The magnetic poles do indeed flip flop every so often. However, the reversals of the poles is distinctly aperiodic. There are intervals of geologic time measured in the millions of years during which no pole reversals occurred. (on average). If that means anything, the poles reverse on order 10⁵ to 10⁶ years. The last reversal was about 700,000 years ago.

The cause is harder to describe in a few words. The study of the earth's dynamo is quite complicated. The ultimate origin of the earth's field is due to the fact that the earth's outer core is molten metal and when a metal (a conducting fluid) moves by convective stirring due to heat transfer, a magnetic field gets set up. The reversals can be thought of roughly by this metaphor: Imagine going down to your favorite creek and watching the movements of some eddies. The eddies swirl water in a certain direction (clockwise say) and the eddy itself moves down or up stream. this complicated motion is called turbulent flow. Now imagine the same thing going on in the core but because the core is metal the fluid is electrically conducting. The strength and time rate of change of the magnetic field will be strongly coupled to the flow (the so called hydrodynamics) ; in fact the coupling is intimate in the sense that the flow is both determined by and determines the orientation of the magnetic field

When one actually solves the partial differential equations that govern the flow, it turns out that the orientation of the magnetic field undergoes chaotic binary behavior... that is, it self reverses 'every so often'; the exact timing is non predictable. The subject is called magnetohydrodynamics and its is pretty heavy stuff. Essentially a branch of applied math.

You're right -- the magnetic poles on Earth do reverse, roughly every half-million years or so. The earth's magnetic field is generated by flow of molten iron within the core (molten iron is a good electrical conductor, and its movement generates a magnetic field), so flips in polarity must somehow be caused by rapid shifts in flow. That's the basic idea, but scientists are still trying to workout the details of the movements within the earth core and how these can lead to abrupt changes in the earth's magnetic field.

It is true that the magnetic poles of the earth have flip-flopped. At the present time, the north pole of the earth is like the south pole of a regular magnet (called a "dipole" magnet because it has two poles - a north and a south). Only the magnetic pole is not exactly lined up with the rotation axis of the earth, but is tilted by about 11 degrees, so that it comes out somewhere in Greenland. No one really knows why. We know that the magnetic polarity of the earth has flipped not once, but a great many times in the past, by measuring the magnetism that is contained in the iron particles in rocks.

Let me tell you about ferromagnetism: Iron has the property that it can be magnetized. Here are two examples that you can try yourself: 1) Take some staples or paper clips and leave them on a magnet for a day or so. When you remove them, you will see that they have become little magnets themselves. 2) Wrap a wire around a nail, and strip the ends off the wire. Attach the ends of the wire to a flashlight battery. The current in the wire makes a magnetic field, which magnetizes the nail. The nail can pick up paperclips while it is connected to the battery. (Be careful, because it tends to get hot... can you think why?) At first, when you disconnect the battery the nail will drop the paperclips... but if you leave it connected for along time, eventually the nail will become a magnet itself. This shows that iron can be magnetized.

Igneous rocks are melted inside the earth and are either forced out by a volcano, or eventually get exposed from erosion of the rocks above them. Many igneous rocks contain iron. This iron is not magnetic when it is hot(more than 600 Celcius) but when it cools, the little bits of iron line up with the magnetic field of the earth.

You can measure this tiny bit of magnetism in the rocks (called "remnant magnetism" because it "remains" in the rocks after they cool) in a laboratory. If you carefully measure which way is north when you remove the sample of rock from the ground, then when you take it to the laboratory you can point it in the north direction. When you measure the little bit of magnetism in the rock that remains from when it cooled, you can see if its magnetic north pole points north, south, or some direction in between.

We can tell how old the rock is in several ways:One way is radioactive dating. If we can measure certain radioactive elements, we can tell by the amount that has decayed into something else how old the rock is. Another way is if the igneous rock comes from a layer that lies in between sedimentary rocks that contain fossils. We can date the fossils and tell approximately the age of the igneous rock in between.

The magnetic field reversals were first discovered by observing the magnetic polarity of the ocean floor. The ocean floor is made of a type of igneous rock called basalt, which is pushed out molten from the ridge of volcanoes that goes down the center of the world's major oceans, continually creating new ocean floor. As this basalt cools and spreads away from the center of the ocean, it takes on the magnetic field of the earth at that time.

Scientists from Columbia University's Lamont-Doherty Earth Observatory sailed back and forth across the oceans back in the 1960's, and found that the magnetic polarity (which way the rocks' magnetic fields pointed" flipped back and forth as they went back and forth from shore to shore making their measurements. They did this very carefully, for a number of years, until they had taken "magnetic profiles" across all the world's major oceans(Atlantic, Pacific, Indian Ocean, Sea of Japan, among others).

Since the volcanoes in the middle of the ocean keep putting out lava and the sea floor is spreading away from the ridge of volcanoes, the ocean crust in the middle of the Atlantic is the youngest, and the crust by the coasts on either side is the oldest. The scientists also found that on either side of the mid ocean ridge (as those volcanoes are called, because they run all down the middle of the whole Atlantic) the magnetic "stripes" were mirror images. So in this way they could tell that in the past 200 million years or so the earth's magnetic field has changed directions many times.

The oldest rocks on land are older than the oldest rocks in the oceans, soother scientists who measured the magnetism of igneous rocks on the continents were able to extend the magnetic history of the earth back to about 600 million years.

It is true that the magnetic poles of the Earth switch but the frequency of the switch is quite variable. Each reversal seems to occur over a period of 1000 to 5000 years during which the magnetic field dies down and wobbles around, after which it strengthens again in the opposite orientation (i.e. what we think of as the North Pole is in the south and vice versa).

The period between magnetic field reversals is quite variable, from a few 10's of thousand of years to several millions of years although it seems to average about 200,000 years. Unfortunately, we still don't know very precisely why the magnetic field shifts or how to predict such a shift. The best attempts have been made through computer modeling but this is still research in its early stages and it involves some complex electrodynamics.

A question for you to explore. If the Earth's magnetic field reduces by a factor of 5 during a reversal, how do you think the Earth would be affected?