Readers expand the discussion on whether the Dish should just bump up its $19.99 subscription price to 20 bucks:

Apparently it's is also effective to drop the $ sign on prices.

Another reader:

According to at least one study on the penny (pdf), the true purpose of having prices end in .99 is not to "trick" consumers into believing that an item costs less than it really does.  Rather, stores use these prices to deter theft by employees.  If something is priced with an even amount, say $20, the consumer is likely to pay with exact change.  It's fairly easy for a cashier to just pocket the $20 bill without ever ringing up the purchase.  If an item is priced at $19.99, the cashier will probably have to make change.  It'll be pretty obvious if she takes a penny out of her pocket, so she will have to enter the purchase into the register.

Thought this theory might be of interest to you.  And, if true, it implies that it is pretty pointless to price something at $19.99 over the Internet.

Another:

Steven Landsburg offered this explanation of 99-cent pricing in his 1993 book, The Armchair Economist: Economics and Everyday Life:  

The phenomenon of "99-cent pricing" seems to have first become common in the nineteenth century, shortly after the invention of the cash register. The cash register was a remarkable innovation; not only did it do simple arithmetic, it also kept a record of every sale. That’s important if you think your employees might be stealing from you. You can examine the tape at the end of the day and know how much money should be in the drawer.

There is one small problem with cash registers: They don’t actually record every sale; they record only those sales that are rung up. If a customer buys an item for $1 and hands the clerk a dollar bill, the clerk can neglect to record the sale, slip the bill in his pocket, and leave no one the wiser.

On the other hand, when a customer buys an item for 99 cents and hands the clerk a dollar bill, the clerk has to make change. This requires him to open the cash drawer, which he cannot do without ringing up the sale. Ninety-ninecent pricing forces clerks to ring up sales and keeps them honest." (http://books.google.com/books?id=qTBgMMxeJ5IC&pg=PA19)

It's a cute theory, but even Landsburg notes some holes. What about states with sales tax, where an item priced in whole dollars would still require change? And why has it persisted, even after the advent of scanners and electronic registers that record sales even when no change is required?

The question seems particularly troubling to economists, because a strong consumer preference for $19.99 over $20 would seem to wreak havoc with the notion of consumers as rational actors. Nevertheless, over the last twenty years, a growing number of empirical studies and experiments have confirmed that 99-cent pricing actually works.

One explanation is for that success is that 99-cent pricing signals to consumers that the item is a bargain, because it's most often used on bargain goods. Consumers aren't buying these items because they mistakenly think that missing penny makes them meaningfully cheaper, the theory suggests, but because they've been flagged as discounts.

A second is that consumers – or at least, some significant subset of consumers – process prices in two discrete units, before and after the decimal. Some people, perhaps don't find it worthwhile to invest the time and effort necessary to pay close attention to prices, and merchants calibrate their pricing to take advantage of their inattention.

You've actually been running your own experiment, with interesting results. There is, practically speaking, no real difference between those sums. At the moment, all contributions are voluntary – even the leaky paywall hasn't yet been put in place – and it's hard to believe that readers who will voluntarily hand over $19.99 would begrudge you the additional penny. And although subscribers are prompted with the $19.99 minimum, they have to manually enter the amount, and it's easier to just put down $20.

Indeed, that's exactly what nearly two thousand of them did. The only catch? Twice as many took the trouble to type out $19.99. I'd say that's fairly solid data in support of the theory that $20 just feels like more money – and that many readers are more likely to take the plunge and invest in the site if that threshold isn't crossed.

Previous parts of this thread here and here.