Assessing Aircraft Noise Conditions Affecting Student Learning, Volume 2: Appendices (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

APPENDIX D. Development of Alternative Research Designs D.1. Datum—Macro-Analysis (Top 60 Airports) Description The plan is to conduct a nationwide macro-analysis of the relationship between noise exposure and student performance taking into account the effect of school sound insulation and other confounding factors. We will use the top 60 airports on the US MAGENTA airports list; sorted by the number of schools exposed to DNL 55 dB and higher. Our student performance measure is the standardized test scores (reading and mathematics) available from the NLSLSACD. Outcomes This plan answers the project research questions as follows: 1. To what extent is student learning affected by aircraft noise? We will examine the exposure-effect association between aircraft noise level and standardized test scores to quantify the magnitude of the noise-induced impairment and to discover a statistically significant relationship between test scores and aircraft noise. However, there is a chance that we will not find an effect above DNL 65 due to small sample of schools. 2. What is the most appropriate noise metric for describing aircraft noise as it affects learning? Through modeling, we will have a variety of aircraft noise metrics to analysis with standardized test scores to find the best metric-score correlation. 3. What is the threshold above which the effect is observable? The critical statistic to answer this question will be the differences in mean test scores between target schools at varying levels of aircraft noise and control schools (not exposed to aircraft noise). The key assumption in being able to answer this question is that the effect of aircraft noise on learning becomes significant near or above DNL 55 dB. 4. Has insulation meeting existing classroom acoustic criteria improved student achievement? Two analyses will supply the answer. A before-and-after analysis will provide the difference in mean test scores before and after insulation. A comparison with control schools will provide the difference in mean scores between insulated schools at varying levels of aircraft noise and control schools. 5. How does aircraft noise affect learning for students with different characteristics? The answer will come from analyses of subpopulations of students in each school, by race, gender, poverty level, grade, English proficiency, learning disability, and proficiency on the standardized tests when these subpopulation sizes are sufficient. Methods 1. Airport Selection The scope of this effort is the top 60 airports from the FAA’s US MAGENTA list found in the Overview as ranked by the number of public schools exposed to DNL 55 dB or higher in 2000. D-1

2. School Selection School information will be obtained from the CCD. After a preliminary examination of our databases, we find the numbers of target schools (public schools exposed to aircraft noise) around the top 60 airports are as follows: Noise Bin # Schools DNL 55-60 694 DNL 60-65 240 >DNL 65 76 Total 1010 We also expect to capture 99% of the insulated schools at these top 60 airports. 3. Student Performance Measure We will use school-level student test scores from the NLSLSACD. We will focus on Grades 3-8 since the 2002 No Child Left Behind Act (NCLB) calls for testing children in these grades every year. 4. Noise Measure As stated in the Overview, we will examine the following exterior noise metrics: • Arithmetic Average LAmax • Energy Average SEL • Leq(School) • TA(55), TA(60), TA(65), TA(70), TA(75), and TA(80) • NA(55), NA(60), NA(65), NA(70), NA(75), and NA(80) 5. School Characteristic and Student Population Measures We will draw these variables from the NLSLSACD and the ED CCD databases. 6. Analytical Techniques According to our preliminary power analysis (see Appendix D.6), we have sufficient sample sizes of target schools below DNL 65 dB to find an effect; but not above DNL 65 unless the actual size of the effect is much larger than our estimate based on the RANCH finding. At 95% confidence interval; we estimate that the probability of not finding an effect above DNL 65 (when it exists, type II error) increases by almost 20%. Plan Assessment Pros Cons • Largest sampling of schools for a study of this kind, which should produce more precision (and confidence) in drawing inferences about the effect. • The power analysis supports the probability that the study will find a statistically significant relationship where such a relationship exists. • Data gathering and analysis workloads fit • No insight into the mechanism of how aircraft noise affects learning. • Above DNL 65, probability of type II error is around 40% unless actual effect is much larger than the RANCH finding. • No follow-up study on what makes atypical schools different, which would have provided insight into any study design issues. D-2

