Thejesh, great questions. Here are some answers. 1. Physical because our intent is to get the opinion of aam janta. And less than 10% of them have Internet connections. The biases in only-online surveys is unbelievable. 2. Non-OCR for Hindi was an operational decision that some of our surveying teams took - something we (and they!) are regretting now. We'll stick to OMR everywhere in future, with an offline app for the surveyor in a large number of cases. 3. We have multiple vendors, some volunteer-based and some commercial. Karnataka was actually easy, thanks to our years of works here - we use a network of Govt teachers. Happy to introduce you/others to any/all of our vendors, if you're interested. 4. The survey process started in mid-Dec, to create lists of 15-or-so stratified-randomized locations In each MP constituency. Field work, driven off of a short guideline document and training, started in early Jan 2014 and will complete by the first week of March. 5. By "validation", if you mean the data that is being received, we have a number of check-points we go through, to ensure that surveyors are actually doing what they're expected to do. We begin with the random locations that people are supposed to visit and ask for a travel plan at the lowest level of detail; we have volunteers calling random surveyors daily, checking their location via questions, call-backs, etc.; we have field visits by non-interested parties to validate that surveyors ARE where they say they will be and are doing what they say they are doing; surveyors also submit images and video from the locations they visit; I can go on. The data itself is being analyzed for surveyor-bias, outliers and other typical problems, as it is received.
Again, my intent in posting here is to get ideas on how we could USE this data. Would love to hear from the group here about that! On Thursday, February 13, 2014, Thejesh GN <i...@thejeshgn.com> wrote: > Kishore, > There is so much to learn from how you conducted (operations part) such a > large survey. > > 1. Why paper based? Why not completely electronic? is it because you > wanted a paper trail? or just the availability? > 2. Why no OCR for Hindi? Specially because the questionnaire is multiple > choice type. > 3. Whom did you hire to do this? volunteers? How difficult was it manage > this? did you hire multiple vendors? Specially curious about whom did you > hire in Bangalore and Why? Could help some of us who want to do some kind > of survey in future. > 4. How long did it take just to conduct survey? > 5. What was the validation process. > > > Thank you so much for sharing this with us. > Thej > -- > Thejesh GN *⏚* ತೇಜೇಶ್ ಜಿ.ಎನ್ > http://thejeshgn.com > GPG ID : 0xBFFC8DD3C06DD6B0 > > > On Mon, Feb 10, 2014 at 3:36 PM, Kishore (Narasimhan) Mandyam < > kish...@dakshindia.org> wrote: > > Srini, I agree the 500 responses is overkill - 390 would've > done,statistically. But in our past surveys, we've found 6% to 7% of the > responses being "outliers" that we've had to discard. And this time being a > national effort, we wanted to overdo things, if need be, since re-doing a > constituency is not really an option. > The survey does attempt to get some sense of the respondent's > socio-economic background - age, education, rural/urban, roof-type, basic > assets, stove-type, etc. And we've debated various ways of asking > questions, trade-offs between questions that are easily understood and that > are comprehensively stated, the impact of translation into Hindi and other > languages, etc. We've also stayed away from "incidental" assessments of > things like leadership qualities; instead, we've added a direct option > about whether the PM Candidate is an issue the respondent uses in deciding > whom to vote for. > I want to point out that the intent of this survey is NOT to be a > barometer of how India will vote but to identify issues that people can > verbalize (and MPs can use) in relation to their constituencies. The impact > of having over 2.5 lakh people discussing these issues with their friends > and family will be significant, we believe, in people making an informed > choice in voting. > I'll post a link to the questionnaire later today. > > > > > Kishore. > ----------------------- > www.dakshindia.org > > > On Mon, Feb 10, 2014 at 2:02 PM, Srinivasan Ramani < > srinivasan...@gmail.com> wrote: > > Some thoughts - > > > * The survey of 2.5 lakh would certainly be exhaustive but I personally > feel it could be over-done. Sampling can be somewhat lesser than that to > yield more focussed results (for easier post-survey processing, I suppose). > > * But that said, the survey could seek to not to fully replicate efforts > such as what the CSDS' Lokniti already does. One weakness I believe that > the Lokniti surveys have is that while the questionnaires gauge background > and attitude towards politics variables quite well, they don't necessarily > capture reasons and expectations that much well. > > So, for e.g.,if the questionnaire and data can be more than just "what are > the reasons you vote for a candidate" .. maybe, "why do you think your > choice of party/candidate is better than the alternative".. "what schemes > have you benefitted from during MP's tenure / or hope to benefit from".. > Also, opinions such as "who do you think could be an ideal leader?" to > reflect opinions on the kind of representatives they would like to have ... > is another thing that would be helpful. > > * Beyond political opinion, better knowledge of class backgrounds will > also be useful (in terms of profession, homes, access to facilities etc). > > > On Mon, Feb 10, 2014 at 11:06 AM, Kishore (Narasimhan) Mandyam < > kish...@dakshindia.org> wrote: > > -- > For more details about this list > http://datameet.org/discussions/ > --- > You received this message because you are subscribed to a topic in the > Google Groups "datameet" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/datameet/axFGNa5cQtA/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > datameet+unsubscr...@googlegroups.com<javascript:_e(%7B%7D,'cvml','datameet%2bunsubscr...@googlegroups.com');> > . > For more options, visit https://groups.google.com/groups/opt_out. > -- Kishore, Daksh. -- For more details about this list http://datameet.org/discussions/ --- You received this message because you are subscribed to the Google Groups "datameet" group. To unsubscribe from this group and stop receiving emails from it, send an email to datameet+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
