The Feasibility of a Street-InterceptSurvey Method in an African-AmericanConmunit

Kevin W Miller, MPH, RN, Lora B. Wilder, ScD, RD, Frances A. Stillman, EdD,and Diane M. Becker, ScD, MPH.

IntroductionTraditional population survey meth-

ods may produce inadequate sociodemo-graphic profiles in certain settings.1-5 Inlow income and minority populations,random-digit-dial telephone surveys havedemonstrated respondent bias due todisproportionate representation of personsof higher income, education, and employ-ment levels.6-10 Refusal rates are oftenhigher with telephone methods,"",12 par-ticularly in younger African-Americanmales.7 Mail surveys are characterized bylow response rates and similar biases.'1'6Face-to-face interviews have higher re-sponse rates and greater participant respon-siveness to interviewer questions.'2"14However, in urban areas with high rates ofcrime, face-to-face surveys may beavoided or may result in biases overrepre-senting higher socioeconomic communitymembers. 4 Thus, populations at highestrisk may not be adequately represented bytraditional survey methods.

We evaluated the feasibility of anonquota, street-intercept method, utiliz-ing random selection of interview sites toaccess a representative sample of an urbanAfrican-American population. We com-pared the sample's sociodemographiccharacteristics with US Census data andwith a random-digit-dial telephone surveyfrom the same community.

MethodsInterview Sites and Process

Interviewing sites were based on thesampling unit of a block, defined as thelength of a street from one intersection tothe next intersection. All 3384 blocks inthe catchment area were enumerated fromcensus tract maps and were eligible forselection according to a computer-generated random-number list. Samplingnoncontiguous blocks required frequenttransportation of interviewers from blockto block. To improve efficiency, weconducted interviews in an expandedinterview site, defined as all blockscontiguous to the randomly selected

"index" block. This cluster samplingmodification typically added 6 blocks tothe interview site.

A team of six to eight indigenous,African-American, trained interviewers,and a field supervisor, were assigned toeach interview site. Interviewers workedin pairs for safety, although each interviewwas conducted by a single interviewer;this allowed two interviews to be con-ducted simultaneously along a givenblock. Interviewer pairs moved fromblock to block until all eligible personshad been interviewed, typically taking 45to 75 minutes per interview site.

The street-intercept method was de-signed to access all people on the streetengaged in such activities as sitting onsteps, walking to or from work, runningerrands, performing job-related duties,preparing to use public transportation,visiting, participating in recreation, or just"hanging out." Eligible respondents wereAfrican-American adults, aged 18 orolder. Interviewers were instructed toapproach the first eligible respondent theysaw who was anywhere in the block as theinterview period began.

For refusals, the interviewer re-corded an estimate of the person's age,sex, and the stated reason for refusal.Interviewers were educated about how toavoid unsafe situations (persons who wereintoxicated, exhibiting threatening behav-ior, or presumed to be involved in a drugdeal). All interviews were conductedoutdoors on weekdays, usually between9:00 a.m. and 3:00 p.m., during thesummer months of 1992 and 1993.

Street-Intercept Instrumentand Telephone Survey

The 1992 street-intercept question-naire elicited information about sociode-

At the time of the study, all authors were with TheCenter for Health Promotion, The Johns HopkinsSchool of Medicine, Baltimore, Md.

Requests for reprints should be sent toKevin W. Miller, MPH, RN, Johns HopkinsCenter for Health Promotion, 1830 E. MonumentSt, #8023, Baltimore, MD 21205.

This paper was accepted May 30, 1996.

American Journal of Public Health 655

Public Health Briefs

mographic characteristics and health sta-tus, including items on body image andperceptions of obesity (assessed withpictorial analogues17), smoking, and foodconsumption and a literacy assessment."8It consisted of 64 items and required 10 to15 minutes to complete. In 1993, thequestionnaire was increased to 91 itemsand took 15 to 20 minutes to complete.

