This is a neat article explaining why dyno correction factors are usually for NA cars and tend to overinflate the numbers for forced induction vehicles.

Muscle Mustangs wrote:

Positioned above the door entering the dyno room at Westtech is a simple statement that reads "Polygraph".  For the uninitated, a Polygraph machine is better known as a lie detector.  By measuring electrical signals (the galvanic skin response) in the human body, the machine is able to determine physiological changes that accompany the stress associated with deception.

Along those same lines is a phrase adjacent to the Dynojet chassis dyno at Kenne Bell's place.  The phrase reads:  "The problem with testing is that you are bound to generate data."

Both of these statements basically indicate that the ultimate test of a motor's  performance is the dyno.  The dyno doesn't care who built the motor, how long they've been a prominent figure in the industry, or how many races they've won or lost.  It provides cold, hard facts; data that - much to the dismay of the engine builder (or eventual owner)--often runs contrary to the expected outcome.  But is this instrument really the last bastion of truth?

In it's most basic form, the dyno (be it of the chassis or engine variety) can actually provide accurate data.  This is especially true when looking for what we call relative power gains.  If we run the motor on the dyno with one set of cylinder heads and then again with another, the relative difference should be accurately portrayed.  This, of course, assumes no changes in the calibration, air temp, or any of the many variables that can skew the outcome.  Dyno testing is tricky business, and the data supplied can easily be manipulated through no fault of the dyno itself.  Change the correction factor, the displacement, or even the absorber settings (something we refer to as knobbing), and there goes your direct back-to-back test.

Evil-doers aside, even honest mistakes or dyno errors can cause havoc when performing extensive testing.  The lesson here is that if something seems too good to be true, you'd better check all the data to find the error before bragging about a motor that makes 100 hp more than it should.

There was a time when dyno testing was relegated to serious race teams, and the general public had little access to real-world data.  This have changed dramatically over the last 10-15 years, though.  Thanks in part to the introduction of the Dynojet chassis dyno, dyno testing has become so commonplace that we now have dyno drags, where enthusiasts gather together to post the highest numbers.  Certainly easier on the motor than most forms of racing, the prevalence of dyno drags indicates the importance and acceptance of performance numbers generated on the dyno.

One of the problems associated with dyno testing is comparing results at different locations and especially on different dynos.  While comparisons between runs made on Mustang and Dynojet chassis dynos are truly exercises in frustration, even runs made at different locations on the same type of dyno are fraught with errors.  One of the major areas of frustration is the dreaded correction factor.  Designed to compensate for changes in weather conditions during testing (such as altitude, temperature, and air density), the correction factor basically recalculates the power output of the test motor to a standardized temperature and elevation.  The idea behind the correction factor is to let an owner testing his car a mile up in Denver know what it would make at sea level.  While the idea behind the correction factor is a sound one, the universal application has a few kinks.

The reason for a correction factor in the first place is that air density decreases with elevation.  In lay terms, as you climb, there's less oxygen, and a motor will make less power when tested at an elevation of 3,000 feet than it would at sea level.  The correction factor employed by most dyno systems takes this into account and corrects the tested power output to a standardized temperature and density.

The problem with this system is that the correction factor employed is for normally aspirated engines.  The power output of a normally aspirated motor will drop roughly 4 percent per 1,000 feet of elevation change.  The problem arises when you run a turbo or blower motor on the dyno and the same correction factor is employed, as a motor equipped with forced induction (especially turbocharged versions) do not lose the same 4 percent.  According to SAE standards, the power output of a turbo motor drops only 2.43 percent per 1,000 feet of elevation thanks to the altitude compensation ability of a turbocharger (it basically increases impeller speed due to a change in the pressure differential assuming a fixed boost level).

Thus, there are tuners and enthusiasts running their turbo motors at high altitude to exaggerate their real power outputs.  Running a motor at an elevation of 6,000 feet would provide a correction factor of 24 percent (I'm assuming a lot here, but stay with me), when in reality, a turbo motor should only have a correction factor of just over 14.58 percent.  This disparity in correction factors would knock a 1,000 hp (corrected) turbo motor down to a more realistic 925 hp.

The next you time oooh and aaah over the dyno numbers produced by a turbo motor, ask where it was tested.  The extra 10 percent offered by an inaccurate correction factor really starts to add up.