In case this helps anyone:
When down to 2 ACs that weaken an argument to 2 different degrees, pick the one that matches the degree of the conclusion.
When down to ACs that provide sufficient and/or necessary conditions that could serve as the missing link (SA), pick the one that clearly triggers or fails something. If it's a mystery, that doesn't help at all.
Don't second guess yourself when only one AC is right. If all other ACs have been confidently eliminated, flagging that question will only cost you valuable time.
When down to 2 ACs that both mention the key word or concept you know will be in the correct AC, only one is in precisely the right context. Make sure the key element is performing/describing the correct thing.
If you're confused when piecing together a list of facts, some with numbers, some with %s, give the situation real round numbers and apply them to the contending ACs. Don't mistake many for most. Many could be some, which could be a different subset from some other some. (Some historians claim X, many historians are wrong - do not assume overlap.) Some can mean just one.
When looking for a NA in an argument that strikes you as just plain weak, say to yourself, "Within the universe of this shitty argument, which AC points out something that matters, something that absolutely has to be true or else the shitty argument has no leg to stand on in the first place?"
When 2 ACs have the proper conditions to satisfy what MBT, pick the one that matches the stimulus in terms of what is sufficient versus what is necessary. Don't get turned around by the language. What is literally required? Put everything methodically into S->N. Don't overthink.
Parse out the conclusion of convoluted arguments. Sometimes it's just stating that an action will lead to a goal, the NA is that it's possible for said action to lead to said goal.
With parallels, remember sentence order NEVER matters and logic order ALWAYS matters. Be sensitive to distinctions such as "any" versus "one instance."
When there's no obvious explanation for a phenomenon in a RRE, look for an AC that would push one element of the equation in the particular direction that would provide an alternative explanation of the phenomenon. Do NOT give in to bringing in outside bias (such as generic costs less than brand names).
Don't let ACs bait you into "attacking" or "rethinking" a premise - you must assume all premises are completely true no matter what. Period. You are only trying to attack the manner in which the premises "prove" the conclusion. Never pick an AC that merely restates a premise. That's not even good enough for PSA. It does NOTHING.
Never settle for, pick or eliminate an AC you don't understand. Never help an AC out and try to make it fit the mold of a particular flaw. ONLY pick it if it makes total sense.
Be sensitive to WHILE as a conclusion indicator. While X (concession), really Y (conclusion).
Argument parts sometimes can be assumptions or denials of assumptions. Label them as you go.
Don't assume the exact same number of people need to be tested in an experiment. Pay attention when a stimulus starts to compare apples to oranges (or bone samples to blood samples).
Be sensitive to subtlety. (Saying it's wrong for a country to diminish prosperity isn't the same as saying it's wrong to hinder the growth of prosperity.)
If you're spending too much time stuck between 2 ACs, SKIP, read again on Round 2 with the rest eliminated.
Always bear in mind that just because someone claims, says, believes, thinks something does NOT make it one of the things in the stimulus that we accept as true. Accept what the author says. Everything else is suspect.
Woo-hoo! Only one more to go!