Pros Cons within the budget. • Provides quantitative answers for the first five research questions. D.2. Alternative—Macro-Analysis (Top 50 Airports) with Follow-up Analysis Description This is the same type of macro-analysis as the Datum except we will use the top 50 airports from the US MAGENTA list instead of the top 60. We shift resources in order to conduct a follow-up study. We will follow up the macro-analysis with a more detailed examination at a small sample of schools that the analysis identifies as atypical. Outcomes This plan answers the project research questions as follows: 1. To what extent is student learning affected by aircraft noise? The analysis to derive the answer is the same as the Datum and the probability of a type II error is the same. 2. What is the most appropriate noise metric for describing aircraft noise as it affects learning? The analysis to derive the answer is the same as the Datum. 3. What is the threshold above which the effect is observable? The analysis to derive the answer is the same as the Datum. The key assumptions are the same. 4. Has insulation meeting existing classroom acoustic criteria improved student achievement? The analysis to derive the answer is the same as the Datum. 5. How does aircraft noise affect learning for students with different characteristics? The analysis to derive the answer is the same as the Datum. 6. What other knowledge will be gained by this research? Unlike the Datum, the follow-up detailed examination at a selected small sample of schools will produce information on what is it about these schools that make them atypical. This should provide insight into the capability of the study design to account for confounding factors that can influence the exposure-effect association. Methods 1. Airport Selection The scope of this effort is the top 50 airports from the FAA’s US MAGENTA list found in the Overview as ranked by the number of public schools exposed to DNL 55 dB or higher in 2000. D-3

2. School Selection School information will be obtained from the CCD as noted in the Overview. After a preliminary examination of our databases, we find the numbers of target schools (public schools exposed to aircraft noise) around the top 50 airports are as follows: Noise Bin # Schools DNL 55-60 662 DNL 60-65 234 >DNL 65 76 Total 972 We also expect to capture 97% of the insulated schools at these top 50 airports. 3. Student Performance Measure We will use school-level student test scores from the NLSLSACD. We will focus on Grades 3-8 since the 2002 No Child Left Behind Act (NCLB) calls for testing children in these grades every year. 4. Noise Measure We will examine the same exterior noise metrics as the Datum. 5. School Characteristic and Student Population Measures We will draw these variables from the NLSLSACD and the ED CCD databases. 6. Analytical Techniques We will perform the same macro-analysis as the Datum. According to our preliminary power analysis (see Appendix D.6), we have sufficient sample sizes of target schools below DNL 65 dB to find an effect; but not above DNL 65 unless the actual size of the effect is much larger than our estimate based on the RANCH finding. At 95% confidence interval, we estimate that the probability of not finding an effect above DNL 65 is the same as the Datum. Follow-Up Analysis The macro-analysis should produce relationships between aircraft noise and student performance based on our analysis model. The statistical analysis should also reveal data points that deviate markedly from the other data points, which we label atypical schools. We define these atypical schools as falling ±2.5 standard deviations (s.d.) from the mean. Our preliminary estimate is that the size of the atypical sample will be less than twenty schools. We will conduct a follow-up analysis to try to understand why these atypical schools exist. This will require a more detailed look at these atypical schools beyond the databases we used in the macro-analysis. We will look for erroneous data, such as, incorrect coding in the databases we used. We will also look for differences about the atypical schools not captured in the school or student population characteristics we used for examination. For example, we could turn up information indicating that an atypically high performing school was sound insulated some years ago that is not reflected our data. Or we could find that an atypically low performing D-4