A modified Waksberg method'9 was

used to generate a list of random tele-phone numbers within the same census

tracts used in the street survey. Eligiblehouseholds were noncommercial African-American dwellings located in the catch-ment area with at least one adult aged 18or older. To control for respondent bias,the interviewer asked to speak to the adultin the household with the most recentbirthday. Interviewers tried all numbersup to six times, rotating the day of theweek (Monday through Friday) and thetime of day (moming, aftemoon, or

evening). The survey assessed sociodemo-graphic characteristics using the same

questions as the street survey and includeditems on smoking, diet, and health.Trained interviewers were supervised andwere selected from the same pool ofpersons as the street-survey interviewers.

Evaluation Criteria

To evaluate the feasibility of thestreet-intercept survey method, we as-

sessed response rate, catchment-area resi-dence rate, interview completion rate,interview interference rate, and, as an

estimate of cost, interview yield rate (alldefined in the footnotes of Table 1).

The survey examined representative-ness by comparing sociodemographiccharacteristics of the sample to unadjustedUS Census data for the same census

tracts. Census tract data were obtainedfrom the 1990 Census Summary Tape FileJA20 and were converted to Epi Info2' datafiles. Relevant denominators and rateswere calculated. Comparison was alsomade with the population characteristicsof the random-digit-dial telephone survey.

ResultsFeasibility Criteria

A total of 994 interviews were

conducted; 578 in 1992, and 416 in 1993.Interviews were conducted in all 30census tracts, in 77 interview sites and in395 different blocks. Results are presentedin Table 1. Incomplete interviews occur-

red owing to the arrival of public transpor-tation for which the respondent was

waiting or the need to respond to a workdemand. Eighteen interviews were avoidedand three were interrupted because ofperceived risks or threats (e.g., drugdealing, drunkenness, a man brandishinga knife, and rowdy behavior). Unsafeconditions in three interview sites causedthe team to abandon the interview sitebefore all potential respondents could beapproached. In a fourth interview site, nointerviews were conducted because poten-tial respondents fled as the team membersdisembarked from their van. Sex and agedistributions were not significantly differ-ent between respondent and nonrespon-dent groups. The response rate for therandom-digit-dial survey was 61.3% (928/1514), and the interview completion ratewas 85.6% (794/928).

Sociodemographic Characteristics

Sociodemographic characteristics ofcatchment-area residents were comparedwith the random-digit-dial telephone sur-

vey and the 1990 US Census data for thesame census tracts (Table 2). More men

and younger respondents were accessed inthe street survey compared with the

656 American Journal of Public Health

TABLE 1-Feasibility CriteriaResults forStreet-interceptSurvey Method

No. Rate

Response ratea,f 416/519 80.2%

Residence rateb,f 355/416 85.3%

Interview comple- 973/994 97.9%tion ratec

Interview interfer- 21/519 4.0%ence rated,f

Interview yield ratee 942 2.5(interviews perinterviewer per hour)

aRespondents consenting to be inter-viewed out of eligible respondentsapproached.

bRespondents with residence in catch-ment area out of respondents consent-ing to be interviewed.

Clnterviews completed out of numberconsenting to be interviewed.

dinterviews interrupted or not started(owing to safety concerns) out ofeligible respondents identified.

eCompleted interviews per interviewerper hour.

f1993 survey only.

TABLE 2-Demographic Characteristics of Street-intercept SampleCompared with Random-Digit-Dial Telephone Survey and 1990Census

% of Random-Digit-Dial% Street-intercept Telephone

% of 1990 Sample (n = 942)b Sample (n = 928)US Censusa (95% CI)C (95% Cl)

Male 43.8 50.7 (47.2, 54.2) 25.5 (22.7, 28.3)

Female 56.2 49.3 (45.8, 52.8) 74.5 (71.7, 77.3)

Age18-44 63.4 68.3 57.445-64 23.8 21.3 27.8>65 12.8 10.4 14.8

High school graduate,d 45.9 59.2 (55.9, 62.6) 59.1 (55.6, 62.6)

Employed 47.5 36.2 (29.3, 43.1) 58.9 (55.7, 62.2)

Not employede 52.5 63.8 (56.9, 70.7) 41.1 (37.9, 44.4)Incomef<$10 000 35.1 43.7 ...

$10 000-$24 999 30.2 31.3 ...

