You can use a potentiometer, but this will give you greatly reduced torque at lower speeds, and you'll probably need a pot with a high power rating.
You'll have much better results using PWM, increasing the duty cycle will increase the speed. The best implementation of this would be to use a microcontroller with a pot on an analogue input, and a MOSFET to power the motor via an output pin. Program the microcontroller to change the duty cycle depending on the value of the voltage divided by the pot.
You can use a switch to control the direction if you're controlling it manually, but if you want your microcontroller to automatically control direction then you'll need an H bridge.
First of all resistors aren't used to regulate voltages of any significant consumer.
There are several reasons for that, but the most important ones are that the resistor itself is dissipating all the dropped voltage and consuming power. That will have an impact on the battery life. The second equally important point is that resistors drop voltage but they do not provide voltage regulation! The amount of dropped voltage is dependent on the amount of current that passes through the resistor! So if you have a motor running with no load, the resistor will drop one voltage but when you put load on the motor, the resistor will drop higher voltage (assuming your power source can provide enough power and 9 V batteries aren't the best option here, especially for motors).
You can use a potentiometers and rheostats to obtain variable resistors that will give you different speeds for a motor, but the main problem with them is in general potentiometers are designed to dissipate small amounts of power and when adjusting voltage with a resistor, you'll have large power dissipation on the resistor which makes potentiometers unsuitable for directly adjusting voltages of large loads.
Also note that THERE IS ABSOLUTELY NO WAY TO USE A RESISTOR TO INCREASE VOLTAGE!!! This one is important! I'd not go too much into physics behind that here, but I think that the idea is basically equivalent of truing to produce oil by pushing your car backwards.
On the other hand, the linear voltage regulators behave like a special type of resistor which automatically adjusts its resistance (within certain range) so that the output voltage is (more or less) constant. They too dissipate the extra voltage as heat and aren't a good solution for large loads especially on battery power. Voltage output of linear voltage regulators can be controlled (on some regulators) and you can use them to control speed of a motor.
Now about the voltage drop using H-bridge: It's a bit more difficult to explain, but the main point is that when analyzing voltage coming to a motor you have basically two voltages: voltage in a single moment of time and average voltage over some time. Usually with H-bridge circuits, you're providing full instantaneous voltage to the load, but you're constantly turning the load on and off. This happens so quickly that the average voltage will look like a voltage lower than the input voltage and that way you can provide speed control for a motor by changing the time during which the motor is provided full voltage and time during which the motor has no power. The main advantage of that approach is that you are wasting very little power for voltage regulation. The transistors in an H-bridge will usually have low on resistance and when they're on, they are fully on and when they're off, they are fully off, so only little power is dissipated by them.
Another way of getting the right voltage is to use a switch-mode regulator. They are often more complicated and require more components or are more expensive if they come in same form factor as linear regulators. The good sides however make them very interesting. They can (depending on specific device) decrease or increase output voltage compared to input voltage and they waste very little energy as heat when doing so. They produce more noise on the output than linear regulators too. Anyway as far as motors are concerned and as far as I can see, there is no major benefit to use of switch-mode regulators compared to say PWM, since motors can survive short exposures to higher voltages with no problems at all (as long as the time is short enough so that the current is below the maximum rated current for the motor).
Now about that PWM motor controller: In general you'll need at least two wires to control it: ground wire to provide reference or ground voltage and a signal line. So if you're going to use an Arduino, you'll need to connect the negative sides of the controller's power supply and the Arduino together and you'll need to find the controller's signal line and drive it with PWM from Arduino.
Next, I see you mentioned stepper motors. They are usually controlled not by traditional H-bridgees but by stepper motor controllers. Basically a stepper motor has several inputs which control individual windings on the motor. You need to provide power to each winding in turn so that the motor will rotate. The speed is controlled usually not by voltage directly but by the amount of time each winding is energized. So to increase the speed of a stepper motor, you "simply" need to switch between the windings faster.
Now a little bit about the 9 V batteries: They are in general a poor choice for running any significant consumer because they are usually constricted by having 6 1.5 V cells connected in series. The cells themselves are very small and have low capacity which limits the capacity of the entire battery. This also affects the maximum current the battery can provide and since motors are significant consumers, the lifetime of a single battery will be very short. Some better options are to get say 6 AA (or C or D) cells and connect them in series for much higher capacity and higher maximum current. Another option (which could be much more expensive if you don't have the appropriate tools) would be to get a 12 V battery, such as a car battery and then recharge it or to get a 3 cell lithium-polymer battery or to get 6 cell NiMH battery.
Best Answer
Just a single switch is not enough. Actually, two switches is not enough. You need a memory element, such as a relay, plus a switch at each end of your weight travel.
The following will do what you need. It consists of a relay (double pole, double throw), two limit switches (one normally open, the other normally closed), and a diode.
simulate this circuit – Schematic created using CircuitLab
The vertical travel of the weight is shown on the left, along with a protruding arm which will contact SW2 at the bottom of its travel, and SW1 at the top. The motor is wired so that with 6 volts on its + lead, the weight rises.
Start by assuming the relay is off. The two sets of contacts are active on the right, so +6 is applied to the - motor lead and the weight is descending. Because the motor + lead is at ground, there is no voltage applied to the diode.
Now let the arm on the weight activate SW2. +6 volts flows through SW2 and SW1 and applies power to the relay. Now the contacts to the left are active and two things happen. First, +6 is applied to the motor + lead, and the weight starts rising. Two, +6 volts flows through the diode, and also tries to drive the relay. So when the weight starts to rise and the arm no longer closes SW2, the relay is still driven and the motor continues to drive the motor to raise the weight.
This continues until the arm contacts SW1 and opens the relay circuit. With no power to the relay, the contacts shift to the right and the motor reverses direction. The weight starts down. Since the motor + lead is now connected to ground, there is no voltage on the diode, so when the weight drops and SW1 closes again, there is still no drive for the relay coil.
The cycle will continue as long as the battery holds out.
Speaking of which, I really recommend either much bigger batteries or a power supply you can plug into a wall. I believe you'll find that you drain AA cells much faster than you think.