school conducts most of the classes in temporary buildings with little sound insulation. We will also examine the role of any limitation in the aircraft noise modeling, such as, gross over- or under-prediction of aircraft noise levels due to incorrect model input assumptions. The idea is to look for patterns or trends in the information on the atypical schools and the neighboring airport that would help in future study designs. As this is a very labor intensive effort, the follow up will be limited to a handful of atypical airports. Plan Assessment Pros Cons • Data gathering and analysis workloads fit within the budget. • Provides quantitative answers to the first five research questions. • The power analysis supports the probability that the study will find a statistically significant relationship where such a relationship exists. • The follow-up analysis should reveal weaknesses in the research design. • No insight into the mechanism of how aircraft noise affects learning. • Above DNL 65, probability of type II error is around 40% unless actual effect is much larger than the RANCH finding. D.3. Alternative 2—Macro-Analysis (Top 40 Airports) with Observation Case Study Description This is the same type of macro-analysis as the Datum and Alternative 1 except we will use the top 40 airports from the US MAGENTA list instead of the top 60 or 50, respectively. We shift resources in order to conduct a case study. In the case study, we will observe changes in classrooms when exposed to aircraft noise and measure the noise events. Outcomes This plan answers the project research questions as follows: 1. To what extent is student learning affected by aircraft noise? The analysis to derive the answer is the same as the Datum and Alternative 1 and probability of type II error is the same. 2. What is the most appropriate noise metric for describing aircraft noise as it affects learning? The analysis to derive the answer is the same as the Datum and Alternative 1. With the added case study, we will use the classroom observations and aircraft noise measurements to discover which noise metric best matches up with the degree that aircraft noise disrupts the classroom environment; looking for confirmation of the correlation finding from the macro-analysis. 3. What is the threshold above which the effect is observable? The analysis to derive the answer is the same as the Datum and Alternative 1. 4. Has insulation meeting existing classroom acoustic criteria improved student achievement? D-5

The analysis to derive the answer is the same as the Datum and Alternative 1. 5. How does aircraft noise affect learning for students with different characteristics? The analysis to derive the answer is the same as the Datum and Alternative 1. 6. What other knowledge will be gained by this research? Unlike the Datum, which focused on answering the first five questions, and Alternative 1, which added a follow-up study of atypical schools; this plan includes a case study to provide insights into the mechanisms of the aircraft noise impacts upon classroom learning. Methods 1. Airport Selection The scope of this effort is the top 40 airports from the FAA’s US MAGENTA list found in the Overview as ranked by the number of public schools exposed to DNL 55 dB or higher in 2000. For the case study, we will choose a single airport; one with a high frequency and mix of aircraft operations with several schools nearby. Los Angeles (LAX) and Miami (MIA) International Airports are the leading candidates. 2. School Selection School information will be obtained from the CCD as noted in the Overview. After a preliminary examination of our databases, we find the numbers of target schools (public schools exposed to aircraft noise) around the top 40 airports are as follows: Noise Bin # Schools DNL 55-60 624 DNL 60-65 219 >DNL 65 74 Total 917 We also expect to capture 95% of the insulated schools at these top 40 airports. For the case study, we will choose one elementary school nearby the airport selected. In the pre- selection process, we identify candidate schools around both LAX and MIA. An important consideration in the school selection are the processes to obtain school district cooperation and participation and then to obtain parent permission and informed consent. There also the need for an institutional review board (IRB) approval for a study of this kind. Dr. Hervey, as a member of the faculty of Liberty University, has experience with their IRB process (https://www.liberty.edu/index.cfm?PID=12606). Since minors are involved, this case study automatically comes under the expedited or full review process, which takes about 2 months; after which we would begin the process of approval with the school district. 3. Student Performance Measure We will use the same school-level student test scores (from NLSLSACD) as the Datum and Alternative 1 for the macro-analysis. In the case study, as was done in the Crook and Langdon 1974 study at schools close to London Heathrow, we will categorize classroom interruptions due to aircraft events. We plan to videotape the classroom sessions for later analysis. D-6

