Scotopic Advantage, a term that began in the early years of LED streetlights, has become very popular to help promote LEDs as a lower luminance equivalent to conventional lighting.  In those days, LED output was low but color temperature was higher than the HPS (High Pressure Sodium) lighting it was replacing.

The basic premise is this:  At low luminance conditions, our visual sensitivity shifts toward a bluer color. So generally it makes sense that a bluer light^[1] source would allow us to see better during lower luminance conditions. The key is low luminance conditions and the advantage is usually cited without defining wavelength or luminance level.
I may have been one in the first 500 people in the nation to promote this science. 

Today you may hear that LEDs are rich in Scotopic Lumens^[2] and that 70 Watts can replace 200.   Be careful.
If you could have accepted less light, you could have simply used a smaller bulb.

Some LED promotional material overlooks that luminance^[3] is the selector between photopic and scotopic vision. Some material relates scotopic to color, e.g. rich in Scotopic Lumens, and some will say that bluer light is advantageous at any luminance.  Some marketing material claims that LED scotopic advantage can improve your sharp focused vision, and they call that pupil lumens.^[4] Some literature has an S/P^[5] ratio (scotopic/photopic) for use with multiplier tables to show the scotopic effectiveness of their light source.   What can you believe?

Note that most graphs for Human visual response show a photopic response curve overlayed with a scotopic response curve for relative color comparison.  The problem is that scotopic and photopic are different luminance levels, two distinct places along the vertical plot; they can't exist at the same time.  This overlay can be misleading to the non-technical and may explain some statements about scotopic color enhancement with photopic luminance levels.

Human vision hasn't changed due to technology or national economy.  But "scotopic" brings on a blue light frenzy, as if we couldn't see the color blue with that third cone.

The scotopic spin might have recently backfired.

During approximately the Fall of 2009, the National Lighting Product Information Program (NLPIP) at Rensselaer Polytechnic Institute's Lighting Research Center (LRC) purchased LED streetlights to make an "apples-to-apples" comparison between HPS, LED and MH (Metal Halide) lighting technologies. This study launched a year-long discussion thus far. One comment about the evaluation that deserves attention is that for this evaluation, all LED streetlight samples submitted were of lower power and luminous output than required to meet IESNA RP-8 in a one for one exchange. RP-8 is the standard they were to meet. Consequently, the LRC adjusted pole spacing to accomodate these LED lights, which skewed the cost and payback, certainly not in favor of LEDs. (Note: The LRC re-ran the evaluation after acquiring adequate LED lights. Please see the report and comments.) Ironically, the low-output products first submitted were all selected by the LED manufacturers' representatives. They must have believed a lower powered product could do the job, but the NLPIP did not agree. You can follow the web links from here and from LEDs Magazine here.

The Rensselaer Lighting Research Center is an advocate of scotopic lighting improvement, within certain limits. They have an extensive paper (Volume 6, Issue 2; Outdoor Lighting Visual Efficacy Application) teaching a unified system of photometry based on scotopic/photopic ratios and lower luminance level. They provide a table of multipliers. The paper has one rather noteworthy caveat: "For photopic luminances equal to or greater than 0.6 cd/m² the unified luminance simply equals the photopic luminance."
Basically if you require 0.6 cd/m² or more, your lighting must provide that luminance without the aid of scotopic multipliers.

A Photopic Matter:      When luminance is ≥ 1 cd/m2 Blue is a primary color received very well by the cones. Photopic response is shown to the right as a relative percentage. This is not a scotopic scale. Comparing the same input power, cool white appears brighter than warm white at photopic levels because 5000K CCT (approx. 510nm center band) is higher on the curve than 2700K (approx. 610nm center band).  It's one reason cool white LEDs have higher efficacy than warm white LEDs. This phenomenon is not scotopic; it's merely that different wavelengths will naturally have a different luminous efficacy on the photopic response curve.

