The cross-tabulation tool allows you to measure the interaction between two questions (variables). The table will only show respondents that answered both of the questions, meaning the frequencies shown may differ from a standard frequency table. The cross-tab report will also show Pearson’s Chi-Square Statistics, which shows the level of correlation between the variables using the chi-square, p-value, and degrees of freedom.
You can filter the data based on the following:
What are row and column percentages?
What does the Pearson’s Chi-Square Statistics table mean?
Pearson’s chi-square statistic is used to determine the goodness of fit between the two questions being correlated. In order to measure this, we need a null hypothesis. The null hypothesis measured using the cross-tab tool is that the two questions included in the cross-tab are correlated.
If the Chi-Square value listed in the table is higher than any of the critical values for any of the significance levels (p=.01, p=.05, or p=.1), then we can reject the null hypothesis established. In other words, if the Chi Square value that is listed in the table is higher than any of the other values shown for the 1%, 5%, or 10% significance levels, then the two questions being measured are not highly correlated. The p Value listed under the Chi-Square value in the table (second row) is the level at which the two questions are considered highly correlated. The degrees of freedom is the reference point for a chi-square table. This is determined by calculating the number of observed responses versus the number of expected responses, then subtracting 1 from that value.
The general standard confidence level used is p=.05, or the 95% confidence level. The cross-tab tool gives values at the 99% confidence level (p=.01), 95% confidence level (p=.05), and the 90% confidence level (p=.1). There are times when the chi-square is less than or equal to the critical value at the 99% confidence level but not at the 95% confidence level. It is generally stronger to use the value at the 90% or 95% confidence level.
Chi-square analysis should only be used for categorical values which typically include nominal or ordinal data.