Is the R-Index losing its seat at the sensory table?

Proportion correct (𝘗𝘤), 𝘱-values, proportion detectors (𝘗𝘥), Thurstonian delta (𝘥'), and the 𝘙-Index...

When it comes to estimating the size of a sensory difference, measuring units come in numerous shapes and colors! 🟦🟢🔶❌➕

But are they 𝗮𝗹𝗹 𝗲𝗾𝘂𝗶𝘃𝗮𝗹𝗲𝗻𝘁, or is one of them the 𝗯𝗲𝘁𝘁𝗲𝗿 𝗰𝗵𝗼𝗶𝗰𝗲 and why?

To be reliable, a sensory index must have a 𝗿𝗼𝗯𝘂𝘀𝘁 𝘀𝗰𝗶𝗲𝗻𝘁𝗶𝗳𝗶𝗰 𝗳𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 that allows for consistent 𝗰𝗿𝗼𝘀𝘀-𝗺𝗲𝘁𝗵𝗼𝗱𝗼𝗹𝗼𝗴𝘆 comparisons.

- Scientific reports and publications have long shown why indices like p-values, proportion correct, and proportion detectors (Pd) are not reliable estimates.

- They can vary widely based on 𝗺𝗲𝘁𝗵𝗼𝗱𝗼𝗹𝗼𝗴𝘆 or 𝘀𝗮𝗺𝗽𝗹𝗲 𝘀𝗶𝘇𝗲, even when comparing the same two products.

So, what about the more popular options?

• The 𝙏𝙝𝙪𝙧𝙨𝙩𝙤𝙣𝙞𝙖𝙣 𝙙𝙚𝙡𝙩𝙖 (𝙙') stands out with a strong scientific foundation and a wealth of peer-reviewed research supporting its models and assumptions.

• The 𝙍-𝙄𝙣𝙙𝙚𝙭, while useful within a single method, is method-dependent.

o For instance, an 𝘙-Index estimated with the A/Not-A method will 𝗰𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝘁𝗹𝘆 𝗯𝗲 𝗵𝗶𝗴𝗵𝗲𝗿 than one estimated using the Same-Different method, even for the same samples.

o While the A/Not-A 𝘙-Index can estimate the proportion correct in a 2-AFC, this mapping doesn't hold for the 𝘙-Index in Same-Different or Ranking tests.

o The 𝘙-Index can be useful in some situations, but the practitioner should be aware of its potential limitations.

While it was highly popular in our field in a not too distant past, I do not see the R-Index used or mentioned as much recently.

Interested in others’ experiences:

1️⃣ Were you aware of the 𝘙-Index as a unit of sensory difference?

2️⃣ Do you use it for specific applications?

3️⃣ Do our colleagues in academia teach it in their sensory science curriculums?