In all other months the sample sizes match up if you exclude non-interview households from NBER, and ASEC households from IPUMS. But in January 1977 there are 116,974 observations in the IPUMS data and only 99,341 in NBER data?
Which observations do I need to drop IPUMS in order for these to match and than do a merge based on line number?
The 1977 CPS samples with supplement questions also included an oversample of respondents which were not present in the basic monthly file. The 1977 January sample included the Job Tenure supplement (currently being integrated by the IPUMS-CPS for these early years), which is why the IPUMS-CPS sample (drawing from the supplement file) has more respondents than the NBER counterpart (which is the basic monthly file). Unfortunately, the unique household identifiers are not shared between the 1977 January basic and supplement files making linking the two very difficult.
I am sorry I do not have better news.