Alliance for Lighting InformationAlliance for Lighting InformationWhile there is some discussion about the meaning of Confidence Intervals and Bound Error, there is little explanation as to what they are, how they can be applied, and what the actual values stand for in relative terms. The confidence interval is the likely range of the true value. Note that there is only one true value, and that the confidence interval defines the range where it's most likely to be. Confidence intervals assume that the data being analyzed falls into a 'normal distribution' and relies on the outlying data to compute the error bounds. Data that has been culled or binned into ranges cannot be subjected to Bound Error analysis because the boundaries have been set and the outlying data dismissed.
In most statistical analyses that employ Confidence Intervals the Interval is usually expressed as a 95% confidence. This report uses a 90% confidence interval.
Many of the tables presented list percentages of FUA's or Sites that fit into data ranges. The problem is that the percentages reported do not correspond with the number of sample sizes listed. For instance, Table 34 lists the percentage of sites with parking lots for different building types that have a uniformity ratio within a certain range of values. This means that a parking lot either does or does not fit a the set value range, yet the table reports that for a sample size of 2 (Full Service Restaurants) that 90.4% of the sites fit in the 5.0 to 9.99 range and 9.6% of the sites have a uniformity ratio in the 15 to 19.99 range. These percentages are in error as the only mathematically available percentages with a sample size of 2 are 0%, 50% or 100%. This type of calculation error can be seen in Tables 24, 26, 28, 30, 32, 34, 36, 38, 40, 55, 56, 58, 60, and 62. These calculation errors indicate that, while the input data may have been reviewed for accuracy, the results of the data manipulation were not rigorously reviewed.
Table 55 illustrates another type of calculation error. Looking at the estimates for lighting controls installed at ATMs, this report estimates that 26.8% utilize manual controls, 56% utilize time clocks, and 90.8% utilize photocell controls. These estimates indicate that 173.6% of the ATMs use these three control methods. This is physically impossible - the addition of the control percentages should sum to no more than 100%.
It is also evident that the data sets are not complete. Comparing Table 12 with Table 24 it becomes apparent that less than 50% (387 out of 778) of the functional use areas surveyed include lighting measurements. From Table 24 it is apparent that the surveyors concentrated their efforts on Parking Lots and Walkways as these two Functional Use Areas represent 87% of the sites with illuminance measurements. The other nine functional use areas are represented by meaningless sample sizes. Any scaled-up conclusions drawn about illuminance values cannot be substantiated by the data.
This site is provided as a public service by resodance publishing co. and no guarantee is implied or provided.
This page is served & maintained by &
last changed on 7 May 03 by