4. Noise Measure We will have same INM input files and ETMS operational data as the Datum and Alternative 1 and can calculate the same noise metrics as we proposed in those other candidates. For the case study, we will conduct measurements of aircraft noise outside and inside the school classrooms where observations are being performed – see below. Digital time-histories of noise levels will be obtained for subsequent analysis and correlation with observations. 5. School Characteristic and Student Population Measures We will draw, from NLSLSACD and ED CCD databases, the same characteristics for analysis as the Datum and Alternative 1. 6. Analytical Techniques We will perform the same macro-analysis as the Datum and Alternative 1. According to our preliminary power analysis (see Appendix D.6), we have sufficient sample sizes of target schools below DNL 65 DB to find an effect; but not above DNL 65 unless the actual size of the effect is much larger than our estimate based on the RANCH finding. At 95% confidence interval; we estimate that the probability of not finding an effect above DNL 65 is the same as the Datum and Alternative 1. Observation Case Study Like the Crook and Langdon study, we will collect observations of how aircraft noise disrupts the classroom, such as, teacher pauses, teacher speech masking, pupil speech pause and masking, classroom behavior, and pupil distraction. While Crook and Langdon plotted frequency of disruptions, such as, teacher pauses against peak aircraft noise level; we will examine other metrics (TA, NA, etc.) that we can derive from the noise measurements. Plan Assessment Pros Cons • Data gathering and macro-analysis workloads fit within the budget. • Provides quantitative answers to the first five research questions. • The power analysis supports the probability that the study will find a statistically significant relationship where such a relationship exists. • The case study should provide insight into the mechanisms of how aircraft noise affects classroom learning. • The case study should also confirm findings of the macro-analysis on the best noise metric to represent the relationship with test scores. • Above DNL 65, probability of type II error is around 40% unless actual effect is much larger than the RANCH finding. • Process to obtain school cooperation and parent process for the case study might not fit in the project schedule. • No information on whether case study is representative. • No follow-up study on what makes atypical schools different, which would have provided insight into any study design issues. D-7

D.4. Alternative 3—Macro-Analysis (Top 30 Airports) with Follow-up Analysis and Case Study Description This is the same type of macro-analysis as the Datum and Alternatives 1 and 2 except we will use the top 30 airports. We shift resources in order to conduct both a follow-up study and observation case study. The follow-up analysis is like Alternative 1. The case study is the same as Alternative 2. Outcomes This plan answers the project research questions as follows: 1. To what extent is student learning affected by aircraft noise? The analysis to derive the answer is the same as the Datum and Alternatives 1 and 2. Chance of type II error above DNL 60 is slightly higher than the other candidates mentioned. 2. What is the most appropriate noise metric for describing aircraft noise as it affects learning? Like the Datum and Alternatives 1 and 2, we will analyze a variety of aircraft noise metrics to find the best metric-score correlation. Through the case study, we will use the classroom observations and aircraft noise measurements to discover which noise metric best matches up with the degree that aircraft noise disrupts the classroom environment; looking for confirmation of the correlation finding like Alternative 2. 3. What is the threshold above which the effect is observable? The analysis to derive the answer is the same as the Datum and Alternatives 1 and 2. 4. Has insulation meeting existing classroom acoustic criteria improved student achievement? The analysis to derive the answer is the same as the Datum and Alternatives 1 and 2. 5. How does aircraft noise affect learning for students with different characteristics? The analysis to derive the answer is the same as the Datum and Alternatives 1 and 2. 6. What other knowledge will be gained by this research? Like the Alternative 1, the follow-up analysis will help us understand the characteristics of the atypical schools. Like Alternative 2, the case study will provide insights into the mechanisms of the aircraft noise impacts upon classroom learning. Methods 1. Airport Selection The scope of this effort is the top 30 airports from the FAA’s US MAGENTA list found in the Overview as ranked by the number of public schools exposed to DNL 55 dB or higher in 2000. D-8

