Loading...
HomeMy WebLinkAboutAgenda - 08-10-95 - 3 (No abstract sheet available) ITEM #3 1 MEMORANDUM • TO: OHN K, COUNTY MANAGER FROM: ONES, PURCHASING AND CENTRAL SERVICES DIRECTOR RE: JAIL/COURTHOUSE PROJECT DISCUSSION OF OPTIONS DATE: AUGUST 1. 1995 1 have spent the last month exploring any and all perceived options to the dilemma we find ourselves in on the Jail/Courthouse project. I have attempted to put together a report which will provide the following information: • Background information of project need; • An articulation of options we have explored and the results of our analysis; • Funding options for the additional money needed for the project; • Recommendation. Background There are two components of this project, the Jail additional and the addition of one courtroom for use by Superior Court. Courtroom This project was necessitated by the extreme overcrowding in the Court system, particularly as it regards the District Court system. The implementation of Administrative Court on Fridays has reduced the size of the Monday/Wednesday courts, however, the Friday courts are far in excess of the District courtroom capacity(82). While the District and Superior Court Judges have worked well together to coordinate the use of the Superior Courtroom in the New Courthouse when District Court dockets were out of sight, the problem has not been eliminated, nor does it show signs of diminishing in the future. The courtroom to be built will be specifically for Superior C_.:urt and the large courtroom in the New Courthouse currently assigned for Superior Col.. will be designated for District Court. The Old Courthouse courtroom will continue to be used for Superior Court as well. Jail The problem with the Jail has gotten worse in the past few months. Average daily population is at 118; the rated capacity of the Jail is 71. The Sheriff has decreased the number of Federal inmates which he is holding, which in turn reduces the revenue we will receive. While the pre-trial release program funded through the Criminal Justice Partnership Act is in its infancy, there has been imperceptible impact of the program on the daily population to date. As is the case in the Court numbers, there are no signs of future population decreases. Overcrowding is not a matter to be dealt with lightly. It's impact makes us vulnerable in a number of areas. There may be increased incidences of altercations among inmates, staff stress is increased and the potential for other inmate situations is increased when the population is significantly over capacity. The entire $2,000,000 allocated to this project comes from the Federal Marshal's office and guarantees them 75 beds for 15 years. • JaWCourthouse Project 1 2 • Our Actions To Date The Jail and Courthouse projects were combined for bidding purposes. This proved a very successful approach in the Government Services Center/Tax and Records Renovation projects. A significant cost to any project is the construction administration (inspections of project by architects) and the supervision of the job by the contractor's superintendent. Since both jobs would be under one contractor, the supervision cost would be significantly decreased. Bids were solicited for this project first on May 23, 1995. Under NC Statues three bids per discipline (general, plumbing, electrical and mechanical) must be received in order to open. Response was insufficient to open. Bids were re-solicited for opening on June 8, 1995, at which time only one general contractor bid as a single prime contract. Under a single prime contract, the general is the only contractor and all others are subcontracted by him. Bids may be opened on the second solicitation regardless of the number received. Bids were opened with the resulting bid $3,944,684 in excess of the budget available; the Board rejected these bids on June 28, 1995. OPTIONS We have explored the options as articulated to the Board on June 28 as follows: Option 1. Split the project and re-bid. Advantages. • • You could probably get sufficient bidders for the Courthouse project. However, it continues to be doubtful that they would come in under the amount appropriated. Disadvantages. • Duplicate costs. To split the project will require duplicate effort on all personnel related fronts, including the architect's inspection schedule, the contractors superintendent set up and the County staff contract administration and inspection. Each of these equate to more dollars than would be the case under a combined project. 9 Staging area (materials stored for use on the project) would be more extensive. By having both projects under one contract the materials could be stored together, thereby requiring less area for staging. Since space is at a premium in this area, any opportunity to minimize the amount needed is an important consideration. • Time. Re-directing the specifications and drawings to split the projects and then to re- solicit bids would cost us at least an additional 60-90 days. This time table would put us into a questionable time of year for good building weather as well. Option 2. Wait to re-bid the project until a later date. Advantages. • We might luck out and find a hungry contractor if we wait long enough. However, it is important to note that correctional facility construction is a specialized field and not all contractors will be interested nor qualified to perform the service. • JaWCourthouse Project 2 3 • Disadvantages: • Construction costs. There is a very real possibility that construction costs could escalate further. The most recent trends seem to indicate that costs could increase by 8-12% by January 1, 1996 and then level off to inflationary increases thereafter. While this is a smaller increase than what we have seen in the past year, it is still an increase and would require even more money than what we have now. • Time. Overcrowding is the rule rather than the exception at the jail right now. As pointed out earlier in this document, we are averaging 118 inmates in a 71 inmate capacity jail. This consistent significant overcrowding makes us vulnerable to various inmate problems. It should be also noted that recent deaths in the jail, coupled with constant overcrowding could send a signal to the Prison Legal Services, who constantly monitor such inmate situations. The necessity to deal with inmate litigation would be a real personnel drain and one we can hopefully avoid if the new addition is constructed. Although we are out of compliance on some points of inspection, the State Jail Inspectors have to date been very understanding because they knew of our construction plans. This is a point of negotiation with them we do not wish to see jeopardized. • Revenue. Again, as I pointed out earlier in this document, the Sheriff has had to reduce the number of Federal inmates held in order to reduce the Jail population. The County is reimbursed $49.96 per inmate per day for each Federal inmate. As a point of reference, the loss of just five inmates per day would reflect a revenue decrease of over$90,000 per year. Option 3. Alternative Service Delivery in how the Project is built. • The intent of the General Statutes governing bidding is to ensure that competition creates the best