Check out the LimeSurvey source code on GitHub!

Complex Data... Best Practices

5 years 7 months ago - 5 years 7 months ago #63684 by bmdavis
I'm implementing LimeSurvey as a data collection method, as a part of a database migration from MS Access to mySQL/PHP. I say a "part of" because I'll be making other external scripts and data tables which I'll have to work with that are outside the scope of LS.

I'll mostly be using the data entry screen for responses taken on paper forms and/or importing responses through CSV files. Although the actual online survey form will be a useful feature for the future, it will likely see very little use in the immediate future. So you could say that I'm really using LimeSurvey for it's database architecture and data entry methods (apparently what it was originally designed for long ago).

In terms of number of responses, I'm expecting 100,000+ entries per year based on historical data. I'll also be migrating about 600,000 responses from previous data. These responses fit into a hierarchy of related records. Typically, 15-20 responses make up a "class" (about 7500 classes per year), and each of the classes belongs to 1 of the 200 or so "locations". There are also related tables of information relating to "programs", "trainers", etc. Obviously this is where my custom scripting and external data tables come into play.

My main question concerns managing the quantity of responses, and organizing them into "classes" and "locations". I'm not sure if it would be best to use one survey to store all the data, use one survey for each class (probably impractical due to number of classes per year ~7500), use one survey for each location, or some other method I'm not yet realizing. In all these cases, I'm trying to see how best to utilize tokens. I don't really want the survey open to the internet for just anyone fill out, but I don't want it to be cumbersome to manage. A new feature I'm not familiar with since my 2 years experience with 1.80 are the multiple use tokens... I'm thinking this could be a good use maybe. Also, how ugly would the process of data entry and/or importing data into the above survey cases (combined, per class, or per location) become?

So, I'm looking for ideas, experiences, etc. for using LimeSurvey on such a scale. Lessons learned greatly appreciated!
Last Edit: 5 years 7 months ago by bmdavis. Reason: typos

Please Log in to join the conversation.

5 years 6 months ago #63891 by bmdavis

Maybe my question was too generic. I think I have it figured out how I will proceed, but I'd still be curious to know any opinions/experiences on managing such a large dataset of responses.

Please Log in to join the conversation.

Imprint                   Privacy policy         General Terms & Conditions         Revocation information and revocation form