Chi Square Flashcards
What does the chi-square test?
It is whether the distribution of frequencies in the different samples are varied enough that it is unlikely to have come from the same population.
What are the two different chi-square tests?
Standard chi-square- two nominal variables
One-sample chi-square - one nominal variable
How do we work out the chi-square value?
- Find total frequencies for each row and column and the overall value.
- Using the totals we work out the expected frequencies of each cell if there is no difference.
- Once found we enter the information into the formula for each cell. Observed freq. - expected freq., squared, divided by expected freq.
- To find chi-square value we sum up these results
How can we infer the chi-square?
Bigger chi value means there is a bigger difference. More statistically significant
Once we’ve calculated our chi-square value, how do establish its significance?
We calculate the degrees of freedom. Done using this equation
(no. Columns - 1) x (no. Rows - 1).
Once found, we check the significance tables for chi-square. If our figure exceeds the value given then it is significantly different.
How do we calculate chi-square in SPSS?
Analyze -> descriptive statistics -> crosstabs
Put row variable into row box and same for column.
Select statistics, then chi-square then continue.
Then click cells, select expected, unstandardised, then continue and ok.
What is partitioning chi-square?
This is what happens when we have a table larger than 2x2. We do this so that we can find out which comparison the significance, if there is one, is coming from. If we have a 2x3 table, we partition this into 3 2x2 tables and then conduct separate chi-squares for each table.
What do we have to change when partitioning in chi-square?
The degrees of freedom becomes 1.
The significance level changes to as it’s the same set of data but it’s been split so level has to change.
What is a one-sample chi-square?
This is done when we know the null populations distribution already.
We can then calculate the expected frequency for our data set and do the chi square for just our one group data set.
DofF is calculated by how many conditions with this test.
What is the McNemar test?
This is used in chi-square when we have two related samples of nominal data.
Known as the test of the significance of change.
When conducting this, we will only use the changed data from the data set.
This test states that if condition has no affect, null, then the change will be equal, 50% one direction, 50% the other way. (Expected data)
Data is then calculated using one-sample chi square
What is chi-square used with?
It is used with nominal data in the form of frequency counts. Used when there are one/two nominal variables