circumstance under which to spend the Public's money on any given project. When it becomes clear that there is no competition and in fact the Public is being held hostage by the bidding process, one must explore what other options are available. The option for alternative service delivery which we have explored during the past month is that of a negotiated bid. The Institute of Government Faculty consulted support our position that a negotiated bid is acceptable as long as we are negotiating on the same scope of project for which bids were solicited. Grier-Fripp Architects has worked with Trafalgar House on other North Carolina prison construction projects. Trafalgar House Construction is British owned and is one of the largest construction companies in the world . As a point of reference for resources, they own the bonding company which would write the performance bond for the job. They have offices world- wide, including one in RTP. Don Combs, Senior Vice President of Operations, is responsible for operations east of the Mississippi River and lives in Hillsborough. Because of their high standing in the field, their experience with correctional institution construction and their boundless corporate resources, Trafalgar House Construction was requested to quote their cost for the project. They submitted a cost of$3,588,000 on the exact set of plans which yielded the $3.9 million bid. Jail/Courthouse Project 3 4 In addition, Trafalgar House suggested other areas where we could reduce cost without functional modification of the facilities. These are as follows: **J=Jail only; C=Courthouse only; B= Both projects J DESCRIPTION OF MODIFICATION ACCEPTABL NOT POSSIBLY Or E ACCEPTABL ACCEPTABLE C E CONTINGENT UPON SHER. Or OR ENGINEER B REVIEW C 1. Use acoustical ceiling in courtroom $1,250.00 B 2. Revised backfill material $3,000.00 J 3. Eliminate padded cell $10,000.00 B 4. Tap into existing utilitiesino temporary $4,500.00 service C 5. No cove lighting $3,200.00 J 6. Eliminate retaining wall; use landscaped $10,730.00 berm instead J 7. Eliminate trusses on Jail roof design.(HDC $98,312.00 approval would be required) J 8. Eliminate stainless steel plumbing fixtures $9,000.00 $10,000.00 (Only recommended in minimum security cells.) J 9. Change hot water capacity.(Continues to $20,000.00 guarantee sufficient hot water for kitchen use.) J 10. Eliminate Flemish bond on brick. $10,000.00 B 11. County pay for building permits.(Would $11,685.00 require BOCC approval?) B 12. Eliminate bond and accept a letter of credit in $27,814.00 lieu. J 13. Change electrical switchboard to $10,500.00 accommodate new addition only. B 14. Modify light fixtures/ballasts(increases $16,566.00 energy efficiency) J 15. Change out maximum security light fixtures $1,500.00 $1,500.00 (modify only the fixtures in minimum security cells) B 16. Modify HVAC system to manual dampers. $4,000.00 Totals $207,172.00 $34,700.00 $11,685.00 JaiUCourthouse Project 4 • 5 • Further Trafalgar House Construction indicates that the can build both projects in 10 months rather 9 Y P 1 than the 14 months it was estimated to take originally. The Architect indicates that with their enormous pool of resources this is quite possible for them to do. Advantages: One Contractor, reliable, experienced-irr Jail-construction, excellent resources. Economical purchasing by combining our materials purchases with other jobs they have on-going; • One contract to administer, so the lines of communication and responsible parties would be clear; • Project could stay close to the time line set forth earlier. • Construction now would save us money in the long term. • 10 month construction time rather than 14 month construction time would let us realize revenue from Federal inmates four months sooner. • The original scope of the project would be left intact. Disadvantages. • Would require more funds than are currently available. Option 4. Scratch both projects and wait to build under the recommendations brought forward by the Justice Facility Study Group. • Advantages: Can't identify a single one. Disadvantages: • Continued significant court and Jail overcrowding. Potential legal action by Judge against Commissioners to force Commissioners to build additional court facilities(Alamance County style). • Continued over crowding in Jail. • Lose$2 million in Federal Marshal funds; • Future construction would cost us more in the long run;the problems will not go away. Option 5. Are there current projects which are no longer valid from which funding could be shifted? I don't know of any, but wanted to consider any options that we have. FUNDING OPTION: Funds available in CIP now $2,208,552 Negotiated Bid amount w/$207,172 additional cuts $3,380,828 Deficit $1,172,276 We have pretty much exhausted the pay-as-you-go resources,so I have not even considered them in these calculations. We do have a stream of revenue however,which is attached permanently to this project. The CAP agreement with the Marshal's office is for 75 guaranteed beds per day for 15 years. I have spoken at length with the Marshal's office, both in Greensboro and in Washington D.C. and they indicate that they will stay as close to 75 prisoners in the Orange County Jail as possible. When Orange County opens they will • consolidate inmates held in other smaller surrounding Jails into ours. After our addition is completed their goal in the Middle District is to hold prisoners only in three Jails; Forsyth, Guilford and Orange Counties. Jail/Courthouse Project 5 The Marshal indicates that if the Middle District cannot use all of the beds,the Eastern District(Raleigh)will • be in a position to bring inmates to us. I have tried to demonstrate in the table below,if we sought private placement for the$1,200,000 funds necessary to complete the project,the revenue from 9 inmates per day will completely offset the debt service amount. Keep in mind, our potential revenue from the additional beds is$638,239 per year;this is over and above the revenue we currently receive from the Marshal's office,which was around$537,000 in f.y. 94-95. Assuming private placement of$1,200,000 the estimated payment schedule (per Ken Chavious)will be: Assumptions: 15 year term 6.75% interest rate $1,200,000 principal amount Semi-annual payments ::.::.::.::.::.;:.;:.;:.;.;.:..:.. ) }Pt> :..::.;::: PAE 'i":.::.;:.;:.;:.;:.:.:IRINIPAt� iNTE1EST.:::.: :Adol it rrr es::.>::.:::.;;: :...:::::::.:;:.:;:.:.;;:;;;::<:.:::::::.;:) [1 ::>:::::.: «.;:<::::::::.... .....::;:::.::;:<.;>:.;«;::::::::...:n ded a�y to t€ a> .;::>::: :««:>::»::::<:<:>::>:::: >::»::>:::<««<:»::»:........>::::::>::::>::::;:«:><:::>:»»::>::;:::<:»>:<::<:::>:<:«::>::;<:>::»::>:<::::<:>::<::>:»:>;;<>:::: ......:;::.>::>::>:;:::>:>::««<:.>;;::>::>::>: . ffset. eb. er e. os .. 