UPCOMING WEBINAR: Optimize product ranges using TURF, Factor, and Correspondence Analysis The table above shows a crosstab of two Nominal: Mutually exclusive variable sets. It shows that respondents aged 65 or more are more likely to have said they Like the brand (Diet Pepsi) than the overall percentage (13%).
Although this intuitive understanding is correct regarding how to interpret the data, that is not the case o the technical level, because the 24 respondents in the 65 or more category are also included in the total sample of 105. Thus, if we compare the 24 with the 105, we would double-count respondents - to use the more formal statistical language, we would violate the assumption of independent samples.
Method
The only way respondents 65 or older can be different from the total is if they are different from respondents younger than 65. To test whether that is true or not, you need to collapse all the categories other than Like and 65 or more.
- Combine all categories other than 65 or more into a single category.
- Combine all categories other than Like into a single category.
The test used to highlight the 43% on this table is the same test used in the table above (i.e., when creating the larger table). Displayr automatically creates many smaller tests in the background and uses these to compute significance. However, when using Multiple Comparison Corrections, it is possible to get situations where the collapsed table will have a cell marked as significant and the non-collapsed table will not (and vice versa).
Please note that when working with Banners that have missing data, the Total column will be different than the overall NET column, see Banners in our technical reference for more details.
Next
How to Investigate Your Statistical Significance Testing
How to Compare Significant Differences Between Columns
How to Apply Multiple Comparison Correction to Statistical Significance Testing