The "more often than not" locution is useful for the simple reason it's a "short and dirty" way of describing an aspect of reality without having to resort to cold, hard statistics.
Now if a colleague in a math-driven, statistic-oriented work environment asks you how often Joe Public votes in local elections, she expects an accurate statistic from you, and not a "more often than not" answer. In "normal" conversation with your neighbor, however, he might, out of curiosity, ask you
How often do you wind up washing your car on Saturday?
In answering your neighbor's question you are permitted to say,
More often than not.
Very rarely would a neighbor say in return,
I don't agree with you. I've been keeping track of what day of the week you wash your car for the past two-and-a-half years. In point of fact you have washed your car on Saturday only 45 percent of the time, on Sunday 48 percent of the time, and on various other days of the week the remaining seven percent of the time. Clearly, you do not wash your car on Saturday more often than not. What a liar you are!
Only severely neurotic, anal-retentive types are going to bust your chops about not being precise about trivial things such as what day of the week you wash your car. Normal people don't ask you about the statistical probability that you'll be washing your car on Saturday this week. If you tell them, "More often than not, I wash my car on Saturday, so chances are good I'll be doing so this Saturday," then you'll likely believe them.
As an interesting aside, in American jurisprudence the burden of proof in a criminal case is different from the burden of proof in a civil case. Before the jury deliberates the fate of the accused in a murder trial, for example, they are instructed by the judge to find the defendant guilty only if the evidence proves the defendant guilty beyond a reasonable doubt.
The judge does not, I assure you, say to the jury,
If your level of doubt is between one percent and 11 percent, then your verdict should be "guilty," but if your level of doubt is 12 percent or higher, then your verdict should be "not guilty."
"Beyond a reasonable doubt" cannot be decided on a sliding scale. A reasonable doubt is just that: reasonable, and not mathematically precise. An unreasonable doubt in a fairly clear cut murder trial, for example, would be if a juror insists that even though there were several witnesses who swore under oath they saw the defendant take aim at the victim and shoot him six times in the chest from a distance of less than 10 feet, this juror feels strongly that the real shooter, an expert marksman, was perched on the roof of an 80-story building, and he's the one who killed the victim, not the defendant! This, in spite of there not being one single piece of evidence that there was a shooter other than the defendant. Now that's unreasonable.
On the other hand, in a civil trial the burden of proof is quite different. In order for a judge or jury to find a defendant guilty, that guilt must be proved by a preponderance of evidence. Despite the inappropriateness of using statistics, even in such an important and formal setting as a civil trial, judges have been known to say,
The preponderance of evidence for either guilt or innocence can be as little as 51 percent to 49 percent in favor of one or the other. As long as you find the evidence is more in favor of guilt than innocence, then you have to find the defendant guilty. If on the other hand, as long as you find the evidence is more in favor of innocence than guilt, then you have to find the defendant innocent.
We can forgive the judge, I suppose, for giving the impression that two percent more (or less) constitutes a preponderance of evidence toward guilt or innocence. In a sense, the judge's attaching a number to evidence is unwise since jurors essentially weigh evidence, not on a literal scale but on a mental scale with which, using their own criteria for attaching weight to evidence, they attempt to determine guilt or innocence.
As a rhetorician, I must insist that in both a criminal trial and a civil trial, the role of persuasion is--or at least can be--equally important. In the hands of a bright and persuasive lawyer, if a crucial piece of evidence or a line of argument is packaged in such a way that it becomes compelling, moving, and yes, persuasive, that evidence can tip the balance in favor of guilt or innocence.
In other words, the packaging of proof can sometimes have an inordinately large effect on how evidence is processed by a group of supposedly neutral and fact-finding jurors. Statistics are rarely by themselves persuasive. By using a "more often than not"
locution at the right time and in the right way can, more often than not, prove to be more powerful than cold and bare statistics.