$25 000-$39 999 17.09 18.9>$40 000 17.79 6.2 ...

Note. Cl = confidence interval.a60 337 Black residents aged .18.bEmployment status and income assessed in 1993 survey only (n = 351).C95% confidence intervals for street-survey variables reflect cluster-design effect.dPersons aged .25 completing at least grade 12 or equivalent.eStudents, homemakers, retirees, and unemployed persons."'Household" income for US Census; "family" income for street method.gExtrapolated from $35 000 through $49 999 category.

April 1997, Vol. 87, No. 4

Public Health Briefs

Census and random-digit-dial surveys.Both street-intercept and random-digit-dial surveys identified more persons witha high school education than did theCensus. The street-intercept method under-represented employed persons while therandom-digit-dial survey overrepresentedthis group. Income distributions wereshifted toward lower incomes in thestreet-intercept sample.

DiscussionThe street-intercept method was em-

ployed to capture a representative sampleof the eligible population within a geo-graphically defined catchment area. Street-intercept methods have been used withspecific population sectors for focal stud-ies of high-risk categorical problems only,such as transmission of the human immu-nodeficiency virus,22 adolescent drug use,23and illegal drug sales.24

A population-survey method must beboth feasible and representative if it is tobe useful. The response rate of 80% in thestreet-intercept method is comparable toothers22 and superior to the 61% responserate of the telephone survey. The inter-view completion rate is higher in the streetmethod, perhaps because respondents areless likely to break off a face-to-faceinterview with an indigenous inter-viewer. 145 The 85.3% residence rateindicates the method is fairly efficient inidentifying residents of the catchmentarea.

Consistent with the literature, thetelephone survey underrepresented men;25.5% compared with an expected 43.8%in the Census. This was not a problem forthe street method, where men wereoverrepresented. Both the street and thetelephone surveys found higher educa-tional attainment compared with theCensus. While telephone surveys havebeen reported to overrepresent those withhigher education, it is not clear why thestreet method would do so, although thishas been found before.22

Employment rates vary among thethree survey methods. Disproportionatelyhigh rates of employment among AfricanAmericans interviewed by telephone com-pared with in person have been foundbefore.9 In contrast, employed personswere underrepresented in the street methodcompared with the Census. Discrepanciesin some demographic variables may alsoresult from Census data limitations whichresult from undercount problems, espe-cially in inner city areas.25

Small differences were also ob-served in income distributions measuredin the street survey compared with thosefound in the Census data. Definitiondiscrepancies may account for some ofthe difference, although finding morerespondents with lower incomes is consis-tent with the higher numbers of unem-ployed persons identified in the street-method sample. We were able to assessincome level in 93% of respondents in thestreet survey whereas in the telephonesurvey, congruent with other studies,'0'1'the nonresponse to this item was soconsistently high that it was dropped fromthe questionnaire.

Several limitations of the street-intercept method are worth noting. Clustersampling, while improving survey effi-ciency, typically introduces greater sam-pling error.Y6(P204) We observed about a1.8-fold increase in variance for theemployment variable and minimal in-creases in other variables compared withwhat would be observed if the data weretreated as a simple random sample. Asecond limitation is the potential for biasintroduced by interviewers' passing byeligible, but "undesirable," respondents.This problem can be monitored withsupervision and documentation of allinterviews avoided for whatever reason.Such documentation would allow calcula-tion of a nonresponse rate distinct fromthe refusal rate. Additionally, the methodis dependent on fair weather, underrepre-sents the homebound, and may overrepre-sent homeless persons and the unem-ployed. Traditional survey methods oftenunderrepresent or entirely miss homelesspersons because they rely on a residenceor a telephone. Indeed, being out on thestreet may be the only way to reach somepopulation groups.24'27

Safety is a concern in most urbansettings. The interview interference rate of4.0% is low and may have resulted frominterviewers and respondents feeling saferin public than if the interviews had beenconducted in private residences.4 Al-though there were no harmful incidents,the risk increases with larger surveys, andinterviews were avoided during late after-noon and evening hours when interview-ers perceived greater risks.