For the case study, we will choose a single airport; one with a high frequency and mix of aircraft operations with several schools nearby. Los Angeles (LAX) and Miami (MIA) International Airports are the leading candidates. 2. School Selection School information will be obtained from the CCD as noted in the Overview. After a preliminary examination of our databases, we find the numbers of target schools (public schools exposed to aircraft noise) around the top 30 airports are as follows: Noise Bin # Schools DNL 55-60 576 DNL 60-65 199 >DNL 65 70 Total 845 We also expect to capture 95% of the insulated schools at these top 30 airports. We will choose one elementary school for the case study nearby the airport selected as we would in Alternative 2 involving the same IRB, school district, school and parent approval processes. 3. Student Performance Measure We will use the same school-level student test scores (from NLSLSACD) for the macro- analysis as the Datum and Alternatives 1 and 2. The case study captures the same classroom observations as Alternative 2. 4. Noise Measure We will have same INM input files as the Datum and Alternatives 1 and 2 to calculate the same noise metrics as we proposed for these other candidates. For the case study, the noise measurement protocol is the same as Alternative 2. 5. School Characteristic and Student Population Measures We will draw from NLSLSACD and ED CCD databases the same characteristics for analysis as the Datum and Alternatives 1 and 2. 6. Analytical Techniques We will perform the same macro-analysis as the Datum and Alternative 1. According to our preliminary power analysis (see Appendix D.6), we have sufficient sample size of target schools at DNL 55-60 to find an effect; but fall a little short at DNL 60-65. At 95% confidence interval, we estimate that the probability of not finding an effect at DNL 60-65 is about 2% higher than the Datum or Alternatives 1 and 2. Our sample size above DNL 65 is about the same as Datum and Alternatives 1 and 2. At 95% confidence interval, we estimate that the probability of not finding an effect above DNL 65 is about the same as the Datum and Alternatives 1 and 2. Follow-Up Study The follow-up study is like Alternative 1 involving less than twenty atypical schools. D-9

Case Study The case study is the same as Alternative 2. Plan Assessment Pros Cons • Data gathering and macro-analysis workloads fit within the budget. • Provides quantitative answers for the first five research questions. • The power analysis supports the probability that the study is likely find a statistically significant relationship where such a relationship exists at lower noise levels. • The follow-up analysis should reveal weaknesses in the research design. • The case study should provide insight into the mechanisms of how aircraft noise affects classroom learning from the case study. • At DNL 60-65, probability of type II error is 2% higher than Datum, Alternatives 1 and 2 unless actual effect is much larger than the RANCH finding. • Above DNL 65, probability of type II error is around 42% unless actual effect is much larger than the RANCH finding. • Process to obtain school cooperation and parent process for the case study might not fit in the project schedule. • No information on whether case study is representative. D.5. Alternative 4—Macro-Analysis (Top 15 Airports) with Follow-up Analysis and Expanded Case Study Description This is the same type of macro-analysis as the Datum and Alternatives 1, 2, and 3 except we will use the top 15 airports. We shift resources in order to conduct both a follow-up analysis and expanded case study. The follow-up analysis is like Alternatives 1 and 3. The case study involves classroom observations as proposed in Alternatives 2 and 3, but now includes two schools with the addition of student and teacher questionnaires given through focus groups. Outcomes This plan answers the project research questions as follows: 1. To what extent is student learning affected by aircraft noise? The analysis to derive the answer is the same as the Datum and Alternatives 1, 2, and 3. However, the probability of finding an effect is substantially less than the other candidates mentioned. 2. What is the most appropriate noise metric for describing aircraft noise as it affects learning? Like the Datum and Alternatives 1, 2 and 3, we will analyse a variety of aircraft noise metrics to find the best metric-score correlation. Through the case study, we will use the classroom observations and aircraft noise measurements to discover which noise metric best matches up with the degree that aircraft noise D-10

