ncrncornell / ced2ar

Comprehensive Extensible Data Documentation and Access Repository (CED²AR) was designed to improve the documentation and discoverability of both public and restricted data
https://www.ncrn.cornell.edu/projects/#Metadata_Standards_and_Tools
Other
6 stars 1 forks source link

Compare variables - values "reports" inconsistent or misleading #14

Open sandaionescu opened 7 years ago

sandaionescu commented 7 years ago

When I compare two variables (V520121 and V520170 from ANES TS 1952) I get a "categories report" for each variable saying "4 total" and "1 total" respectively. But this is misleading. It is in fact a report of how many values are labeled and does not indicate how many valid values are actually present in the variable. Variable V520170 is numeric (continuous) and has in fact 130 values - and this fact is correctly documented in the "values report" appearing on the individual variable display page.

sandaionescu commented 7 years ago

Another example: https://demo.ncrn.cornell.edu/ced2ar-web/compare The comparison page states that "The selected variables do not have values" but this is NOT true. They do have values, but the values are not labelled (there are no value labels in the documentation). I think in such a case there should be a statement about the number of values on each variable, and the fact that they are unlabelled.

CSimmer commented 7 years ago

I confirmed sandaionescu's initial issue comment with the three screenshots below.

The Values section in the Comparison Overview page (https://demo.ncrn.cornell.edu/ced2ar-web/compare) is showing a comparison of labels (/var/catgry/labl), not values (/var/catgry/catValu). The Total counts are for labels too.

Column 1 has a heading of Values, but actual values listed are Label values. The totals in Columns 2 and 3 are counts for Labels with values.

CSimmer commented 7 years ago

Screenshots: 1) Comparison Overview - https://demo.ncrn.cornell.edu/ced2ar-web/compare ss comparison overview 2017-03-15

2) Variable Name V520121 - https://demo.ncrn.cornell.edu/ced2ar-web/codebooks/icpsr072134/vars/V520121 ss var v520121 2017-03-15

3) Variable Name V520170 - https://demo.ncrn.cornell.edu/ced2ar-web/codebooks/icpsr072134/vars/V520170 ss var v520170 2017-03-15

CSimmer commented 7 years ago

re: Another example: https://demo.ncrn.cornell.edu/ced2ar-web/compare sandaionescu can you please post the screenshots for this example?
The URL you posted does not give me enough information to pull up the pages you are talking about.

CSimmer commented 7 years ago

I don't have the business expertise to determine the extent of the issue.

A) Is this report is always off? OR B) Is this report off under certain conditions?

Stated another way, which is correct for this Comparison Overview page? A) You always compare values and you never compare Label values. B) Sometimes you compare values. Sometimes you compare Labels.

sandaionescu commented 7 years ago

the second example: image variable V520020A: https://demo.ncrn.cornell.edu/ced2ar-web/codebooks/icpsr072134/vars/V520020A variable V520019A: https://demo.ncrn.cornell.edu/ced2ar-web/codebooks/icpsr072134/vars/V520019A

sandaionescu commented 7 years ago

my answer to your question above: you always compare values. if they are labeled, they should appear with labels. if they are unlabelled, that should be somehow made apparent.