The problem with defining how much people think about things like sex is like the
Hawthorne Effect.
When was the last time I thought about sex? I'm not sure. If it was a casual thought, it's like asking 'When did you last think about blowing your nose?'. I didn't think about thinking about it. If, after hearing that The Average Male Thinks About Sex Every Seven Seconds, you decide to monitor your thoughts and see if you are thinking about sex every seven seconds, the test will be contaminated: You are thinking about sex in the context of looking for thoughts about it.
The statement can neither be proven true nor false, because any monitoring the subject is aware of will effect the result, and the subject must be aware in order to mointor.
Probably the best way to investigate the statement would be to get a brain activity scanner, record the area showing the most activity when the investigate-e is shown pornography, and recording how often activity is shown in that area. The problem would be stopping the people partaking in the investigation figuring out the subject of the investigation, or for that matter getting them to agree to the investigation without knowing what was happening.