1 Reply Latest reply on May 18, 2016 6:31 AM by erolst

    overall standard deviation of several field contents

    nordlicht

      I have a Filemaker file with several fields (say 10) to fill in numbers for statistics (say n=100 in each field). Have calculated a mean and a standard deviation for each field.  Question: How can I create an overall mean and overall standard deviation of all numbers of all fields (10 x 100 numbers) with a Filemaker formula. Statistically it is wrong to calculate the overall standard deviation as mean of all single standard deviations of every field. My only idea was to export all fields + numbers to "excel", arrange all numbers into one column and send them back to filemaker via import to use them in my layout in one field. But that is unprofessional, isn`t it? And if one new number is given in one field there is no way that the standard deviation is automatically actualized.

      Is there any idea in the community to manage that problem via FM?

        • 1. Re: overall standard deviation of several field contents
          erolst

          nordlicht wrote:

          I have a Filemaker file with several fields (say 10)  […]  export all fields + numbers to "excel", arrange all numbers into one column […]  Is there any idea in the community to manage that problem via FM?

          The problem is your data model; it's not properly normalized.

           

          If you use related records (where one field denotes the attribute – what now is the field name – and another holds the value) instead of fields, you have exactly the same arrangement as in Excel – all values in one row.

           

          Now you could use either a summary field with a found set of all records, or a Cartesian relationship and apply a function to all related values, as in StdDev ( MyValueTable__Cartesian::value ). With either method, new entries and / or new attributes can be easily / are automatically subsumed.