The Invisible Bubble: Is your reformulation 'close enough'?

🫧 𝗧𝗵𝗲 𝗜𝗻𝘃𝗶𝘀𝗶𝗯𝗹𝗲 𝗕𝘂𝗯𝗯𝗹𝗲: 𝗜𝘀 𝘆𝗼𝘂𝗿 𝗿𝗲𝗳𝗼𝗿𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻 ‘𝗰𝗹𝗼𝘀𝗲 𝗲𝗻𝗼𝘂𝗴𝗵’? 🧋

Sensory “matching” is 𝗻𝗼𝘁 𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲. The real question is: How do we establish that a reformulation is ‘𝘀𝗮𝗳𝗲’ 𝗳𝗼𝗿 𝘀𝘂𝗯𝘀𝘁𝗶𝘁𝘂𝘁𝗶𝗼𝗻?

Only consumer input can answer that question. Statistics alone 𝘸𝘰𝘯’𝘵 𝘩𝘦𝘭𝘱, psychology is needed, and the 𝗧𝗵𝘂𝗿𝘀𝘁𝗼𝗻𝗶𝗮𝗻 framework again plays a central role.

𝗖𝗮𝘀𝗲 𝗦𝘁𝘂𝗱𝘆: 𝗧𝗵𝗲 𝟮-𝗔𝗖 𝗠𝗲𝘁𝗵𝗼𝗱

Revisiting data from Braun 𝘦𝘵 𝘢𝘭. (2004) on Badoit water carbonation provides an illuminating illustration:

 - 𝗧𝗵𝗲 𝟮-𝗔𝗙𝗖 𝗺𝗲𝘁𝗵𝗼𝗱: “Which is 𝘮𝘰𝘳𝘦 𝘤𝘢𝘳𝘣𝘰𝘯𝘢𝘵𝘦𝘥?”. Significant differences (𝘱 < 0.05) for all three pairs, but were those differences 𝗺𝗲𝗮𝗻𝗶𝗻𝗴𝗳𝘂𝗹? - No easy way to answer this question

 - 𝗧𝗵𝗲 𝟮-𝗔𝗖 𝗺𝗲𝘁𝗵𝗼𝗱: “Which is 𝘮𝘰𝘳𝘦 𝘤𝘢𝘳𝘣𝘰𝘯𝘢𝘵𝘦𝘥, or is there 𝘯𝘰 𝘤𝘢𝘳𝘣𝘰𝘯𝘢𝘵𝘪𝘰𝘯 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘤𝘦?”

 - Both methods resulted in 𝗲𝗾𝘂𝗶𝘃𝗮𝗹𝗲𝗻𝘁 𝘀𝗲𝗻𝘀𝗼𝗿𝘆 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝘀 (𝘥’ values), Thurstonian framework 👍 .

𝗧𝗵𝗲 𝗕𝗿𝗲𝗮𝗸𝘁𝗵𝗿𝗼𝘂𝗴𝗵: 𝗧𝗵𝗲 𝗧𝗮𝘂 (τ) 𝗖𝗿𝗶𝘁𝗲𝗿𝗶𝗼𝗻

Using the 𝟮-𝗔𝗖 𝗺𝗼𝗱𝗲𝗹, we can estimate the τ 𝗰𝗿𝗶𝘁𝗲𝗿𝗶𝗼𝗻 = The size of the momentary sensory difference required before a 𝘤𝘰𝘯𝘴𝘶𝘮𝘦𝘳 𝘳𝘦𝘱𝘰𝘳𝘵𝘴 𝘢 "𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘤𝘦" (not estimable with forced choice methods such as the 2-AFC.)

