Nope, no change in methods! This is a known issue that is currently under investigation. The PTDOBMM and PTDOBYY fields have the similar issues. For the time being, I would recommend using the earliest available information for each participant within those fields.

All participant demographic data is based on self-report, so it is possible (although it seems unlikely for information as basic as year of birth) that participants answered the questions differently in ADNI4.

In the case of PTEDUCAT, there could also be issues arising from inter-phase variance in the procedures for recording that information. The in-clinic protocols and procedures manuals for each phase (available on the ADNI homepage and/or on the IDA) can offer some more detailed information on how participant responses have been categorized, and note that in some cases site staff may have recorded an exact number of years rather than abiding by the coding scheme.

Certainly! it should be in the procedures manuals for each phase, which can be found on the IDA. In the study files interface, use the menu on the left to navigate to 'study info' -> 'Protocols and CRFs', and they should be right there - alternatively, typing 'procedures' into the search bar on the top-left of the interface should pull them up.

The ADNI1 and ADNI4 procedures will probably be most relevant here; I believe the information on education coding is on page 105 of the ADNI1 manual (by page number), and it should be organized similarly in ADNI4.

Thanks for pointing me to the correct documents for the PTEDUCAT coding. Upon reviewing them, the coding schema appears to be the same for ADNI1 and ADNI4, except for one word:
ADNI1: Code 18 years even if MA/MS completed in one year.
ADNI4: Code 18 years only if MA/MS completed in one year.
This one-word difference doesn't make sense -- maybe a typo?

Also, the ADNIGO,2 and 3 procedures manuals do not include a version of the education coding table. Does that indicate the ADNI1 coding table was used for those phases as well?

And if the coding schemas are the same across all phases, does that mean the value differences are all caused by site staff not following the coding scheme between phases?

That is a strange wording change - agreed that it is most likely a typo. The ADNI1 procedures generally take precedent when there are no new instructions, so it is safe to assume that the coding scheme for those phases aligns with the ADNI1 coding scheme.

We've done some digging around about the root cause of the value differences, and it appears that the ADNI4 data entry manual (the internal document used as guidance by site staff when filling out eCRFs) contradicts the procedures manual, stating "If the exact number of years is unknown but level of education is known, convert into estimated years as follows:" - followed by the coding scheme from the procedures manual.

Conversely, the instructions in the pen-and-paper CRFs for ADNI4 instruct clinical staff to refer to the procedures manual for instructions on how to report education, which would give the coding scheme. Depending on which document site staff are referencing, they could run into either set of instructions.

So, at least for ADNI4, there appears to be some inconsistency in how instructions were passed along to site personell. The CRFs and data entry manuals for historical phases are currently being dug up to check and see if that same disconnect has existed since the beginning of the study, and we should be able to offer some more specific guidance on dealing with the field once those are found and reviewed. My best guess is there's been some similar level of historical inconsistency in direction, since there are values in the PTEDUCAT field that fall completely outside of the coding scheme.