Here's Kazdin's summary in Slate ...
How to really change your kid's behavior. - By Alan E. Kazdin - Slate MagazineIf you've done any dog training lately you'll discover the same approach is very popular with canines. No more yanks on the training collars, effusive praise and food rewards for correct behaviors. I was very skeptical of that approach for our dog Kateva, but I must admit she's an awfully sweet and reasonably well behaved 1.5 year old.
... You begin by deciding what you want the child to do, the positive opposite of whatever behavior you want to stop. The best way to get rid of unwanted behavior is to train a desirable one to replace it. So turn "I want him to stop having tantrums" into "I want him to stay calm and not to raise his voice when I say no to him."
Then you tell the child exactly what you would like him to do. Don't confuse improving his behavior with improving his moral understanding; just make clear what behavior you're looking for and when it's appropriate, and don't muddy the waters by getting into why he should do it. "When you get mad at your sister, I want you to use words or come tell me about it or just get away from her. No matter what, I want you to keep your hands to yourself."
Whenever you see the child do what you would like, or even do something that's a step in the right direction, you not only pay attention to that behavior, but you praise it in specific, effusive terms. "You were angry at me, but you just used words. You didn't hit or kick, and that's great!" Add a smile or a touch—a hug, a kiss, a pat on the shoulder. Verbal praise grows more effective when augmented via another sense.
If you don't see enough of the desirable behavior, then you can work on it using simulation play. Wait for a peaceful moment and then propose an exercise. "Let's see whether you can stay calm and just use words when I say no to you. I'm going to say no—remember, this is just pretend—and you stay calm, OK?" You can even switch roles as part of the game. Most kids delight in playing the parent and saying no to the parent playing the child.
Your objective is to arrange for as much reinforced practice as possible, which means you want your child to have many opportunities to practice doing the right thing and then be reinforced in the habit by receiving rewards. Your praise is the most important reward, but you can also add little age-appropriate privileges (staying up for 15 more minutes before bedtime, choosing the menu for dinner), goodies (little five-and-dime gadgets for younger children, downloads or cell-phone minutes for older ones), or treats. And, yes, you reward successful let's-pretend simulation sessions, too. This won't go on forever. A brief but intensive period featuring lots of reinforced practice, often somewhere between a couple of weeks and a month, can make long-lasting or even permanent changes in a child's behavior.
Going ballistic never helps, but explanation aimed at improving a child's understanding can actually play a useful part in this approach. When combined with reinforced practice, explanation has been proven to speed up the acquisition of behavior. So, yes, go ahead and explain why it's important to show respect to parents or to play nicely with others. The understanding your child achieves will resonate with the experience of doing the right thing and being rewarded for it. The deep, nuanced science on this topic all points to reinforced practice as the key, but the greater understanding that comes from explanation is an optional add-on that can help good behavior develop more quickly.
Kazdin doesn't mention it in his summary, but this conditioning approach comes from animal training, where it's coordinated with with the use of "extinction". As Amy Sutherland discovered in her immensely popular NYT article, it works for adults and killer whales alike. For a complete guide to child rearing, I'd suggest the above summary in Slate together with my summary of Sutherland's article. You can read the two within ten minutes and be ready -- for neurotypicals anyway!
Of course I shall venture my opinion on this topic, I would say I am as expert on the childhood portion as most. For the top 80% of the temperament curve any non-abusive approach to child rearing will more or less work for almost every child. So for most family's Kazdin's approach, or any reasonable approach consistent with the temperaments of family and child, will work well enough.
For much of the next 18% I will guess that something like Kadzin's approach, augmented by basic extinction methods and some loss of privileges for various sins, is required for a reasonably successful childhood. (As to adolescence I have no experience and no predictions.)
Most children with high functioning autism fall into this 18% range.
For the most challenging 2% ... well, a miracle is helpful. In my reading list post I recommend Greene, The Explosive Child combined with extinction, positive direction, and rewards. I agree with Kadzin that for these children punishments have little impact on behavior, but some symbolic measures they may be needed to provide a sense of justice (and comfort) to both the sinner and the injured.
We've recent added to this with my wife's version of "applied behavior analysis". She did the reading on this so I can't provide references, but there's psychological judo involved. With the most challenging 2% one very quickly runs out of positive incentives. The range of behavioral interventions needed to provide health, safety and education are so large that all practical measures of praise and reward are piled so high as to be irrelevant. It's the behavioral analysis of "printing money"; the currency is so devalued that a wheelbarrow is needed to buy a loaf of bread.
The trick then is to "revalue the currency". All privileges are removed, and even things like "sugar on cereal" are redefined as privileges. Rights are limited to nutrition, shelter, education and love. Then one seeks actively for all signs of good behavior, no matter how small and new, and rewards them with praise and concrete measures. We've returned to using a sticker chart of about 7 columns by seven rows. Each act is rewarded with praise and a sticker, a column gets a bigger-than-average privilege (ex: 20 minutes on the Wii) or a quarter (the sibs prefer the money, of course the same system must apply to all), a full chart gets a treat or $2.00.
As of today, then, we apply a mixture of Greene, Kazdin (before we knew of him), reward/extinction, Applied Behavior Analysis modified, symbolic punishments in the cause of a child's sense of "justice", as much exercise as a child will tolerate, medications, and many leaps of imagination and experimentation. The experiments usually fail, but some will work, at least for a time, when retested at a later date.