We conclude that the street-surveymethod is feasible, as evidenced by the80% response rate, high interview comple-tion rate, low interview interference rate,and a reasonable interview yield rate. Thestreet method is likely to be most cost-effective in urban areas where members ofthe target population are geographically

clustered as was the case for this survey.28When eligibility characteristics are visu-ally discernible, the method allows inter-viewers to identify potential respondentsquickly, in contrast to mailed and tele-phone surveys.29 Visual characteristics(race, gender, age) of nonrespondents canalso be assessed, thus allowing for anestimate of response bias.

The method reached a broad sampleof the target population on most sociode-mographic indicators. Although somepopulation groups such as the homelessand the unemployed may be overrepre-sented, the ability of the street-interceptmethod to survey those who are hard toreach can be considered an advantage.There is an increasing need to assesshealth status in persons who bear adisproportionate burden of disease, andthe street-intercept method providesgreater access to these groups. Thus, morecomplete assessments of the health issuesin traditionally high-risk, urban, minoritycommunities are made possible. Z

AcknowledgmentsThis research was supported by a grant from theNational Heart, Lung, and Blood Institute,#HL46757-03.

The authors wish to thank the interview-ers and supervisors for their help in conductingthe survey, and Cheryl Finney, RN, MPH, forher recommendations and review of the manu-script.

References1. McGraw SA, McKinlay JB, Crawford SA,

Costa LA, Cohen DL. Health surveymethods with minority populations: somelessons from recent experience. In: BeckerDM, Hill DR, Jackson JS, Levine DM,Stillman FA, Weiss SM, eds. HealthBehavior Research in Minority Popula-tion: Access, Design, and Implementation.Washington, DC: National Institutes ofHealth; 1992:149-167. NIH publication92-2965.

2. Shumaker SA, Jackson-Thompson J, TaskGroup VII. Survey measurement withminority populations. In: Becker DM, HillDR, Jackson JS, Levine DM, Stillman FA,Weiss SM, eds. Health Behavior Researchin Minority Populations: Access, Design,and Implementation. Washington, DC:National Institutes of Health; 1992:194-206. NIH publication 92-2965.

3. Rice DP, Drury TF, Mugge RH. Householdhealth interviews and minority health. MedCare. 1980;18:327-335.

4. Word CO: Cross-cultural methods forsurvey research in black urban areas. In:Burlew AKH, Banks WC, McAdoo HP,Azibo DA, eds. African American Psychol-ogy. Newbury Park, Calif: Sage Publica-tions; 1992:28-42.

5. Mays VM, Jackson JS. AIDS surveymethodology with Black Americans. SocSciMed. 1991 ;33:47-54.

April 1997, Vol. 87, No. 4 American Journal of Public Health 657

Public Health Briefs

6. Freeman H, Kiecolt K, Nicholls W, ShanksJM. Telephone sampling bias in surveyingdisability. Public Opinion Q. 1982;46:392-407.

7. Weaver CN, Holmes SL, Glenn ND. Somecharacteristics of inaccessible respondentsin a telephone survey. J Appl Psychol.1975;60:260-262.

8. Marcus AC, Crane LA. Telephone surveysin public health research. Med Care.1986;24:97-112.

9. Aquilino WS, LoSciuto LA. Effects ofinterview mode on self-reported drug use.Public Opinion Q. 1990;54:362-395.

10. Groves RM, Kahn RL. Surveys by Tele-phone: A National Comparison with Per-sonal Interviews. New York, NY: Aca-demic Press; 1979:79-120.

11. Aquilino WS. Telephone versus face-to-face interviewing for household drug usesurveys. Int JAddict. 1992;27:71-91.

12. Groves RM, Lyberg LW. An overview ofnonresponse issues in telephone surveys.In: Groves RM, Biemer PP, Lyberg LE,Massey JT, Nicholls WL II, Waksberg J,eds. Telephone Survey Methodology. NewYork, NY: John Wiley & Sons; 1988:191-211.

13. Dillman DA. Mail and Telephone Surveys:The Total Design Method. New York, NY:John Wiley & Sons; 1978:39-78.

14. Brambilla DJ, McKinlay SM. A compari-son of responses to mailed questionnairesand telephone interviews in a mixed mode

health survey. Am J Epidemiol. 1987;126:962-971.