1 $1,200,000.00 $161,000.00 $80,000.00 $81,000.00 9 2 $1,120,000.00 $155,600.00 $80,000.00 $751600.00 9 3 $1,040,000.00 $150,200.00 $80,000.00 $70,200.00 8 4 $960,000.00 $144,800.00 $80,000.00 $64,800.00 8 5 $880,000.00 $139,400.00 $80,000.00 $59,400.00 8 • 6 $800,000.00 $134,000.00 $80,000.00 $54,000.00 7 7 $720,000.00 $128,600.00 $80,000.00 $48,600.00 7 8 $640,000.00 $123,200.00 $80,000.00 $43,200.00 7 9 $560,000.00 $117,800.00 $80,000.00 $37,800.00 6 10 $480,000.00 $112,400.00 $80,000.00 $32,400.00 6 11 $400,000.00 $107,000.00 $80,000.00 $27,000.00 6 12 $320,000.00 $101,600.00 $80,000.00 $21,600.00 6 13 $240,000.00 $96,200.00 $80,000.00 $16,200.00 5 14 $160,000.00 $90,800.00 $80,000.00 $10,800.00 5 15 $80,000.00 $85,400.00 $80,000.00 $5,400.00 5 Totals $1,848,000.00 $1,200,000.00 $648,000.00 RECOMMENDATIONS: • Accept the negotiated bid with Trafalgar House Construction, accepting at least the additional cuts of$207,172 and consider waiving the$11,685 in building permits. • Board action to move the process forward at the August 15 BOCC meeting. (Trafalgar House has only offered to hold prices for 30 days,although I think we could negotiate a little past that, maybe to the first meeting in September.) • Seek private placement for the additional funds needed (approximately$1,200,000); • Use additional Federal revenue to pay debt service for the private placement funds. * This private placement would also need to indicate $500,000 in additional funds for the Southern Human Service Project. Jail/Courthouse Project 6 DRAFT A Study of Housing Characteristics and Student Generation Rates for Orange County, North Carolina Prepared for the Orange County Board of Commissioners by: Merritt Clapp-Smith Karen Kristiansson • August 8, 1995 Table of Contents • Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Data Sources and Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Student Generation Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Chapel Hill-Carrboro School District . . . . . . . . . . . . . . . . . . . . 10 Orange County School District . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Analysis by Subdivision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Comparison to Carrboro Subdivision Study . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Analysis of New Housing Owners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 • • Introduction Orange County, North Carolina, established an impact fee to help fund public school construction in the summer of 1993. The fee was calculated using data from the 1990 Census, and the calculations are described in detail in the Technical Report. Calculation of Proportionate Share Impact Fees for Financing Public School Capital Needs. As originally established, the fee was $750 for each unit constructed in the County. Early in the summer of 1995, the County Commissioners raised the impact fee to $1500 for housing units constructed in the Chapel Hill-Carrboro School District while keeping it the same for the Orange County School District. This study was commissioned in the summer of 1995 in order to address questions that arose both from the creation of the Technical Report itself and from public hearings. The main goal of this study was to determine the statistical relationship, if any, between the number of public school children in new housing units and the type, number of bedrooms, size, and value of that housing. For instance, do families with more children have • larger houses, houses with more bedrooms, or houses with a higher assessed value than families with fewer children? This kind of a relationship could be used to justify a schedule of public school impact fees that varies based on housing characteristics, rather than the current flat fee where each housing unit pays the same amount regardless of size or value. In order to perform this analysis, data was collected on all new housing units which received a Certificate of Occupancy between January 1, 1992 and December 31, 19941. Spring 1995 enrollment information for public school students was also collected', and this was matched to the housing unit data. The result was a data set containing a list of all new housing units in Orange County, their characteristics, and the number of public school children in each housing unit. Two separate data sets, one for the Chapel Hill-Carrboro School District and one for the Orange County School District, were created in this way. These data sets were then analyzed to test for the statistical relationships between each of the housing characteristics and the number of public school children in each unit. The tests performed tell whether or not there are statistically significant relationships between the • 'For the remainder of this report,the phrase"new housing"refers to this set of housing units. =For the remainder of this report,the terms"children"and"students"refer to children enrolled in the public school system only. housing characteristics and the number of children, and the strength of those relationships. Basic summary statistics, such as the average number of public school children per household (student generation rates) and the percentage of new housing units occupied by people without children in the public school system, were also compiled. Each of the steps used to attain the results are described in further detail in the body of the report. The key findings of this report are summarized below. • In the majority of statistical tests run for the report, the results were insignificant due to the large variation in the observations. • The only notable relationship between the number of public school children in new housing and housing characteristics found in this study is for the number of bedrooms in the Chapel Hill-Carrboro School District. This relationship indicates that there is approximately half a child more on average in new housing with four or more bedrooms than in new housing with three or fewer bedrooms. • The student e g neration rate calculated in this report for the Chapel Hill- Carrboro School District is higher than that calculated in the Technical Report, but the rate for the Orange County School District is lower than in the Technical Report. • 2 • Data Sources and Accuracy Data was obtained from five main sources: the Chapel Hill-Carrboro School District, the Orange County School District, the Carrboro Planning and Inspections Department, the Chapel Hill Planning and Inspections Department, and the Orange County Data Processing Department. Each data source offered different types of data and varied in method of collection and accuracy, as detailed below. Data errors encountered were corrected where possible, and if not corrected, were accounted for in the analysis. A few housing unit records were missing data on one or more of the housing characteristics. In these cases, the records were omitted from the analysis of the missing characteristic but were included in the analyses for which the data was complete. When data for several characteristics was unavailable, the record was included in the summary characteristics but omitted from the statistical analyses. 