normal distributed data, what does this mean?
the mean is less than the median
if the data is skewed to the left in non-normal distributed data, what does this mean?
the mean is greater than the median
Define sampling error
the difference found when you get slightly different mean and standard deviation (i.e. descriptive statistics)
what are the 3 types of null hypothesis significance tests?
- tests for difference
- tests for relationship
- tests for frequencies
in null hypothesis statistical testing, what are the 4 steps to it
- establish the null hypothesis
- decide on a critical significance level a (usually 0.05)
- choose your test and calculate your test statistic (p-value)
- accept or reject the null hypothesis
what is the alternative hypothesis?
when we reject the null hypothesis, this the alternative hypothesis that there is a difference
what is the critical significance level (a)?
the level of uncertainity we are prepared to accept
- this means we are likely to get it wrong 5% of the time
outline the 5 steps to calculate p value in null-hypothesis significance testing
- decide type of data (ordinal, nominal,ratio etc.)
- decide if you are looking for differences or relationships
- work out if you satisfy certain assumptions
- identify appropriate test
- calculate the P-value and decide if it is less than critical significance level
how do you decide if you reject or accept the null hypothesis?
If P ≤ a –> reject the null hypothesis as there is a statistically significant difference
If P ≥ a –> accept the null hypothesis as there is NO statistically significant difference
what are the 4 decisions needed to make when choosing the right stats test for looking at differences between two groups?
- related or unrelated? (means paired or unpaired)
- two samples or more than two samples?
- parametric or non-parametric?
4.predicted direction of change?
what is an example of related data?
- matched up
- paired individuals
- repeated measurements (e.g. measure height of year 5s, then height of the same children in year 6)
what is an example of unrelated data
non-repeated measurements E.g. measure heights of a group of year 5s and a group of year 6s
what does parametric mean?
when the data is well-described by a mean and standard deviation (means it is normally distributed)
when is data non-parametric
- when data is better described by a median and IQR
- if data is ordinal
- in scale data, if data is not normally distributed (skewed) or unequal variance
if you are expecting the difference of data to be in a particular direction (increase or decrease), what test do you use
a one-tailed test
if you only expect a “difference” and don’t know if there will be an increase or decrease, what type of test do you use
a two tailed test
when do you decide a test is one or two-tailed?
after you decided what specific test you are running (t-test, kruskal-wallis etc)
