Programming :: Association Lists Are Faster Than Hash Tables?

I'm writing an interpreter and it used simple association lists for mapping varaible names to their values. Here's the code:

Code:

#include "assoc_array.hpp"
#include <string.h>
using namespace LANG_NAMESPACE;

[code]....

I thought that I would replace it with a hash table to increase performance. Note that I decided to store linked lists in the buckets instead of the actual values, in case the hash function outputs the same index for multiple variable names:

Code:

#include "hash_table.hpp"
#include "assoc_array.hpp"
#include <string.h>

[code]....

It runs in about 0.333 seconds on my machine. Since I'm using git version control, I decided to bring back the old version that used association lists. To my amazement, it ran 3 times faster, completing in 0.116 seconds! Is my hash table implementaion really that bad, or is this a really poor benchmark (and real-world code actually will be faster using the hash table)?