Hi,I am a total novice using MDAT. However, for a research project I am trying to download a CSV of a custom table I've built with several recoded variables with a given state as my selected geography.I'm haven't been able to download a CSV that has a number of unique rows/instances that correspond to the COUNT total of the observations I see when my custom table is rendering through the browser. I have tried download the CSV with only 'PUMS Person Weight' toggled **I do ultimately want person-oriented results rather than household** as well as downloading with only the 'Housing Unit Weight' toggled, and with both active. Despite my stabs at various selections the most robust file I've been able to extract contains only 17,188 rows compared to the 40,447 records I would expect based on the table view that google chrome generates.
The output does not contain any kind of SERIAL or primary key type descriptor either, so I am feeling as though some kind of tabulation or aggregation is happening that I am not desiring.
Additionally, one of my recoded variables contains select occupations from OCCP, and I'm noticing that the CSV file is pushing out numerous instances of blanks for that field, despite not indicating blank responses in the restricted universe portrayed in the table view from google chrome.Any insights and wisdom would be hugely appreciated, thanks!-Mary
In the days since posting this question I have gained some understanding into the weighting system that helped me appreciate the discrepancy in the 40,447 record value as being a projection of the *actual* surveyed respondents. I am still a little perplexed however, when selecting the "unweighted" drop down from the custom table view with my given variables --presumably showing actual responses that conform to my restricted universe-- and seeing that whenever I download the CSV I will still have "blank" fields for my OCCP_RC1.Does anyone have any insights into why an unweighted selection might still contain observations that don't conform to the variables I have selected? Is there a way to ensure the records in a custom data set that is unweighted have complete responses? I can imagine having small degrees of missing data to reconcile in an outside statistical package is not uncommon, I'm just getting a substantial amount of blanks in the occupation field and have a hard time appreciating why they would show in the unweighted, restricted universe on the table view to begin with.Thanks for any feedback,Mary