Browse Prior Art Database

Automatic Line Length Display, Recording, and Passing Algorithm for Magnitude Estimation Disclosure Number: IPCOM000130322D
Publication Date: 2005-Oct-19
Document File: 3 page(s) / 19K

Publishing Venue

The Prior Art Database

Related People

Jason Cohen: AUTHOR


An automated system is described for screening panels of human subjects who make judgments related in some way to spatial perception of objects. The method employs customized software tools to improve known magnitude estimation procedures in which panelists are screened based on their estimations of line lengths. The software provides a computer interface that displays line lengths, records the estimated lengths, analyzes the data and determines if subjects pass based on a mathematical algorithm. In this system, screening can be done rapidly and objective criteria for accepting or excluding panelists can be enforced.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 41% of the total text.

Page 1 of 3

  Automatic Line Length Display, Recording, and Passing Algorithm for Magnitude Estimation

  Jason Cohen Kimberly-Clark Corp. Neenah, Wisconsin

Many commercial and academic studies rely on panels of human subjects who make judgments related in some way to their spatial perception of objects. One example may be a panel responding to the aesthetic and functional properties of products such as diapers, disposable gloves, or product packaging. One approach to such testing is known as magnitude estimation, in which subjects are asked to provide positive numerical scores for the items being evaluated, wherein the scores are not from a predetermined range, have no arbitrary upper bound, and are based on a subjective personal scale that can differ widely from one subject to another. For example, one subject may rank the brightness of a tissue product using a subjective scale with values such as 1000 for a low-brightness tissue and 100,000 for a significantly brighter tissue, while another subject may give the same tissues scores of 0.5 and 3, respectively. For each subject, the scores need to be normalized to the apparent scale of the user in order to allow the results from multiple users to be processed statistically. Generally, the panelists are not experts who are trained to use a particular scale, but are non-experts creating their own internal rating system in evaluating items. In many cases, the non-expert panelists are selected to represent a target population of consumers or users.

In such testing, it is important to ensure that panelists are able to follow instructions and that they are not impaired in terms of estimating magnitudes. This is typically done by screening the panelists with a line length estimation test. The panelists are shown a series of lines, randomly selected from a group of varying lengths, and asked to provide numbers to describe the lengths of the lines. The estimated line lengths (using the arbitrary internal scale generated by the panelist) are then plotted against the known line lengths on a log-log plot. A straight line relationship is generally expected, indicating a power-law relationship between the internal human scale and the actual length. The power-law exponent is unity when the user's estimated length is a linear function of the true length, but the exponent in practice can be greater or smaller than unity. After plotting the data on a log-log plot, the goodness of fit (e.g., R2 value) is examined, and if it is poor (i.e., R2 less than a predetermined cut-off such as 0.6), then the panelist may be excluded from the study, while the others are passed on to the rest of the study.

The use of line lengths for screening prior to or part of a magnitude estimation study has been well documented in the literature (e.g., Verrillo et al., (1999) "Subjective Magnitude of Tactile Roughness," Somatosensory Motor Research, 16:352-360, and J.J. Zwislocki and D.A. Goodman, (1980) "Absolute Scaling and Sensory Magnit...