Why is ArrayList faster than HashSet in java?

I'm used to the fact that HashSet is a pretty fast implementation of a data structure because it uses hashes (and in turn is implemented through HashMap). I was solving some problems and decided to check the performance, so here it is:

You are given an array of numbers - [11, 3, 11, 11, 3, 2, 0, -2, 2] You must write a function that returns a number that appears "odd" times.

Here is my solution:

public class OddNumInArray {

public static List<Integer> oddNumList(int [] ar){
    Collection <Integer> l = new ArrayList<>();
    for (int n: ar) {
        if (l.contains(n)) {
            l.remove(n);
        }
        else {
            l.add(n);
        }
    }
    return (List) l;
}

public static Set<Integer> oddNumHSet(int [] ar){
    Set <Integer> l = new HashSet<>();
    for (int n: ar) {
        if (l.contains(n)) {
            l.remove(n);
        }
        else {
            l.add(n);
        }
    }
    return l;
}

public static void main(String [ ]arg) {
    int [] a1 = new int [10000000];
    for (int i=0; i<10; i++){
        a1[i]=(new Random()).nextInt(5);
    }
    long cur= System.nanoTime();
    System.out.println(oddNumList(a1));
    long c1 = System.nanoTime()-cur;
    System.out.println("TIME CONSUMED:" +c1);
    cur= System.nanoTime();
    System.out.println(oddNumHSet(a1));
    long c2 = System.nanoTime()-cur;
    System.out.println("TIME CONSUMED:" + c2);
    System.out.println("c1/c2*100: "+ (new Double(c1)/new Double(c2)*100));
}

}

      

        
        
        
      

    

And here's the output:

[1, 0]
TIME CONSUMED:101804000
[0, 1]
TIME CONSUMED:183261000
c1/c2*100: 55.55137208680516

      

        
        
        
      

    

So why the implementation with ArrayList is almost 2 times faster than one with HashSet? Thank.