Forum Discussion

daniellf's avatar
daniellf
Frequent Visitor
6 years ago

Survey Results Analysis

My Company has run an employee opinion survey (keep in mind over 100k surveyed, 31 questions in the agree/strongly agree/strongly disagree etc. format) 

 

I've been given a report that sumarizes the information into a hierarchy, informing me how many valid responses were collected for each question in each level of the hierarchy and the % of those valid results per answer option. 

 

The problem is that there are over 5000 organizational units (see simplified example below), but i can't create a simple hierarchy on power bi because my results are already summed up in the next higher org. level.

 

Additionally, the levels below an org unit wont necessarily add up to what the level already shows: Sweden, Greece, Germany, France and Italy dont add up to Europe because there are people counted only at the europe level; the same loghic applies to Europe and Africa not adding up to the Europe & Africa pre-sum. 

 

The data set is too big for manual changes, what can I do to make this process easier? 

 

 

L1L2L3L4Valid Responses% Favorable 
Europe & Africa   10078
Europe & AfricaEurope  5361
Europe & AfricaEuropeFrance 1084
Europe & AfricaEuropeItaly 1096
Europe & AfricaEuropeGermany 1052
Europe & AfricaEuropeGreece 1078
Europe & AfricaEuropeSweden 1063
Europe & AfricaAfrica  4398
Europe & AfricaAfricaWest Africa 952
Europe & AfricaAfricaWest AfricaMorroco978
Europe & AfricaAfricaEast Africa 2091
Europe & AfricaAfricaEast AfricaKenya1084
Europe & AfricaAfricaEast AfricaEgypt561

1 Reply

  • Anonymous's avatar
    Anonymous
    Not applicable

    Ideally you'd model your Geography table as a parent-child hierarchy and then join that to your survey responses table.  A parent-child hierarchy can handle the situation you have.

    Marco Russo has an excellent article on the technique here: https://www.daxpatterns.com/parent-child-hierarchies/

    Eric