1. Public School Children The school districts supplied information on students enrolled as of May 1995, including students' addresses, schools attended, and dates of entry into the school system. Data on children at private schools was not collected since those children do not attend the public school system for which the impact fees are collected. Incomplete addresses provided the only difficulties for this data. In the few cases where an address lacked a street number or apartment number, it was filled in when possible using telephone directories, data from the Orange County tax appraisal system, and by comparing last names of students and homeowners. In a handful of cases, incomplete data may have prevented a student from being appropriately matched with a new housing unit. 2. New Housing Units The Chapel Hill and Carrboro Planning and Inspections Departments provided information on all new residential units which received a Certificate of Occupancy in 1992, 1993, and 1994. The Orange County Data Processing Department supplied a similar list for • Orange County. Because the data for 1995 was in the midst of processing and was 3 incomplete, we did not include new units for this year. The data on new housing units appeared and resented only occasional problems with addresses, like those encountered in the . PP P Y P school children data. These were able to be corrected in almost every instance by using the taxmap number of the lot to find a complete address. 3. Number of Bedrooms For consistency, the number of bedrooms for most housing units in both school districts was taken from the County's tax appraisal database. An exception was made for the apartment complexes, for which the management supplied the data. 4. House Area (Square Footage) Estimates for the square footage of new housing units were gathered from the towns' building permit records and the County's tax appraisal system. Some of these estimates varied widely between sources, differing as much as 4,000 square feet or more for a single house. In order to improve the consistency of the data, the square footage of living area for all housing units was calculated based on data from the Orange County tax appraisal system. The area was determined by multiplying the area of the first floor of the main portion of the unit by the number of stories. 5. House Value Values for all units came from the County's tax appraisal database. The data represents the value of the main portion of the housing unit only, and does not include the value of the lot, of additions, or of accessory buildings. This data was available for all units excluding approximately 50% of the county's mobile homes. For these units, the cost of the mobile home was obtained from the building permit and was used as the value of the unit. • 4 • Analysis Summary Statistics One type of analysis done in this study is a set of summary statistics and student generation rates for each school district. These are reported in the tables. Cha e/Hi!/-Carrboro School District Single I Mobile Modular I Apart- Town- Duplex ::::: ._ Family Homes Homes ments houses ° > TAL<• NITS ; < »<€> : >' '"< >'I>»:9 ..... ;................. Units with Public 402: 4 0 26 1 .... ............. School Children ........---•................................. ...................... ...................... ...................... ...................... ...................... ....................... W;;: Units with No Public 524: 15 1: 29: 5 571 ' >«':<<: 3: : School Children ............................................. ....... ...................... ...................... % of Units with 43.4%: 21.1 %? 0.0%? 47.3%? 16.7%' 6.6%: Public School Children Y. Elementary 56.3%: 50.0%:: 70.5 /: 10 .0/ : 0.0/ <? >`.:.....9.::.:.:°:.a.: School Children Middle School 22.2%: 25.0%: - 15.9% 0.0% 0.0%I '>_< `<' dren ::.. t.......................... •--.......... ............... •.. ...:..::::.:.:::.;:.:.::;.::. High School 21.5%: 25.0%: - 13.6%: 0.0%: 100.0%i . Children :<>: « ........:: ... Oran a County School District Single Mobile Modular Apart- Town- Duplex <'AC> ' Family Homes Homes ments houses ........:: ::::.:;;.::.:::;.; rr.uwrr.u:�iYui:.w.r....�rrw.•+r.r..w.r �1r•••^•••r.w..r..w.rr•'vrrr..e..r.......rr..F rq,.....rr..w....:.:i:.:....rr.u...�....����:�i�i:::::i..'.ii.)w....i••••• .v.,f!>��:{1fffY^tfY!!.^tt�:. Units with Public 126: 108: 7: 15: 0 0 25ea School Children ......................._.................... ...................... ...................... ...................... ...................... ...................... ........................; :> : Units with No Public 413: 400: 18: 20� 0: 6 85 School Children t.•..•• •..,• ........................ ...................... ...................... ...................... ...................... ...................... ........................' s»` « % of Units with 23.4%? 21.3%' 28.0% 42.9%:I - 0.0°0 .... .. .. .. .. ..... Public School / �•����-����� ����q Children M "L1...�1•filt.l�>^# I�.............................�.�`f� .. .............1.. ...v;.�:.:;�.::...-.a.....a...,,.,.......,a...,,. ,.,a....,.....a.,.a... .,v.......,,....... ««:;�:«.:: .«.:;� },a. ... aat:.:. a.,yaau vaa.w.C�..........;•:;;; :;•::::;•: % Elementary School 66%: %:' 58%: - - Children ............................................. ...................... ...................... ...................... ...................... ..................... T ......................... »» . ..... % Middle School 15%T 19%? 25% Children ........................................... ...................... .....•-••--------..... ...................... ...---•------------... ............--------- ----------------...... :..;.:.:: . ,:; r. :.:::-: High School 19%: 19%: 17% - - Children • 5 In addition, the graphs on the following pages illustrate the distribution of families with distribution among housing units of • different numbers of public school children and the g g different types. The most surprising result of these analyses is that fewer than 50% of new housing units in both school districts are occupied by families with children currently in the public school system. The other units, however, may have children in private schools or pre- school children who will someday attend public schools. In addition, the percentage of units with children varies among the two school districts, with the percentage of units with children markedly higher in the Chapel Hill-Carrboro School District than in the Orange County School District. Most of the public school children in new housing units are elementary school children, with smaller numbers in middle school and high school. Compared to the distribution of all public school children in the 1994-1995 year, new housing units have a greater proportion of elementary school students than the population at large. This indicates that new housing units tend to have more elementary school students than average. • Student Generation Rates In spite of the large number of families in new housing units without children, there is still a high student generation rate for new housing units as a whole in the Chapel Hill- Carrboro School District: 0.63 children per unit. Comparing the student generation rates in the table below for different housing types may be misleading, however, since single family homes are the only residential type for which there is a wide base of data. One problem with the student generation rates for the Chapel Hill-Carrboro School District is the presence of two large developments which may not be typical. One is the Carol Woods Retirement Community, which makes up almost half of the duplexes built in 1992- 1994. These units were included in the student generation rate calculation but removed from the data set before performing the more advanced statistical analyses described below. The other development which may be atypical is the Dobbins Hill Apartment Complex, which is an affordable housing project. It was the only apartment complex built in the Chapel Hill-Carrboro School District during the three years of this study. This means that the student • 6 Orange County School District Distribution of Families with Different Numbers of Public School Children Among New Housing Units 900 800 - 700 - C 600 - sm 300 - 200 - 100 - 0 0 1 2 3+ Public School Children Distribution of Housing Types Among Families with Different Numbers of Public School Children 450 400 - 350 300 CIO Children El Child 250. 