15. Orleans C(, Schoenbach VJ, Salmon MA,et al. A survey of smoking and quittingpatterns among Black Americans. Am JPublic Health. 1989;79:176-181.

16. O'Toole BI, Battistutta D, Long A, CrouchK. A comparison of costs and data qualityof three health survey methods: mail,telephone and personal home interview.Am JEpidemioL 1986;124:317-328.

17. Stunkard AJ, Sorenson T, Schulsinger F.Use of the Danish adoption register for thestudy of obesity and thinness. In: Kety SS,Rowland LP, Sidman RL, Matthysse SW,eds. Genetics ofNeurological and Psychi-atric Disorders. New York, NY: RavenPress; 1983:115-120.

18. Wide Range Achievement Test (WRAT).Wllmington, Del: Jastak Associates Inc;1984.

19. Waksberg J. Sampling methods for randomdigit dialing. JAm StatAssoc. 1978;73:40-46.

20. Census ofPopulation and Housing, 1990:Summary Tape File JA on CD-ROM.Washington DC: Bureau of the Census;1991.

21. Epi Info, version 5.01b. Atlanta, GA:Centers for Disease Control and Preven-tion; 1991.

22. Nebot M, Celentano DD, Burwell L, et al.AIDS and behavioural risk factors inwomen in inner city Baltimore: a compari-

son of telephone and face to face surveys. JEpidemiol Community HealthL 1994;48:412-418.

23. Spooner C, Flaherty B. Comparison ofthree data collection methodologies for thestudy of young illicit drug users. Aust JPublic Health 1993;17:195-202.

24. Hopkins W, Frank B. Street studies thatwork and what they show in New YorkCity. In: Frank B, Simeone R, eds. WhatWorks in Drug Abuse Epidemiology. Bing-hampton, NY: Haworth PressInc; 1991:89-97.

25. National Research Council Panel on Cen-sus Requirements in the Year 2000 andBeyond. Population coverage and its impli-cations. In: Edmonston B, Schultze CL,eds. Modernizing the US Census. Washing-ton, DC: National Academy Press; 1995:30-43.

26. Cochran W. Sampling Techniques. NewYork, NY: John Wiley & Sons; 1961.

27. Watters JK, Biernacki P. Targeted sam-pling: options for the study of hiddenpopulations. Soc ProbL 1989;36:416-430.

28. Sudman S, Sirken MG, Cowan CD.Sampling rare and elusive populations.Science. 1988;240:991-996.

29. Sebold J. Survey period length, unan-swered numbers, and nonresponse in tele-phone surveys. In: Groves RM, Biemer PP,Lyberg LE, Massey JT, Nicholls WL II,Waksberg J, eds. Telephone Survey Method-ology. New York, NY: John Wlley & Sons;1988:247-256.

.......: .: :' "': --:--: .... ::-----:---:-:-----:-:-::-::-:--:-:-:-::::-:-:::--:::..:...::--::.:::..:...:..:.::::%.::..::...:..:..:..::,.:...:,..:..::% :- I.:...!-.-....:..:....-- "...-:- ...:.. :"' .:...:: ....::::--:-:-- ...:.:.....,:..:...::......::%...:.%-.... .::.:.::..::::.::.....,.......... .:..::-:--::::-:::::'':..:.:::.::.:.:.::..:..::..:....:.::....%.:....:...:.....::..-:....:....% %-..%...:...:.................::::.:.....::: %.-- .:...::...:..:................::..:.":.:.:.'':::.:...:.:......:......::..:.::.:....-...:..-..%.--....-:.%-----.........:......'''--:::.:''.'.' ..:.:.:..:.:" :...'':.:.......:.......:..:...::::-: ... :.:.:.:" ::.::.:::::..:.:''''."."........- '......-':::':'.':.:':..: :......

