- Data must be in the form of frequencies.

- Random sample data are assumed.

- A sufficiently large sample size is required. There is no accepted cut off. Some researchers would set 50 as the minimum sample size, others as low as 20.

- Must be calculated using actual count data not percentages which would have the effect of pretending the sample size is 100.

- Adequate cell sizes are assumed. A common rule is that 5 or more frequencies in each cell is required.

- Frequency data must have precise numerical values and be organised into categories or groups.

- Observations mst be independent.

- Sum of all the frequencies must be the same as the number of subjects.

- Assumes a known distribution of expected frequencies calculated without reference to the observed values.

- Non directional hypotheses are assumed. Chi square tests the hypothesis the the two variables are related by chance. It does not show causal relationships - i.e. that A causes B or vice versa.

- Assumes normal distribution of the deviations, not the data itself (non-parametric).

- Nominal, ordinal or interval data can be used.