Some software (e.g., Q, Displayr) shows a NET on any table that presents categorical data.
The NET is the proportion of people that have chosen one or more of the categories on the table. Where the categories are mutually exclusive and exhaustive, such as with a typical nominal or ordinal variable, the NET will always be 100%. However, it is possible to have NETs of less than 100%, when dealing with summaries of binary - multi and binary - grid data. When this occurs, it signals a potential data integrity issue.
Note that SPSS Statistics automatically rebases data if the NET (shown as a total) is not 100%.
The below table shows the breakdown of ethnicity in a survey. The NET is 99%. This indicates that 1% of respondents have no recorded ethnicity data. The data is essentially missing, even though no missing value appears in the data.
In the table below, the NET is 0%. This is because the NET is always computed based on respondents with complete data. Only 7 respondents have complete data, and none of them have selected any of the brands shown in the table. These types of data issues are discussed in more detail in Checking and Understanding Missing Data.
Please sign in to leave a comment.