02 Children 200 3+Ch Idren 0 150 r SFR Mobile Duplex Modular Apt Residential Housing Type Source: New Housing Units are based on Certificates of Occupancy issued in 1992, 1993, or 1994. Public school children include all those enrolled in the Orange County school system in Spring 1995. Chape| HIl-CarbO[O School District Distribution of Families with Different Numbers of Public School Children Among New Housing Units Ko ��¥ Ko /. g � mo 9�» � � 7 \ 300 o aw : _ ) $ 200 ' &�m _M�m ��A 100 - 7 m « k ) m Q 0 1 2 » Public School Children Distribution of Housing Types Among Families wpm Different Numbers of Public School children 500 ,45o 400 . ao CIO Children \ 300 k ■IChild f ■2Children ; 30 2 ■+Children » 200 e z 15 / 50 A 7 . . «< sFR Mobile vA, t_6 a! Residential Housing Type � Note: New Housing Units are units that received aCertificate U Occupancy in 1992, 1993, or 1994. Public School Childre n includes all those enrolled in the Chapel +&Carb r School system h Spring 199E • generation rate calculated for apartments in this study may not be representative of new apartments as a whole. Chapel Hill/Carrboro School District Single Mobile ::Modular:: Apart- Town- Duplex Family Homes: Homes : ments houses AL �t1...... .......�>''`:� ':'::'_: `«>< >'�>:`: ><::::;':(�:6. ..... 0 ......f#.$ �..... .......:..E'1r07......,..,.d.lx3 Elementary School Students 0.38:;...... ....0....��.1: .......0....._0..0: .......0......5...6 _.....w0.:.1� 0.0.0 06 ..»»......... ..................i . : Middle School Students 0.15: 0.05 0.00: 0.13: 0.00: ................................:.................................. ................. ................. <«. : High School Students 0.14: 0.05: 0.00: 0.11 0.00: The overall student generation rate for new housing in the Chapel Hill-Carrboro School District is substantially higher than the estimate of 0.29 children per unit used in the Orange County Technical Report. The difference is due to at least two factors: the five year time difference between the two reports, and the fact that the Technical Report included all existing housing rather than focusing on new housing units. This means that the large number of housing units in the area which are occupied by University students (and have a much lower • public school student generation rate) were included in the Technical Report estimate but not the current study. In addition, the subdivision study described later in this report indicates that there may be a higher student generation rate for new housing than older housing, even within subdivisions. Oran a County School District Single : Mobile !Modular:: Apart- Town- Duplex fi ' ' vi�Family Homes: Homes :: ments houses TY_P >I :::+3.11 :9<. ::::::....:::: ..::..:..(�.. .. ..........:. .... ............ . _:::::.>;::... ......................................�............................ ..............w.r....._......w..»r......»..w..».....�..v..».......»...w y..»......».­..».y..»..w.............. ... chool Students -0.21: 0.19: 0.28: - 0.00: 0.00 < «<Q<2 .............. ........................................ .....••••••...... ...........t:;;:;z.>: Middle School Students 0.05: 0.06: 0.12: - 0.00: 0.00 ............................................................... ........,................ ................... ; ,High School Students 0.06 0.06 0.08 0.00 0.0 0 N <>Ot6 In contrast to the Chapel Hill-Carrboro School District, the Orange County School District has a much lower student generation rate for new housing at 0.31 children per unit. This is lower than the Technical Report estimate based on the 1990 census of 0.45 children per 7 housing unit. The difference in this case may be due to the time gap between 1990 and 1995, or some unknown factor. Student Generation Rate Comparison Technical Report Carrboro Study (1994) This Study (1995) 1990 data Orange County 0.45 - 0.31 School District Chapel Hill-Carrboro 0.29 0.68 0.63 School District Statistical Analysis In the second phase of analysis, statistical analysis techniques were used to determine the relationships between housing characteristics and the number of children in new housing units. Bonferroni (means difference) tests were carried out to test the relationship between the number of public school children and each of the housing characteristics. These tests calculate the difference in the averages (means) for each set of observations within a certain category. For instance, what is the difference between the average number of bedrooms for families with 0 children and for families with 3 children? The tests also consider whether the difference in the averages is meaningful. If the observations are similar to each other, the average is useful, and a difference in two averages from data like this will be meaningful, or 'significant.' In contrast, if the observations vary dramatically, then an average does not mean much because of the amount of variation in those observations. Although two averages from data like this may be quite different, this difference will be meaningless if there-are large variations in the data. For instance, examine the two samples shown in the table below. 8 • Number of Bedrooms for Two Samples Families Families Families Families Families Average lA & lB 2A & 2B 3A & 3B 4A & 4B 5A & 5B Sample A: 1 3 1 1 4 2 0 Children Sample B: 3 3 3 3 3 3 3 Children Sample A consists of five families with 0 children. Three of these families have housing units with 1 bedroom, one has a house with 3 bedrooms, and one has a house with 4 bedrooms. The average number of bedrooms for Sample A is 2 bedrooms per housing unit. Sample B consists of five families with 3 children. All five of'these families have housing units with 3 bedrooms. The average number of bedrooms for Sample B is 3 bedrooms per housing unit. Although the average number of bedrooms is higher for families with 3 children (Sample B) than for families with 0 children (Sample A), this difference is not meaningful because the data for Sample A has so much variation in it. Therefore, it would not be accurate to say that the families with 3 children tend to have housing units with one more bedroom than the housing units of families with 0 children, even though the average for Sample B is one greater than the average for Sample A. In this example, the wide variation in the data for families with 0 children makes the difference between the two averages 'insignificant.' In the majority of the tests run for this report, the results were insignificant due to the large variation in the observations. In cases where the data was significant, the difference in the averages was very small. The results of the Bonferroni means difference tests were confirmed by an advanced statistical test called a MANOVA. The MANOVA test was suggested and interpreted by the staff of the Statistical Analysis lab at the University of North Carolina's Institute for Research in the Social Sciences. • 9 Chapel Hill-Carrboro School District • For the Chapel Hill-Carrboro School District, the statistical analyses show a weak but statistically significant relationship between the number of children and the number of bedrooms, the area, and the housing value. 