....... ..%:..::..::-:-:---::.":..:":::- -:-::-::: -:-:-:-:-:: ....". "...-':':-'..:.:''"-' .............."..:":::--:---:::-:,:::.:.:.::-::::....::.....: :.. .-:.-..-.:-.....:.:':":'..:'.'::-::::- .... :..:.::::.::.:.: M.- .::::.... ..........................---":': .........,_...,- .:..%: .:......._._........ ...:.:.:. ,--%..... .: .. .:..... -V--':--:' .:.::: ----:: "': .:.:":.::.:.:.:..: -..........:

... ........ ::-': ....... .:..'.'..'..'.!.:......:.......%--..-.--..-..--.-.-..-.- --....-'':-:. .:::..:.:.-....-:-..-:-..-.:-..-...:...:.....................................

... :.'..........I.:........: :---..-:..x%

..::... .::::. -... ...-:_......:.....::.:_.:-.-.--:.-... ._............. ".......:.'". ...... .:...' -:-::.:..- .::....:............:............. ': -::. .:..:....-......_. -:::..: :...:.'''

_.: :.:....-................................:.......:..................:......7.,%....:.:..:..::.:.

...............-':.' ..-:........

.::::.:. ............%.........."....".:....%................... ....%.................:.::...::.:................:....................... -:::..:...........'.".%--.-.-.-.-.--.----.---.-.:.:. ....:

.......:....:...:-.:-:.:..:%.........:.:...'.."..-.-..'..: ...:.:'::.:'.:..:...:.-:_..:....:._...:.:...:.: ::.......................

..:::.::::.::--.-..::::.-: ....: .......:..: ..................... ... .........v..... .... ..... .: :...:.:

.:.::.:. .................:.....:......:..-....--...............-...... .:......... :'OAIL -AIWV ... '."..:....::I:,W.. ::;;::;;. ..,::: .... .; ..... .....:...,..-..:: .'IM r....... .-, -........ ,.::..:....... ." "'..;.;:.:I ......"..'! ..........N.-:..". .....,.:::.;:.;...:.::...:.:....:V::..:.iu,: .:..!:.-:.Vw..:::...:..:.;...----_--_.-.:::: :..::::..::;.::.:.: .,'':,*.:'. :!:.:.--! ...-......W.....,..::.,-......-......:...;....., r:IFli r F` -................... .........!i; .,....m;-:'::.:.:.%AU-,--.,.w. .:-.!:.....:.. .:.....::::%..:::-......-..-. ...:. ..:.:.: .............

N'1!-........ii;i..,......'..:V: .iAMW. .:.:..:..:.. i............-'-.ogn j.......I.. .'..:!fir....'., Wm.:: -:VddW9i6A..:.,.,::: ",..:. .--....1:.:,:..:e V.-. .:...n ... -.1 .., :: ......1...l,...% ......on .., ne'l-.1:1-:1 :::,%:M-:-::: ...............:. .......V:'::.'".:..._:-::.:...:::.:.-.... .:.::..::::.::!:::.::..:::.-::".:"...:":.:":.":.":.".:.:.: :'..:::!::.%::..:.:.:::.:.::.:.::....::

...........,.....E....:.M....-.:.:---..:%:. _._x_%.. ::.:::.::.::.::.::..:..::.::::.::.::%..:::...::.::."".......................::.::::..:.:::.:.....:%::......:.... --. ..:..:.:.::..:..:..:.-.-:.-.-.--..:'-:.:-.-:..-::-.- :..-"":.'.':.': :.: .:.."-::-::.':.:.:...........%................:.....::::::::.:...:....-

'..:.....,

::::.....................N:. :.......,

.....::... :. .:::.:..%:...:..:::.:: ::,.: ::.::.:::.::::: ::...::::..:.:::.,..-.:..:::-.. .:--::..:.....: :.--.-.%.::%.:.%-..-.%-.-.%-:. ...- -..--.:-.- :: :. ': ..:..:. .: ..... :- ::::..::....:::.::::::.,::.:::.:::..:.:..::.:.....,....,................................::--::-::...: :......--:--.............%.:.--......:...........................................

.:.:::.:....:%:"....:.......:%...............::..::.:.::...................: :::..:::-:..:.:.:::..::.:..:.--------...:'.':....:.:......... .:::::.

