Null Hypothesis And Significance Testing Flashcards
What is a null hypothesis ? (Ho)
A hypothesis that states there will be no difference in the data
What is an alternative hypothesis? (Ha)
a statement that suggests there will be a relationship or a difference in the data
What is Null hypothesis significant testing? (NHST)
A methods of statistical inferences.
Which involves turning a research question into two hypothesis.
A null hypotheses and an alternative hypothesis.
When is a null hypothesis considered to be true?
It is considered to be true until there is no evidence against it.
What type of hypothesis are we attempting to find evidence for?
Alternative hypothesis.
What is a one tailed hypothesis?
State which direction the effect, differences or association will be in.
What is a two tailed hypothesis?
Do not state the direction of the effect or association will be in.
What type of data can we only collect withe nominal level data?
Frequencies
How are descriptive statistics presented?
As frequencies.
What is chi square and what does it asses?
An inferential statistical test that is used for nominal level data.
It assesses if there is an association between categorical variables.
How do we estimate expected frequencies (which categories we would expect people to fall into by chance?
(Row total) x (column total) Divided by N
What does the chi squared test do?
Measures the discrepancy between observed frequencies and expected frequencies.
What results need to show from a chi squared test to show there is an association between the IV and DV?
Also an association between two categorical variables ?
Observed frequencies need to differ largely enough from expected frequencies.
What does the type of chi square test depend on?
-Number of rows and columns
What does every statistical test have?
Assumptions.
What are the assumptions for in statistical tests?
Rules that our data must adhere to in order to be analysed using that test.
We look at our data then make a decision based on certain criteria.
What are the assumptions for chi square?
- Independence - all frequencies must unique. This means it cannot be used for within subjects/repeated measures designs.
- All expected values should be above 5
If there is lots of conditions all expected values should be greater or equal to 1 And no more than 20% of expected counts should be less than 5.
Sample size is important we need to have an appropriate number of data points. But if it is too large this can lead to misleading results.
What are P values?
They evaluate how the well the data in the sample supports the argument that the null hypothesis is true.
What do High P values indicate?
Data is likely with a true null
What do low P values indicate?
Data is unlikely with a true null.
What can we do to the null hypothesis if there is a low P value?
Reject the null hypothesis.
What is the P value in technical terms?
A p value is the probability of obtaining an effect at least as extreme as the one in your sample data, assuming the truth of your null hypothesis.
General rules of P values?
If P value is below 0.5 it is statistically significant.
The level at which we accept a result t o be significant is known as the alpha level (a)
What is a test statistic for?
Let’s the reader know which statistical test was conducted.
What is the the test statistic for chi square?
Chi symbol and a squared symbol
X2
What does the effect size tell us when using chi squared?
How strong an association is between two variables.
What type of effect size would we use for 2x2 chi square?
Phi
What effect size would we report for other chi squared designs?
Cramer’s V.
What problems are there with NHST?
Type 1 and type 2 errors
What is a type 1 error?
Rejection of a true null hypothesis (Finding a false positive)
What is a type 2 error?
Failing to reject a null hypothesis (known as a false negative)
What is the problem of running to many statistical tests?
Increase the chances of a type 1 error occurring.
How do we prevent a type 1 error occurring?
Change the alpha level and divide the alpha level (0.05) by the amount of tests you conduct. (Bonferroni correction)
What is P hacking?
A method of manipulating data to achieve significant results by conducting multiple inferential statistical tests.
Omitting other information
Changing the DV
Stopping to gather data when you have significant result.
How can we stop P hacking?
Preregistration and moving the alpha level down to 0.005 instead of 0.05