1. Residential Type and Number of School Children: Inconclusive Relationship The number of housing units that were not single family residences built in 1992, 1993, and 1994 were too few to be used as a representative sample for the statistical analysis of the relationship between residential type and average number of children. In addition, all of the apartment units built during the three years of the study were in a single affordable housing complex, and the majority of duplex units built were in one retirement community. Although it may be possible to determine a relationship between housing type and the number of students by using data from all existing housing units, this would not provide student generation rates for new housing units, which may differ from older units. • 2. Number of Bedrooms and Number of School Children: Weak Relationship The graph below shows the average number of bedrooms for housing units with different numbers of school children. The vertical bars indicate 95% Confidence Average Number of Bedrooms versus Number of Public School Children 4.10 E 3.90 A g + z m 3.70 3.50 + 0 1 2 3+ Num ber of Public School Children 10 Intervals, which indicate the level of variation in the data and show the range in which • the true population mean should fall. The Bonferroni test show s t hat there are statistically significant differences in the number of bedrooms between all family sizes except between families with 1 child and families with 3+ children, and between families with 2 children and with 3+ children. The following table shows the statistically significant differences in the average number of bedrooms for families with different numbers of public school children. Significant Differences in Average Number of Bedrooms by Number of Children 0 Children 1 Child 2 Children 3+ Children (Avg = 3.6 (Avg = 3.8 (Avg = 4.0 (Avg = 4.0 0 Children 0 -- -- -- (Avg = 3.6 1 Child 0.22 0 -- -- (Avg = 3.8 • 2 Children 0.44 0.22 0 -- (Avg = 4.0 3+ Children 0.43 Not Significant Not Significant 0 (Avg = 4.0 This table tells the statistically significant differences in average number of bedrooms for families with different numbers of children. For example, families with 0 children have an average of 3.6 bedrooms while families with 1 child have an average of 3.8 bedrooms, for a significant difference of 0.22 bedrooms. As you can see, most of the differences are significant. The question was then what the differences in average numbers of children would be for housing units with different numbers of bedrooms. To determine this, another Bonferroni test was run. The results are summarized in the table below. 11 Differences in Average Number of Children by Number of Bedrooms 2 Bedrooms 3 Bedrooms 4 Bedrooms 5+ Bedrooms (Avg = 0.26 (Avg = 0.42 (Avg = 0.78 (Avg 0.83 2 Bedrooms 0 -- -- -- (Avg = 0.26 3 Bedrooms Not Significant 0 -- -- (Avg = 0.42 4 Bedrooms 0.53 0.36 0 -- (Avg = 0.78 5+ Bedrooms 0.58 0.41 Not Significant 0 (Avg = 0.83 As this table shows, the only differences that are not significant are between units with 2 and 3 bedrooms and between units with 4 and 5+ bedrooms. The largest differences in average numbers of children is between houses with 3 and fewer bedrooms and houses with 4 or more bedrooms. Although most of the differences are significant they are quite small. • 3. Area of Housing Unit and Number of Children: Very Weak Relationship The graph below shows the averages with 95% confidence intervals. There are statistically significant differences in the square footage of units with 0 and 1 children, Average House Area versus Number of Public School Children 2,100 2,000 + 1,900 + 1,800 y1,700 + 0 1,600 = 0 1 2 3+ Number of Public School Children 12 • and units with 0 and 2 children. However, the differences are very small. The following table shows the statistically significant differences in the average house area for families with different numbers of public school children. Differences in Average Area b Number of Children 0 Children 1 Child 2 Children 3+ Children (Avg = 1740 (Avg = 1899 (Avg = 2010 (Avg = 1943 0 Children 0 -- -- -- (Avg = 1740 1 Child -158.95 0 -- -- (Avg = 1899 2 Children -269.66 Not Significant 0 -- (Avg = 2010 3+ Children Not Significant Not Significant Not Significant 0 (Avg = 1943 • 4 in Value and Number of Children: Very Weak Relationship . Housing ry p Average values for families with different numbers of students are shown on the graph below together with the 95% confidence intervals. The only statistically significant Average House Value versus Number of Public School Children $180,000 l RI $170,000 T 1 a $160,000 + T $150,000 + $140,000 0 1 2 3+ Number of Public School Children differences for housing value were between families with 0 and families with 2 • children, but the difference is quite small. The following table shows the statistically 13 significant differences in the average housing value for families with different numbers of public school children. • Differences in Average Replacement Value b Number of Children 0 Children 1 Child 2 Children 3+ Children (Avg = 150064 (Avg 160068 (Avg = 175088 (Avg = 166812 0 Children 0 -- — — Av g = 150064 1 Child Not Significant 0 — - -(Avg = 160068 2 Children 25024 Not Significant 0 — (Avg = 175088 3+ Children Not Significant Not Significant Not Significant 0 (Avg = 166812 Orange County School District For the Orange County School District, the Bonferroni statistical tests show no significant relationship between any of the housing characteristics and the number of public • school children in new housing units. Therefore, none of the housing characteristics analyzed- -number of bedrooms, area of the housing unit, house value, or housing type--can be used to predict the number of public school children in a housing unit. Appendix A shows graphs of the averages and the 95% confidence intervals for each of the housing characteristics. • 14 • Analysis by Subdivision In addition to examining the student generation rate for new housing in each school district as a whole, the student generation rates for a number of individual subdivisions in Chapel Hill and Carrboro were also calculated. These are summarized in the tables on the following page. The average student generation rate for new housing units in these subdivisions is 0.70 children per unit in Chapel Hill, although it ranges from 0.15 to 1.63 children, and 0.77 children per unit in Carrboro, with a range from 0.49 to 1.56 children. The methodology for this study and a detailed analysis of the results can be found below. Finally, the student generation rates for Carrboro are compared to the rates calculated by the Planning Director of Carrboro in April 1994. Methodology 1. Choosing the Subdivisions The subdivisions chosen generally included those with more than 30 lots and with • recent building activity. One or two subdivisions fitting this description may not have been included due to difficulties in determining the precise borders of the subdivision. For comparison, two subdivisions completed prior to 1992 were examined for Chapel Hill and one for Carrboro. 