disrupts the classroom environment; looking for confirmation of the correlation finding like Alternatives 2 and 3. 3. What is the threshold above which the effect is observable? The analysis to derive the answer is the same as the Datum and Alternatives 1, 2, and 3. However, chance of finding the effect is less than the other candidates mentioned. 4. Has insulation meeting existing classroom acoustic criteria improved student achievement? The analysis to derive the answer is the same as the Datum and Alternatives 1, 2, and 3. However, the probability of finding an effect is substantially less than the other candidates mentioned due to a much smaller sample of insulated schools. 5. How does aircraft noise affect learning for students with different characteristics? The analysis to derive the answer is the same as the Datum and Alternatives 1, 2, and 3. However, the probability of finding an effect is substantially less than the other candidates mentioned. 6. What other knowledge will be gained by this research? Like the Alternative 1, the follow-up analysis will help us understand the characteristics of the atypical schools. Like Alternative 2, the case study will provide insights into the mechanisms of the aircraft noise impacts upon classroom learning. Through the focus group questionnaire, we will discover what students and teachers perceive to be the effects of aircraft noise. Methods 1. Airport Selection The scope of this effort is the top 15 airports from the FAA’s US MAGENTA list found in the Overview as ranked by the number of public schools exposed to DNL 55 dB or higher in 2000. For the case study, we will choose a single airport; one with a high frequency and mix of aircraft operations with several schools nearby. Los Angeles (LAX) and Miami (MIA) International Airports are the leading candidates. 2. School Selection School information will be obtained from the CCD as noted in the Overview. After a preliminary examination of our databases, we find the numbers of target schools (public schools exposed to aircraft noise) around the top 15 airports are as follows: Noise Bin # Schools DNL 55-60 437 DNL 60-65 154 >DNL 65 59 Total 650 We also expect to capture only 68% of the insulated schools at these top 15 airports. We will choose a target elementary school and a control elementary school for the case study in the D-11

vicinity of the airport selected. As with Alternatives 2 and 3, this part of the plan requires IRB, school district, school and parent approval processes. 3. Student Performance Measure We will use the same school-level student test scores (from NLSLSACD) for the macro- analysis as the Datum and Alternatives 1, 2, and 3. The case study captures the same classroom observations as Alternatives 2 and 3, but now we can compare any difference in-classroom behavior between a target school and a control school. Through use of student and teacher focus groups, we will gather information on their perceptions regarding the influence of aircraft noise on such effects as: • Student stress • Teacher stress • Teacher’s vocal strain and fatigue • Annoyance • Attitudes For the focus group questionnaire, we intend to adapt questions from the form used by Haines et al in the West London Schools Study cited in the literature review. 4. Noise Measure We will have same INM input files as the Datum and Alternatives 1, 2, and 3 to calculate the same noise metrics as we proposed for these other candidates. For the case study, the noise measurement protocol is the same as Alternatives 2 and 3. 5. School Characteristic and Student Population Measures We will draw from NLSLSACD and ED CCD databases the same characteristics for analysis as the Datum and Alternatives 1, 2, and 3. 6. Analytical Techniques We will perform the same macro-analysis as the Datum and Alternatives 1, 2, and 3. According to our preliminary power analysis (see Appendix D.6), we do not have sufficient sample sizes of target schools to find an effect. At 95% confidence interval, we estimate that the probability of not finding an effect at DNL 55-60 is about 2% higher than the Datum and Alternatives 1, 2, and 3; 13% higher at DNL 60-65 and 30% higher above DNL 65. The sample of insulated schools also falls well short of what is needed. Follow-Up Study The follow-up study is like Alternatives 1 and 3 involving less than twenty atypical schools. Case Study The case study expands on the case study concepts of Alternatives 2 and 3 with the addition of a control school and focus groups to try to answer the following: • How do students who are regularly exposed to aircraft noise at school differ from similar students who are not exposed to aircraft noise at school with respect to inhibitory factors including distraction, learned helplessness, memory difficulties, hearing and auditory processing difficulties, stress, health difficulties, noise annoyance, and absenteeism? D-12

