The Planning Cycle

I’ve always been a good test taker. Being able to take tests well and know what the test givers are looking for is a skill I’ve spent many years developing, to the point its almost an art form. 12 years of public school and my first year of college have thought me that not only is good test taking important, it can be even more important than actually knowing the subject; I have recently taken several tests on subjects I know next to nothing about, and manage to get passing grades. I’ve got tons of tricks to help pass tests, I rarely get below a 55% on any test, no matter how little I know about the subject.

The secret is what I like to call meta-testing. From the greek word meta, meaning beyond, its the concept of taking it a step beyond the test, knowing about the nature of tests enough to decide the right answers, even if you don’t know how to get them or why they’re right, there are simple processes to create the right answer out of nothing.

Over the next period of time (weeks, months, years even, however long it takes) I’ll be discussing the subject of meta-testing. I’ve gained immense skill in it, I can walk into a multiple choice test on any math-based subject (physics, chemistry, any form of math) and get at least a 55%, a long answer test, in the same fields, I can usually get at least a 40%, which can easily be raised to 60% with an hour or so of study time and a knowledge of what will be tested.

The secret is meta-testing, tests have become so common in US schools that they almost all resemble each other. They follow similar rules, there grading scales are similar, and the secrets are similar. If you can beat one, you can beat them all.

The first major principle of meta-testing builds on a time honored tradition of students: Christmas Treeing. Its a common practice for students, when presented a Scantron test on a subject they don’t know, to perform what is referred to as Christmas Tree Answering, or just Christmas Treeing. Since Scantron tests are multiple choice tests, with either 4 or 5 possible answers, its a simple matter to make a pattern across the page, and a Christmas Tree’d test is likely to get either a 20% or a 25%, significantly better for your average than a 0%.

The problem with Christmas Treeing is that it is done regardless of the test writer, it does not apply any of the core principles of Meta-Testing. It is still a form of Meta-Testing, in that it goes beyond the test, but it does not take the core features of Meta-Testing in, and can be improved by the application of the three most basic tenets: The test writer is predictable, you know something about the subject matter, and questions are designed to have answers.

Multiple choice tests are great. They give you the answer. They actually tell you what the right answer is, you just need to decide which one is actually right. Since the question is designed to be answered, and you have the right answer sitting right in front of you, it only takes a little bit of a stretch to decide which one is actually correct. Look at all parts of the question, and make a list of all the information given to you; its usually a lot more than it appears at first glance, and can lead you to the answer. I’ll go more in depth about meta-answering later, right now lets just assume you’re able to reason out a handful of questions, lets say 20% of the test, a mere 10 questions on a 40 question test, and we’ll say that at least 75% of those are right (a statistic that has held true on almost every occasion I’ve tested).

This means you’ve got 7 or 8 correct answers, more than enough to make a decent analysis of the test writer’s preferences. The problem with Christmas Treeing is that it is usually done by putting ABCDABCD on the test, answering in order. While you have a statistical 25% chance of getting any single answer right this way, its very easy to have the wrong pattern. One very important thing is that test writers try and make random tests answers, to prevent people from guessing the pattern and building on that. That means they do a couple of things, the most important being an avoidance of regular patterns. Rarely will a test have a string of ABCD in a row, and if they do, it won’t occur more than once. The Christmas Tree pattern is one usually avoided, and strings like that will not occur often enough to make it worth while.

Its not too much trouble for a test writer to completely avoid having a pattern in their answers, at least not an obvious one. One thing they can’t do, however, is not use a letter. Very, very rarely will a test writer completely neglect a certain answer choice throughout a test, since it pops up in the mind as a pattern, which they want to avoid. Despite wanting to avoid not using an answer, they have a tendency of favoring one answer more than others. For 4 answer tests, that answer is usually C, answering straight down a test will often score you at least a 30%, for the 5 answer tests its usually B, which will net you a 25%. Its not a major improvement, but it is definitely more than what you started with, which is what meta-testing is all about.

Lets go back to those deduced answers. You’ve got 7 or 8 questions that or correct, a decent enough sample to get a guess at the writer’s preference. Take the answer most used in the 10 questions you have, and answer it for all the questions you don’t know. You’ve got significant chance at error, but its a whole lot better than nothing. Even with the error, you’re looking at almost 40% correct answers on a 4 question test, or a 38% on a 5 question, as long as the tendency towards a single answer choice holds true, which it always will. It isn’t much, but its something to build off of, and it will keep you well within passing ranges for the class.