I am currently conducting empirical studies to evaluate the run-time complexities of the quicksort, and mergesort algorithms. To do this I run a random number generator that stores whatever amount of numbers I specify in a binary file. The ranges of those numbers are from 1-1,000,000.I then run tests of each algorithm starting from 100,000 numbers, incrementing by 50,000 each time, until 1,000,000 numbers are sorted on the last run. So 20 tests each. I have successfully completed each algorithm but my results are kind of puzzingly. This is a graph showing my results.

I understand that quicksort has a worst case of O(n2) time, but typically O(n·lg(n)) time. Mergesort has Θ(n·lg(n)) time.
Also I would like to note that when I started the timer I just used clock() from time.h, and calculated the time elapsed. I started my timer one line of code before I called my sorting function.
What I dont understand is how my graph shows mergesort is always double the time, and reaching triple the time to sort numbers compared to quicksort.
My only thought is that for my mergesort algorithm every time I divide my array in half I use malloc to create a new integer array for each half. Of course this means a large amount of calls are made to malloc considering the number sizes I am sorting.
int* mergeSort(int* nums, int size){
int* left;
int* right;
int middle = size/2;
if(size <= 1)
return nums;
split(nums, size, &left, &right, middle);
//I dont understand why the code below wouldnt work in place of the split()
//when i run it, in main, nothing gets printed out. I guess i lose my pointer to the beginning of my array.
//left = nums;
//right = nums+middle;
left = mergeSort(left, middle);
right = mergeSort(right, size - middle);
merge(nums,left,right,middle,size - middle);
free(left);
free(right);
return nums;
}
void split(int* nums, int size, int** left, int** right, int middle){
int *lft = (int*) malloc ((sizeof(int) * middle));
int *rght = (int*) malloc ((sizeof(int) * size - middle));
int mid = middle;
int upMid = size - middle;
int i;
for(i=0; i < mid; i++)
lft[i] = nums[i];
for(i=0; i < upMid; i++)
rght[i] = nums[i+middle];
*left = lft;
*right = rght;
}
void merge(int* num, int* left, int* right, int sizeLeft, int sizeRight){
int i,j,k,n;
i=j=k=0;
n=sizeLeft + sizeRight;
while(k < n){
if(i< sizeLeft){
if(j<sizeRight){
insert(num,left,right,&i,&j,&k);
}
else{
append(num, left, sizeLeft, &i, &k);
}
}
else{
append(num,right,sizeRight,&j,&k);
}
}
}
void insert(int* num, int* left, int* right, int* i, int* j, int* k){
/*int i,j,k,n;*/
if(left[*i]<right[*j]){
num[*k] = left[*i];
(*i)++;
}
else{
num[*k] = right[*j];
(*j)++;
}
(*k)++;
}
void append(int* num, int* half, int sizeHalf, int* i, int* k){
while(*i < sizeHalf){
num[*k]= half[*i];
(*i)++; (*k)++;
}
}
I would greatly appreciate any feedback on this question of mine, and any advice on maybe making my mergesort function more efficient. Thanks!!