• How do teachers who are regularly exposed to aircraft noise at school differ from teachers who are not exposed to aircraft noise at school with respect to inhibitory factors including stress, health difficulties, noise annoyance, absenteeism, and vocal strain? • According to students and teachers, how, if at all, does aircraft noise influence teaching and learning? Plan Assessment Pros Cons • Data gathering and macro-analysis workloads fit within the budget. • The follow-up analysis should reveal weaknesses in the research design. • The case study should provide insight into the mechanisms of how aircraft noise affects classroom learning. • The case study could confirm findings of the macro-analysis on the best noise metric if the macro-analysis finds a statistically significant relationship with test scores. • The focus group portion of the case study should provide student and teacher perspectives on the problem. • Do not have statistical confidence that the macro-analysis will be able to provide quantitative answers to the first five research questions. • Process to obtain school cooperation and parent process for the case study might not fit in the project schedule. • No information on whether case study is representative. D.6. Power Analysis For the macro-analysis, we will conduct power analyses to determine if the target school sample sizes are sufficient to provide statistically significant results. The parameters for the power analysis are: statistical significance (α, probability of Type I Error), effect size (z), and power of the test (β, probability of Type II Error). For significance, the confidence interval was set at 95% and a two-tailed α set at 0.05 was used. For power, probability was set at 80% (β=0.20).. The effect size is the minimum deviation from the null hypothesis that it hoped to detect. The RANCH study (Stansfeld 2005) found that adjusted mean reading z score (at 95% confidence interval) fell below zero at exposure greater than 55 dB LAeq16 and the relationship was linear at exposures less than 55 dB (see Figure 1 in Chapter 2). The effect found in the RANCH study is approximately 0.05 standard deviations per 5 dB change in aircraft noise. The RANCH study only included schools in the general vicinity of airports, whereas the present study will include many control schools located away from the nearest large airport. Assuming that the average airport noise difference between the lowest level at which detection is desired and the average of control schools is double the differences found in the RANCH study, leads our choice of a the value of z=0.1σ for this study. Other key assumptions for the power analysis are as follows: • Other than the known insulated schools, schools have equivalent differences between exterior and interior noise levels. D-13

• There is no “state” effect. Test scores are standardized to the same mean and standard deviation in each state in order to enable aggregation of results across states. • Differences in scores between schools at different airport noise exposure levels can be attributed to the effects of noise. Differences in achievement due to demographic and resource differences between schools at different airport noise levels are eliminated by statistical adjustment. • The availability of scores for multiple years (instead of a single year) for most schools reduces the standard errors of estimates by approximately 40 percent.1 This reduction in the standard error translates into the need for a smaller sample of schools. The preliminary power analysis indicated the need for a minimum sample size of 470 to detect an achievement difference of 0.1σ between schools with airport noise exposure between 55 dB and 60 dB and control schools in order to answer Research Questions 1, 2, 3, and 5. For higher noise levels, the required numbers of target schools are smaller, because the expected true value of z, the airport noise effect, is correspondingly larger. If for the interval from 55dB to 60 dB the value of z is 0.1σ, the value for the interval from 60dB to 65 dB might be 0.15σ, the value for the 65 – 70 dB interval might be 0.2σ, and 0.25σ for the interval 70 to 75dB. The corresponding minimum sample sizes would be approximately 210 at DNL 60-65, 120 at DNL 65-70, and 80 at DNL 70-75 to answer these same research questions. For the effect of insulation (Research Question 4), the critical statistic for the first part of this question is the mean difference between achievement scores after insulation and scores in the same schools before insulation. Because scores in the same school are correlated from one year to the next, the standard error of the difference is roughly 10 percent smaller than the standard error of the mean score before insulation.2 The preliminary power analysis suggests a sample of around 300 insulated schools assuming that we are trying to detect the effect when the true insulation effect is 10dB. The preliminary power analyses indicate that the research plan candidates fall short of meeting sample size minimums to varying degrees. However, the power analysis was based on previous research involving only Leq-based aircraft noise metric. The current study is planned to explore metrics that are distinctly different from Leq in hopes of finding one that has a better relationship with learning. Thus, the preliminary estimates of sample size requirements could be viewed as a worst case scenario. 1 Evidence to support this assumption is based on Don McLaughlin’s analyses of grade 4 reading scores for schools in two states and 5 years, California and Illinois in 1998-1999 through 2002-2003. 2 Ibid. D-14