One problem with a test like this, giving you an alignment, is that the D&D alignment system has gone through revisions of meaning over the years. And the source material has very different things to say about Law vs Chaos for example than ever made it into publication.
And of course there's the perennial contentiousness around what certain alignments mean.
And this.A secondary problem is that these personality tests are skewed by the preconceptions of the test maker. The language itself, and the words used in the test, skew the takers perception in different ways whether he's a native or non-native speaker (or if the test was translated!). The specific questions asked may or may not be understood in the way the test maker intended. The answers given may not be exactly what the test maker had in mind. The restrictiveness of a multiple-choice test is compounded by the test maker's choice in potential answers. The very order in which the answers are provided will subtly shift answers over the whole population of test-takers.
A tertiary problem is in the daily variances in our answers. One is not a substantially different person day to day, yet the test may yield different results. Are you tired? Drunk? High? Hungry? Need to pee? All of these may or may not influence your decision-making.
Finally, mapping the outcomes of one bad test with the outcomes of another bad test, which are only superficially related, and which are re-imagined to match better with each other, is of limited value.
Specifically relating to the D&D quiz, I take exception to the statement that "lawful people tend to work at secure jobs -- this doesn't mean that working a steady job makes you lawful". Practicing being good, or bad, or sloppy, or neat, or cowardly, or courageous, will actually change your decision-making process in the future. What we experience changes us, sometimes in fundamental ways. Just look at most people who spend time in the military. They exit with certain habits relating to hygiene and sleep schedule common to members of their unit. That the test maker succumbs to such a simple mistake suggests that deeper, more profound disasters of methodology are possible.
The D&D quiz also seemingly gives you a numerical value in some way for each answer, then averages them. In this way, answering "I like to give money to everyone" and "I like to eat babies and kick puppies" would average you out at Neutral when clearly there is a more complex conclusion.
One specific question in the D&D quiz asks:
A powerful but corrupt judge offers you wealth if you'll testify against a friend. Do you:
1: Condemn your friend and take the money
2: Take the money and testify, but try to keep your testimony ineffective
3: Refuse the offer and refuse to testify
4: Testify on your friend's behalf, no matter the consequences
Nowhere in the answers is "testify truthfully regardless of loyalty to the friend or the judge", "bring a recording device to get evidence and turn in the judge", "take the money and murder the judge", "murder the judge immediately upon his offer", "break your friend out of jail and help him escape", or the strange choice "refuse the money but testify against your friend anyway because you're petty and cruel for its own sake". All of the people who would have answered any of those get left out in the cold, and instead of showing up on the population graph as what they really are, they are crammed into one of the other categories.
I've taken enough of these to understand that they're meaningless. I haven't seen a personality test that's any good unless it focuses strictly on one aspect of your personality, and even then the test is extensive and the possible outcomes have many millions of permutations.