On fitting functions to data for the percent of a community highly annoyed by noise such as impulsive noise

Survey respondents ‘‘highly annoyed’’ (HA) typically are categorized as those responding in the top 30% on an anchored numerical scale or in the top two categories on an adjectival scale. The percent HA typically ranges from 0% to 40%. At higher percents HA (e.g., >15%), the noise stimulus is clear and dominant and the reliability to the percentage values is high. At low percentages, the reliability is low. Some respondents may be annoyed any time they hear a sound. Depending on the number of such respondents, the percent HA can vary greatly on a percentage basis (4% HA versus 2% HA is a change of 100%). Therefore, alternatives to using small percentages HA (e.g., <10%) are considered when fitting transition functions to HA data. One method is to approximate the transition function by three asymptotes. The transition (sloped) asymptote would be fit by the 50% point and the slope when the percentages HA were large. Another way to avoid small percentages HA is to consider the group ‘‘little annoyed’’ or ‘‘h...