2. Determining the Borders of the Subdivisions Two steps were used to determine the precise boundaries of the subdivisions. First, taxmaps were examined for subdivision boundaries. Second, all the subdivisions were visually inspected and roads and addresses confirmed. 3. Delimiting Addresses for Subdivisions In most cases, a road within a subdivision begins and ends in that subdivision and all of its addresses are contained therein. For roads not completely enclosed in one subdivision, taxmaps identified which addresses fell within the subdivision borders. 4. Matching Students and Permits to Subdivisions Using the list of applicable roads and addresses for each subdivision, all public school children and new permits that matched those addresses were isolated and labeled by 15 their respective subdivisions. New housing units (issued a Certificate of Occupancy in 1992 1993 or 1994) were then matched with public school children. The result was a • list for each subdivision of all new housing units with students and without students, as well as all students living in housing units not built in 1992, 1993, or 1994. 5. Determining the Number of New and Existing Units The total number of new housing units was defined as all units issued a Certificate of Occupancy in 1992, 1993, and 1994. The total number of existing housing units was based on a visual count of occupied housing units within each subdivision as of July 1995. The count of existing units includes all new housing units. Results The average student generation rate for new housing units in the Chapel Hill subdivisions is 0.703, which is higher than the average rate of 0.632 children per unit for existing housing. This difference shows a slight trend of newer units having more children on average than older units. For most subdivisions within Chapel Hill and Carrboro, the trend of . higher student generation rates for newer units holds true. For Carrboro subdivisions as a whole, the student generation rate increased from 0.729 for existing units to 0.774 for new units. Following a similar pattern, the percentage of new units with public school children in Chapel Hill and Carrboro subdivisions is higher on average than the percentage of existing units with public school children. For Chapel Hill, 39.9% of existing units have public school children compared to 45.4% of new units. For Carrboro, subdivision housing with children was 46.5% of existing units and 51.6% of new units. The wide range in student generation rates and the percentage of units with public school children between subdivisions shows that each subdivision attracts a different mix of household types. The reasons behind these differences is difficult to determine and is outside the scope of this study. 16 • Sample of CHAPEL HILL SUBDIVISIONS ALL EXISTING HOUSING UNITS HOUSING UNITS BUILT IN 1992, 1993 OR 1994 Total New :Percentage: Total New ::Percentage,: Housing I of New Total#of Student Housing I of New I Total#of Student Units in the: Units with `Kids in New:Generation Units in the: Units with Kids in New:Generation SUBDIVISION Subdivision Kids I Units Rate Subdivision! Kids Units Rate Coker Woods 25; 76.0%: 31: 1.240 19 100.0%€ 31: 1.632 .........................................»................ ..............»..........1........................e..........................,......................... ..........................1........................a..........................,..««.................. Chandler's Green 78' 43.6%:: 60: 0.769 22: /0: 26: 1.182 .....................................................«.... .«.......................,«.......................4..............».».......,......................... ..........................,...... ................... ..........................,..........«............ CheSIeY.......................................... ...................•79: 44.3%@ 611 0.772 38: 44.7%€ 32 0.842 t.........................z..............»..........,......................... ..»..... ,.........................,...... +. Springy Crest «_ 88; 54.5°�; »69 0.784 78.;. 56.4%: 64 0.821 .....».... .................«. ......». .............o.._..........«...». ,......................... ........». o..r......-.--..«..«.........». .. Ironwoods 132' 50.8�_ 105' 0.795 29; 55.2/o' _ 23 0.793 Forest Creek...........««».......»« »«»«......104 »»».45.2% .........«..» »80- ...... 0.769 ......»........ 23: ...... 47.8° .»..»» .»..17€ »»0.739 ««.»»....»»» . .« ««.»....... ....«..............:.....»»». .... .... .« .. ........:... ... » North Haven At Ironwoods 37: 24.3%: 12: 0.324 171 41.2%: 9:» 0.529 «............«..».. ....». ..........................�. .:......................... ..........................:...._...................4..............». ...........»............. .. Oxford Hills 461 34.8%• 28 0.609 » 39• 28.2%' 20: 0.513 ...............................».....................«.« .»....»......«». ..../».....«................8..............»».«.....,......................... ....1.........................4...»»»......----.---------................... Windhover 40 35.0%: 19: 0.475 30€ 33.3%: 15 0.500 ...........................«......»....»..«»..«.«». ...........««..».......1...»«»................0......».»....«««.....1.....».................. «.»«..................,.........................6«....w......«......«.I.......w......«..... Colonx Lake » 106: 25.5%: 35: 0.330 52: 28.8%: 201 0.385 Culbreth Park 46' 32.6/o' 22'• 0.478 17• 29.4 k' _ 6: _0.3_53 Cobble Rid.«e............... ...«..«... »»..».... 39,. »10.3%°........»»...«..«.5«•... ..«... 0.128 ..............«..27` ......»11:1%_*....._ 4s...»« »0.148 Coker«H Hills West(older) ... ....... 44,....... 45.5°h .....».«««» 35=....... 0.795 ............».......- North Forest Hills(older 90: 36.7%: 52: 0.578 - - .................. - - ... .. ........................ T .:A ».z;954:;>::>:<:;:.;::>38.9/0.,:; 6.14..:...::.:. 0.632>:..:.:.:::. .:..:...E ;;. ..:.;45.4./a..;.:;;:.;;;,.:;:.;;;;;;;;;;;:..:.;;;;:.;;; OT LS,.AVERAGES . ..::... .:..:.:.:..:... _...... _....__.......... . • Sample of CARRBORO SUBDIVISIONS ALL EXISTING HOUSING UNITS HOUSING UNITS BUILT IN 1992, 1993 OR 1994 Total New ;Percentage; Total New ':Percentage-. Housing of New I Total#of Student Housing of New Total#of I Student Units in the'=, Units with :Kids in New;Generation Units in the: Units with Kids in New:Generation SUBDIVISION Subdivision: Kids I Units Rate Subdivisioni Kids Units Rate Wexford 35- 45.7% »25: 0.714 16: 100.0%: 25 1.563 Cates Farm 35: 40.0%: 24€ 0.686 21: 61.9%� 221 1.048 »....»_.................... .. ,..»»...................,...................»»....,......».».. ........ .». :.........................y....... . ..».....;.«. ..»«.......... Berryhill »... ......» » 39:_ 66.7%: 42: 1.077 -----------23: 6.... .. 