Background: 
Photopic vision is the term for Human visual response to daylight, or luminance of ≥ 1 cd/m².  The photopic range is sufficient for our cone receptors (red-green-blue) to perceive color, and this peak sensitivity is green. 
Scotopic vision occurs at a luminance level of about 0.01 cd/m²; it relies mostly upon rod receptors and color sensitivity peak is blue.  This luminance is about equivalent to a full moon.
Mesopic vision is that range from 1 cd/m² down to 0.01 cd/m² (photopic to scotopic) in which our color sensitivity peak is moving, it lies somewhere between green and blue. Cone and rod reception^[6] is somewhat overlapping.

Converting^[7] luminance levels to illuminance involves reflectance and several other factors, but if we assume a constant 15% reflectance, illuminance in footcandles will be quite close to 2 times luminance in cd/m².  So making that assumption, the mesopic range is from about 2FC to 0.02FC illumination.

Scotopic Vision (defined as ≤ 0.01 cd/m²), does not occur with daylight, or with ordinary indoor room illumination, or with outdoor lighting of ≥ 2FC, such as stadiums, gas stations, car lots, commercial parking areas, etc. 

The Scotopic Lumen Spin is used regardless of luminance level implying that less LED footcandles can always do the same job as higher footcandles at a lower color temperature. 


Putting it together: 
1) By the LRC's unified system of photometry, there's no scotopic advantage when illumination requirements are 0.6cd/m² (about 1.5 footcandles) and higher.  At that level, the cones in our retina do the work; that's Foveal Vision. 
2) Tasks requiring sharp focus^[8] rely solely upon Foveal Vision (cones only) and will usually have more than 5FC (see tables) with a full spectrum light source; these tasks do not utilize rod receptors. 
3) Given 1 and 2 above, a scotopic advantage is possible only when luminance is less than 0.6cd/m² AND when the LED lighting provides a bluer spectrum than lighting it is replacing. For example, there is no scotopic advantage from LEDs replacing metal halide or fluorescent lighting of 5000K CCT; that's already blue. For outdoor areas already using Metal Halide, promoting less footcandles with LEDs to replace MH would be a costly mistake.

Actual practice in the real world:  That asymmetrical delay with night vision is a problem.  Human vision adapts to mesopic levels over time, several minutes or more, depending on the degree of luminance shift.  We adapt to mesopic sensitivity slowly.  The greater the span between photopic and the maintained mesopic luminance, the longer our eyes need to adapt. From mesopic levels, when exposed to higher light levels, our vision shifts back to photopic in mere seconds. We adapt quickly to the brightest light in our field of view. Any sudden brighter light can do it; could be from car headlights, moving-message roadway signs, commercial signage, glance at a store front or at a streetlight, etc. From photopic to mesopic levels, we are temporarily blind while re-acquiring our mesopic vision.

The IESNA has not incorporated scotopic effects into its lighting recommendations. (Position Statement PS-02-10)
Also it seems the LRC must be on to something by recommending a 0.6 cd/m² upper limit, over which, photopic measurements have no scotopic multipliers. This helps to reduce the span between photopic and maintained mesopic luminance, which helps to reduce the adaptation time to night vision.

Lighting standards or specifications requiring more than about 1.5 footcandles of illumination would not be candidates for scotopic experimentation. Don't be confused by sales hype,^[9] supply lighting by performance; if your customer cannot see as promised with the product you supply, then you have a problem. For safety and for liability, the best practice may be to supply illumination in accordance with peer-accepted practices; LEDs can easily do that today.
Have your proposals verified with IES photometric files by a lighting professional.

Note about the scotopic lumen in commercial & retail spaces:
Many phosphor converted white LEDs are lacking in the red spectrum, even though having a relatively high CRI. If you must use cool white, the LED red spectrum will be even weaker. Field evaluation of CRI originally used a chart with eight colors of intermediate saturation, and this chart has expanded to 14 to include six saturated colors, R9 through R14. When comparing LEDs for commercial and retail lighting applications, be sure to test for "R9" to get the spectral purity you need.