Same year, same author, no disambiguation in bibliography if FirstName field is empty

skjpope · January 30, 2009

There is a problem in Zotero (1.0.9) that occurs in generation of the Bibliography (and citations) that is especially noticeable with Harvard and similar styles.

If you have multiple documents -- all in the same year, and by the same author -- if the author's first name is empty, then the citations will appear as the following:

OECD, 2001. OECD country note: early childhood education and care policy.

OECD, 2001. Starting Strong: Early Childhood Education and Care.

Similarly the citations will both appear as (OECD, 2001; OECD, 2001)
Clearly this is wrong. They should appear as the following:

OECD, 2001a. OECD country note: early childhood education and care policy.

OECD, 2001b. Starting Strong: Early Childhood Education and Care.

with the citations being
(OECD, 2001a; OECD, 2001b)
Interestingly, the problem only occurs when the FirstName field is empty. A workaround is to put an empty space in both of the firstname fields of the documents, and then it formats correctly.

This seems like such a simple problem to reproduce, and should be rather trivial to fix.

Note: I would have logged a bug in Bugzilla or similar, but it seems to be limited to core developers, so I hope this get addressed in the initial release of 1.5 (or 2.0, whatever you call the next release after 1.0.9).

bdarcus · January 30, 2009

While this is sort of a bug, you're not using the field correctly. "OECD" isn't, after all, a last/family name. What happens if you change the contributor field to the field (click on the icon to the right of it)?

skjpope · January 31, 2009

I assume you meant change the field to a single-value field? Well, that didn't make a difference either, though I appreciate the tip!

Curiously, I experimented with it some more and I believe my workaround was not-quite-right, and the problem somewhat more subtle than I thought.

In the bibliography, there were 4 documents by the same author, appearing so:

OECD, 1998. Document One
OECD, 2001. Document Two
OECD, 2006. Document Three
OECD, 2001. Document Four

That's right - "Document Three" of 2006 is interleaved between the two "2001" documents.

Frankly, I'm at a bit of a loss. I've checked that the two "2001" documents both have exactly the same text for the author and dates, and all four have exactly the same authors (and now use a single field).

If I change only "Document Three" to use LastName="OECD" FirstName=" " (that's two spaces but I think one would suffice), then I get the following output:

OECD, 1998. Document One
OECD, 2001a. Document Two
OECD, 2001b. Document Four
OECD, 2006. Document Three

Which is what I wanted. Go figure.

I'm at a loss to understand what is driving the mis-ordering (which is clearly the problem, and again goes to the reference sort key) but fudging it works. Is there anything I can provide that you think may throuw light on the exact causes?

adamsmith · January 31, 2009

I can confirm this issue - it's the same for APSA and ASA, but things works correctly in Chicago author-date as well as APA.
If you want, browse around a bit in
chrome://zotero/content/tools/csledit.xul
to see what's driving this - else you'll have to wait for someone else to take a look.

skjpope · February 1, 2009

Bingo! I found a proximate cause for the error. I have no idea how to fix it though. The author first/last name workaround is unnecessary and will not reliably solve the problem (as we shall see). To replicate, perform the following steps:

1. Create three entries in Zotero as follows:
OECD, 2001. OECD note: early childhood education and care policy in Australia.
OECD, 2001. Starting Strong: Early Childhood Education and Care.
OECD, 2006. Starting Strong II: Early Childhood Education and Care.

The important part here is to ensure that there are colons ":" in the title of the two last references. As we shall see, Zotero is using the colon in a way that is not expected by the user!

2. Open the Reference Test Pane (chrome://zotero/content/tools/csledit.xul) and select the three references. [Thanks Adam]

3. Select the Harvard Reference Format 1 from the list of available styles. Zotero will format these as:

OECD, 2001. OECD note: early childhood education and care policy in Australia.
OECD, 2006. Starting Strong II: Early Childhood Education and Care.
OECD, 2001. Starting Strong: Early Childhood Education and Care.

4. Proof of weirdness (bug?) can be seen by replacing the colons in the titles of the two "Starting Strong" references. Change them to dashes ("-") so they read:

Starting Strong - Early Childhood Education and Care, and
Starting Strong II - Early Childhood Education and Care.

5. Refresh the Test Pane and notice that they now (correctly) format as:

OECD, 2001a. OECD note: early childhood education and care policy in Australia.
OECD, 2001b. Starting Strong - Early Childhood Education and Care.
OECD, 2006. Starting Strong II - Early Childhood Education and Care.

Note: The same result can be achieved without altering the references by the (presumably undesirable) action of removing:
<key variable="title"/> from the sort element

    <sort>
      <key macro="author"/>
      <key macro="year-date"/>
      <key variable="title"/>
    </sort>

so that it reads instead:

    <sort>
      <key macro="author"/>
      <key macro="year-date"/>
    </sort>

Conclusion: That's about the limit of what I can do without going and debugging the back-end Zotero code. Does anyone have any better idea about what is causing the problem from this description? ... And if so, can someone please fix and test, perhaps?

adamsmith · February 1, 2009

there really doesn't seem to be anything wrong with Zotero - this is a problem of the styles.
What I find weird is, that you find
<sort>
<key macro="author"/>
<key macro="year-date"/>
<key variable="title"/>
</sort>

in my Harvard (e.g.) it doesn't have the "year-date" in the first place -
but if I insert it it doesn't work, so something appears to be wrong with the macro.

What _does_ work is instead using <key variable="issued"/>
so that the thing reads:
<sort>
<key macro="author"/>
<key variable="issued"/>
<key variable="title"/>
</sort>

which will only be problematic if you have lot's of refernces with a precise date - mine all just have the year, so this solves it for me.

Can someone in charge of the styles maybe have a look and (presumably) correct them?
(@skjpop - for your purposes I'd suggest you install the modified style as a separate style with a new ID on your Zoter - else it will be overwritten with each update.
Here are instructions how: http://forums.zotero.org/discussion/5104/modifying-word-plugin-using-journal-abbreviation-instead-of-publication-name/#Item_2 )
)

Rintze · February 1, 2009

Out of personal interest I would like to broaden the question a bit: how does sorting using macros as keys exactly work? Are all the variables in the macro used for sorting in the order they appear in the final output, or just the first one? I never encountered any documentation on this (the schema and dev-pages are quite limited when it comes to sorting).