Does Diet*Calc have a mechanism to identify questionnaires with unusual or repetitive response patterns (suggesting that the respondent is not reading the DHQ or taking it seriously)?
This is not built into the software. Different investigators want to do data cleaning different ways: some like calorie exclusions, some like exclusions based on percentile cutpoints of nutrients or missing data, some like to transform data and then exclude, others want to exclude those respondents who did not take the task seriously (however that is defined). By looking at the scanned data file, one can write a program to check issues related to whether an individual always checks the same frequency or portion size. It is our experience that it is difficult to discern these types of errors from kcal exclusions or from those who skip foods that they never consume to begin with. No matter how you look at it, FFQ data are always messy for a few people and prone to measurement error for all respondents. Often many hours are spent trying to develop exclusion criteria, and the conclusions vary. Ultimately, if you exclude extreme outliers based on your nutrient or food group of interest, you
Related Questions
- Does Diet*Calc have a mechanism to identify questionnaires with unusual or repetitive response patterns (suggesting that the respondent is not reading the DHQ or taking it seriously)?
- Are there issues regarding the confidentiality of an email RESPONSE since the respondents email address is attached to the returning message?
- How do I know which mechanism to use in response to a specific Funding Opportunity (FO) or for investigator-initiated applications?