22. 0.957 Hililand Meadows 16: ..««.......... 14:........ 57.1%««»...«.....«.. 12:«»»«. 0.857 Quarterpath Trace » » _ 791• «« 49.4%0° «__« 67: «0.848 » «59: 52.5%: «»» »_»50! «0.847 » ... ;... .» ;»« The Highlands «. 94 «64.9% ».....« 98; .....•«1.043 --...«.•.•..••. 9' ....• 55.6°�: «.....«.«..«. 71 »»« 0.778 Cobblestone 55j49.1%4� 42`•:� 0.764 11: »36.4%r« _7,... «0.636 Bolin Forest 63; «» »0.606 35' 42.9/o. _ 20 0.571 Sprin9_Valley �_�»; »86' 37.2°r6: ;w « 45: « 0.523 « 18: 44.4% »;w 9-; _ 0.500 Weatherhill Pointe -- « 561 39.3% « ».. 321 0.571 33 33.3°�� » »�« «16 0.485 :... « » Fair Oaks%Sudbur�i%V1laverly «_....». 1381• » 48.6%o µ »_1021- »� 0.739 ..................26'.. 23.1%F- �. --- 0.269 Windwood folder subdiv -,�».•751....,m.:28.0°r6� w324 0.427 ..:........».....»......". - ...». o :. a 0:774 TOTALS:'AVERAGES: ;,.:,:;:>:::>::<::;::;:::::;.<:<:;812..:;;;>;::;::.:46.5 h::>:<>:>::;::;::;:<;584. . . . .7 ...:... .. ..:..:......,................. • Source: Based on Residential Certificates of Occupancy issued for Chapel Hill and Carrboro in 1992, 1993, and 1994. Compared to children in the Chapel Hill-Carrboro public schools in Spring 1995. • Comparison to Carrboro Subdivision Study The table below compares the results obtained in this study with those of a quick analysis of student generation rates contained in a memo from the Carrboro Director of Planning of April 14, 1994. COMP arison of Student Generation Rates for Carrboro SubdivisionS3 Subdivision Housing Units Students Student Generation Rate Carrboro Current Carrboro Current Carrboro Current Analysis Study Analysis Study Anal sis Stud Bolin Forest 137 104 40 63 0.29 0.61 Cobblestone 57 55 51 42 0.89 0.76 Fair Oaks 140 138 96 102 0.69 0.74 Highlands 97 94 94 98 0.97 1.04 Quarterpath 80 79 60 67 0.75 0.85 Trace • 1 Total 1 511 1 470 1 341 1 372 f 0.67 1 0.79 There are several differences shown in this table. The number of units per subdivision tend to be somewhat fewer in the current study, probably due to the way the subdivisions were delimited. For all but one subdivision, the number of children enrolled increased in spite of this. While this may indicate that the number of public school children per unit increased between 1994 and 1995, the inference that the student generation rate increases with the age of the unit is contradicted by the evidence presented above that existing housing units in the subdivisions have a lower student generation rate than new housing units. In the current study, the student generation rate is higher for these subdivisions than for the subdivisions as a whole, which is in turn higher than the rate for all new housing units. This may be due to the wide range in student generation rates among subdivisions or possibly because housing units in subdivisions may have higher student generation rates than other new housing units. 'Data for Laurel Springs subdivision is not included because the data was not collected for the current study. 17 Analysis of New Housing Owners to address the question of what • In addition to the above analyses, the data was used q proportion of new housing is built by current residents of the school district. This question arose as a way to test the assumption that most new houses are built by newcomers to the area who have not been paying taxes to the County to support the school district. Unfortunately, gathering data on this topic is difficult, and the results are not clear. The only readily available data were the children's school entry dates and the dates of issuance of the certificates of occupancy. Each new housing unit in the Chapel Hill-Carrboro School District with at least one child was examined to determine whether any children from that household entered the public school system prior to the date the Certificate of Occupancy was issued. School entry dates were not available for the Orange County School District. Forty-six percent (46%) of households with children in new housing units had at least one child enrolled in the public school system prior to the date of issuance of the Certificate of Occupancy, while fifty-four percent (54%) had no children enrolled before the Certificate was issued. These numbers might indicate that less than half of new housing units are occupied by • people who lived in the school district before occupying a new housing unit, while more than 50% of new housing units are occupied by people who moved to the area. There are a number of complicating factors, however. The main weakness of this analysis is that the majority of new housing units had no children in the school system and therefore were not included. Additionally, there are a number of problems with the data for households with children. In some cases, families may temporarily live in a nearby apartment to oversee the construction of their homes and to allow their children to begin school at the start of a new school year. Alternatively, some families with children who enrolled after the date of the Certificate of Occupancy may have lived in the area previously with children who were too young to attend school. Due to these problems, the data is not adequate to answer the question. In order to obtain better data, a survey could be conducted of residents of new housing units asking for their previous address, the date they moved to the current address, and the length of stay at the previous address. 18 Conclusion The most important result of this study is evidence that the student generation rate for new housing units in the Chapel Hill-Carrboro School District is higher than that calculated for the Technical Report using data from the 1990 Census. As a result, the Technical Report, and possibly the impact fee itself, should be updated to reflect the higher student generation rate. For the Orange County School District, this study suggests use of a lower student generation rate than that based on the 1990 Census. While there may be a difference in student generation rates by type of housing, better data would be needed to determine this. The other housing characteristics examined here, especially the area and value of the housing unit, have a statistically insignificant or weak relationship with the number of public school children per housing unit. The strongest relationship found was between the number of bedrooms and the number of public school children in the Chapel Hill-Carrboro School District. In this case, the largest difference in average number of children was about .5 children, and was between housing with 4 or more bedrooms and housing with 3 or fewer bedrooms. 19 Appendix A Means and Confidence Intervals Orange County School District Average Number of Bedrooms versus Number of Public School Children 3.20 3.10 T 0 a 3.00 1 E2.90 -- T Z 2.80 0 1 2 3+ Number of Public School Children Average House Area versus Number of Public School Children • 1,600 1,550 1 1,500 T 1,450 Q a 1 1 1,350 1,300 0 1 2 3+ Number of Public School Children Average House Value versus Number of Public School Children $75,000 $70,000 1 d $65,000 i 0 } s $60,000 } $55,000 0 1 2 3+ Number of Public School Children Note: Each graph shows the difference in means and the 95% Confidence Intervals.