The SC grid in the 2-4 form form sucks unless you have comps that have the same number of units and rooms. Apartment units are rented primarily on the basis of bedroom count, not GLA, so that makes GLA itself a minor adjustment. So if you have a 400sf 1bd and a 700sf 1bd there will be some difference in rent but it won't be anywhere near the difference between a 400sf 1bd and a 600sf 2bd.
Ibd units generate more less rent than 2bd units, but generally not 1/2 as much, so using price/bedroom isn't usually that great. What I've found does work better *in this region* is price/room. If you look you might notice that a 1bd unit (3 rooms) will rent for about 75% of a 2bd (4 rms) and then when you're making adjustments you're using the price/rm indicator and adjusting that as your unit of comparison. Obviously, this can vary by locale. We have some properties in the high demand areas (like a beach community or a ski town) where the 2bd units actually do come close for renting for twice as much as the 1bd units.
As others have said, tenants are sensitive to location, quality, condition, unit size, off-street parking and such and that gets reflected in the rent, which in turn is what an investor is looking for, even if they live onsite and are using the other unit incomes to offset their own housing costs.
But this is key - when appraising multi-family you do need a solid rent survey and you need to project the expectations of the typical buyers for the subject and all of the comparables on what you think they'll actually do with those rents once they take ownership. This is where merely going through the motions to fill in the blanks is counterproductive.
Unless there's an external limitation on rent increases, such as a rent control ordinance - you can expect that the average buyer for units will raise the rents to market rate in whatever length of time they think they can pull it off. Not just for the subject but for all the comparables. So if you have comps that are partially owner-occupied or have vacant units or have family members or long term tenants with significantly below market leases then you need to attribute to those units what the buyer expectations will be - which usually means forecasting their rents at the market rate if they're not already at or near.
THEN the resulting GRMs or GIMs will tend to be more consistent and usable.
And then you can get to comparing the subject's rent/unit or rent/rm to the others to identify how much better/worse the tenants think those comps stack up against your subject as an indicator to how the sale prices should be adjusted. If on a price/rm basis my subject units generate 5% more income than S#1, the multiplying S#1s indicator by .95 will equalize it to the subject. That will account for the cumulative differences in location, quality, condition, unit size, amenties and such. Rinse and repeat for all the comparables and there will be precious little else for which to adjust because it's already accounted for in the rents.
That's far from the only way to analyze rental units, but it is one way.