In this study, τ 𝘄𝗮𝘀 𝟬.𝟳𝟱. This becomes our benchmark:

 • Pair A vs. B: 𝘥' = 0.67 → Slightly 𝗯𝗲𝗹𝗼𝘄 threshold

 • Pair A vs. C: 𝘥' = 1.65 → 𝗔𝗯𝗼𝘃𝗲 threshold

 • Pair B vs. C: 𝘥' = 1.04 → 𝗔𝗯𝗼𝘃𝗲 Threshold

𝙽̲𝚘̲𝚝̲𝚎̲: Of course, final conclusions must consider the variance of all estimates.

𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿𝘀 𝗳𝗼𝗿 𝘆𝗼𝘂𝗿 𝗥𝗢𝗜:

 1. 𝗥𝗶𝘀𝗸 𝗣𝗿𝗼𝗳𝗶𝗹𝗶𝗻𝗴: Use the threshold to 𝘴𝘦𝘵 𝘴𝘶𝘪𝘵𝘢𝘣𝘭𝘦 𝘴𝘵𝘶𝘥𝘺 𝘴𝘱𝘦𝘤𝘪𝘧𝘪𝘤𝘢𝘵𝘪𝘰𝘯𝘴 including internal 𝘀𝗮𝗺𝗽𝗹𝗲 𝘀𝗶𝘇𝗲s and 𝘁𝗲𝘀𝘁 𝗽𝗼𝘄𝗲𝗿.

 2. 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗖𝗼𝗻𝗳𝗶𝗱𝗲𝗻𝗰𝗲: Steer clear from "𝘴𝘵𝘢𝘵𝘪𝘴𝘵𝘪𝘤𝘢𝘭 𝘨𝘩𝘰𝘴𝘵𝘴" that are 𝗻𝗼𝘁 𝗺𝗲𝗮𝗻𝗶𝗻𝗴𝗳𝘂𝗹 to consumers.

 3. 𝗘𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆: Focus R&D resources only on differences that 𝗲𝘅𝗰𝗲𝗲𝗱 𝘁𝗵𝗲 τ 𝗯𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸.

We find the 𝗧𝗵𝘂𝗿𝘀𝘁𝗼𝗻𝗶𝗮𝗻 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 𝗲𝘀𝘀𝗲𝗻𝘁𝗶𝗮l for ensuring product substitution 𝘪𝘴𝘯'𝘵 𝘭𝘦𝘧𝘵 𝘵𝘰 𝘤𝘩𝘢𝘯𝘤𝘦.

𝗪𝗵𝗮𝘁 𝗮𝗯𝗼𝘂𝘁 𝘆𝗼𝘂? Are you leveraging Thurstonian models in your decision process, or is there another scientific framework guiding your risk management?

𝙍𝙚𝙦𝙪𝙚𝙨𝙩 𝙤𝙣 𝙤𝙪𝙧 𝙬𝙚𝙗𝙨𝙞𝙩𝙚: 𝘊𝘰𝘳𝘳𝘰𝘣𝘰𝘳𝘢𝘵𝘪𝘯𝘨 𝘵𝘩𝘦 2-𝘈𝘍𝘊 𝘢𝘯𝘥 2-𝘈𝘊 𝘛𝘩𝘶𝘳𝘴𝘵𝘰𝘯𝘪𝘢𝘯 𝘮𝘰𝘥𝘦𝘭𝘴 𝘶𝘴𝘪𝘯𝘨 𝘣𝘰𝘵𝘩 𝘢 𝘮𝘰𝘥𝘦𝘭 𝘴𝘺𝘴𝘵𝘦𝘮 𝘢𝘯𝘥 𝘴𝘱𝘢𝘳𝘬𝘭𝘪𝘯𝘨 𝘸𝘢𝘵𝘦𝘳. 𝘉𝘳𝘢𝘶𝘯 𝘦𝘵 𝘢𝘭. (2004). 𝘍𝘰𝘰𝘥 𝘘𝘶𝘢𝘭𝘪𝘵𝘺 𝘢𝘯𝘥 𝘗𝘳𝘦𝘧𝘦𝘳𝘦𝘯𝘤𝘦, 15, 501-507.

Thanks for the great research Virginie Braun Secret !