You could look into noise, noiseDetail and noiseSeed.

A while ago I did the same to try out different number distributions. See the results here.

The solution is to create an array and populate it with the things to pick in proportion to their probablity of being picked. Then pick array elements at random whenever you want.

For instance the code below has an array where
20% of the elements are 'A'
30% of the elements are 'B'
10% of the elements are 'C'
40% of the elements are 'D'

If we pick one at random then these are also the probabilities.

The resolution is 10% so if you wanted the following distribution
20% of the elements are 'A'
5% of the elements are 'B'
45% of the elements are 'C'
30% of the elements are 'D'
The the array would need to have at least 20 elements.

1. char[] ch = new char[] {'A', 'A', 'B', 'B', 'B', 'C', 'D', 'D', 'D', 'D'};
2. 
   
3. int n;
4. char cn;
5. 
   
6. void setup() {
7.   for(int i = 0; i < 20; i++) {
8.     n = (int)random(0, ch.length);
9.     cn = ch[n];
10.     print(cn + " ");
11.   }
12.   println();
13. }

Why have you changed the topic type? (I changed it back to Question.)
When you have similar code which is repeated, you can often factor that by using some data structure, starting with arrays.
The nice thing is that is scale well, ie. you should have not to change the logic if you want more steps.
Example:

1. int[] distribution = { 2, 3, 3, 2 };
2. int[] stats = new int[distribution.length];
3.  
4. void setup()
5. {
6.   size(500, 200);
7.   randomSeed(System.nanoTime());
8. }
9.  
10. void draw()
11. {
12.   int rnd = 1 + int(random(10));
13.   int total = 0;
14.   int pos = -1;
15.   for (int i = 0; i < distribution.length; i++)
16.   {
17.     total += distribution[i];
18.     if (rnd <= total)
19.     {
20.       pos = i;
21.       break;
22.     }
23.   }
24.   assert pos >= 0;
25.   stats[pos]++;
26.  
27.   background(255);
28.   int xpos = 20;
29.   for (int i = 0; i < stats.length; i++)
30.   {
31.     fill(100, 255 - i * 20, 128 + i * 20);
32.     rect(xpos, height - stats[i], 50, stats[i]);
33.     xpos += 70;
34.   }
35. }

Doh, I wrote my message some hours ago and forgot to send it...