Alliance for Lighting InformationAlliance for Lighting InformationThe following is a summary review of the Outdoor Lighting Baseline Assessment - Final Report on behalf of California Energy Commission Public Interest Energy Research (PIER) Program. The links lead to more detailed discussions of particular topics.
A copy of the PIER Report can be downloaded from:
http://www.newbuildings.org/lighting.htm
Follow the link to the registration page to download the Element 7 Report.
Many of the generally accepted principles of Scientific Method have been ignored in this study. Peer review, the reproducibility of data, the accurate and appropriate use of references, and the appropriate use and description of new metrics have all been ignored. For a document that will have an impact on legislation in the state of California and possibly nationally, these principles cannot be dismissed.
The methodology of the Sample Design is also rather dubious. The use of zip codes as the main basis for the sample selection may not be very appropriate. Also, the notion that there is some correlation between the number of employees in a facility and the amount of exterior lighting is unfounded. When the Sample Design Methodology implies that an adequate sample size for grocery stores is 3 for a population of 7281 grocery stores in the state of California, there is something wrong with the design methodology. When the methodology infers that an adequate sample size is less than 3 there is also a problem, since less than 3 data points will not allow the data to fall into a 'normal distribution curve' which is necessary for any type of statistical analysis.
This study introduces the new metric of Glare Ratio. This new metric has never been presented or accepted by any lighting technical society. The metric is not described adequately to allow for any verification of the validity of the metric. In addition, the data collected shows that the metric is not at all reliable. The raw data for this metric indicates that a 1000-watt sport light flood can be less of a glare source than a luminaire with a 13-watt compact fluorescent lamp. The collected data and the generated metric are often in conflict with the subjective glare ranking by the surveyor.
The Statistical Data presented has many errors ranging from simple addition and division errors to the misapplication of statistical methodologies. There are an overwhelming number of tables and charts, many without the units identified, included in this report. Many of the tables report percentages that are not mathematically possible given the sample sizes. In addition, some of the statistical methodologies are briefly explained but the significance of some of the results are not discussed in relative terms.
By way of example, Table 1 reports 90% confidence interval of 2381 GWh to 3754 GWh. This represents a 45% margin of error, which is not very conclusive. There is also a subtraction error in this table; the lower limit, given the estimated value and the error bound should be 2380. To put this 1374 GWh range into perspective, this range is equal to the annual energy consumption for over 225,000 American households based on Howard Geller's estimate of 6100 KWh per year, or the annual energy consumption of 1,250,000 street lights with 250 watt lamps.
This large margin of error could be partly due to a problem in the calculation of unit power density for some of the functional use areas. Large error bounds can also occur when the data does not adequately fit a normal distribution indicating the sample size is inadequate for the population.
The Lighting Satisfaction Survey administered as part of this research has problems both in its design and administration that make the data collected and the conclusions reached inappropriate. The survey employs a two-valued logic system that is unsuitable for scientific subjective analysis and does not include some important information about the respondent and the site that could significantly alter the interpretation of the data. The main problem with the survey, however, is in the way that it was administered. In most instances there was only one survey administered at each site. The opinion of one individual is just that, an opinion and cannot be construed as the opinion of the entire population. Unless three or more opinions about the same site are obtained the data is meaningless and no conclusions about the site can be inferred. Then to correlate one opinion about one site and another opinion about a different site just propagates the probability that the results do not represent the population.
Finally, the Lamp Replacement Scenario that is presented misinterprets Dr. Sam Berman's referenced research by inferring that research done under normal interior lighting levels can be applied to exterior lighting. There is no generally accepted research that concludes that this is the case. In addition, when assessing the impact of any scenario in a model, the reliability of the baseline is extremely important. In this case the baseline is not very well established (the 90% confidence interval indicates a 45% margin of error) hence any projected impacts will have a correspondingly high margin of error.
While there appears to have been lot of data collected and presented in this report, the validity of the conclusions drawn from the data are questionable and the information is not sufficiently reliable to be used as the basis for legislation.
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