::---:----....................--.%-..-....% ...........:....................."-!...:.:..,.W .::..:. .... ... '': :'' '"'.::. '':":.:''

%.-.::;;..:.M,---:.:-.--' 'W.-'....%'D-:-:---::::::..:: '.;:.... !..:. .....:-.... .:.:.::...:.''...''...."..------... %..:.ig...-'..-'.-.-'..-..-'.....'..-..;;.o...;.:.:::.;!..;; .....iffi..;..;-.::..:.;;..;;.:.;;:i.....;..; :-':::::.:..:--....----'--'-- ....:.... ..... "...". ..::.:....:...:......-........................................ -..-..-..-....-.-.-..................::... ...:..

.:......:...:'':: -...

-I..... .--..::..::............ :.....:..:........................:..:..: ...... :- "'%. .: .: -- .:.... ....... ,............ ......,....................:.::.....:........-.. .-.......................--:----..:......:.::.-:..::.:...%................................:-.:::.:. .................%..........--:..:.......:: :... :..--.;.ff:%:_.:...:.....:. ... ::: ........ :"...::.:......::...'..:.%:..:: .:.:..'.::..:.............................................".:.:":..::-:--:..-::--: --.-:.:-::---.--::.:-

-........,-::.:::.:.....::..:................%...................- ......-N.:N.

....................%- ..............................:- ':-:..::.:.:..:...:..%.:..:....:. .., .::: ..:. .:......................:.::.::.:.

...%i. ...."'.!.M.M.., .........................:..4. --',i. 'M ]"... .. .."-'-'--'-'.I.-i-i.:' -W .....-... -.. :.V-I;.an

.1!:.......... ...........-.............................:...II.:-,:::.:.:.:.:- ":.......%............::::...:................::----.::...:.:: ...... .....:. .... .. ..

T.7.:.....""'...:.-- ',i; .............. ."-..:::....:::.:....:...............................................I...W..]%%..: .................: .:.. .:.:...:.BHeMff.:::.. %..: .:......................................:''.::....:..:.:-:.:.......................".: ::... -.. :-::....: ...:. :4....,..

.:': ..:.. :-:: N N .. .. N..: ': :--:- ::-:- N % .: ... .:..:: ...... :. ...........: :' ...:. -, % .: ::--::-:- ':: '::.:..":% '.-"::"::":.%;:::.::-- V. ....:.:..:: :" :":. N .: ': ..: ".:...... .: .: ... %....

:.......:.":'

-:---:- .:.:.%.':..:. :' ': .:.::.:,::.:..::.:::::.:. ..:: ........ .....:.:.:....:.-.:-............................................:.%--.%--.%--.%.....:.%...%......::--:-:...:: ..::..-....-............................"...:.:..:..:.::::..:: ........... ............................:....:4:.:.'.....--.:..:..-: %'...':.'.::_. ..:..:.:..-.. :...-......................................:....%.............'""':.:..:::'':.:.::.:..-..-...-.....-....-.--.--.-.-.-...................

.........%..-...

:. 'I.:: :::::-::.::-'::...:.,::-.:.: ::....::::::.::..:.....:..:.......: ............:............:.:......................................:...::.::.:.....:.......:..%:...::....,...:..:.::::.::--:."..:::.::.:..:.-....:.:%::%.:...:..:::::.:...::........................:............:. ..... ::...............................................::-----:-------:::.::.::...........:.....:..................

.:..:.:4:...".::.:::..:":.:::..::::.:..::..: .......:: ...... "..:.:'::.....: ...... ..::.:..:.::.: .......-----:--".:.:"":.:..:-:.::::::........:-.....................-

:....................................... %.:..--.%-.-..--.%.....:.%...%......_,-:_ .............. .:.::..:.::-:-: ..... .....N...:....-:--..'....................:.:....:.: ..-.....-:

.......................................: :.:.....: .:...........,....,...........,....... ":. -'..:..:.:..:...............-....-:-......... ...............:.. '%-......... ........................:.:.: .....::. '-:".:,'.:"-'.::".................................

-.--:-:::-.-::._.:.:..:...:...,..................... ..:......................................: ...................:....................-.- :.....--........................%......:.:.:... .:...._._ -x-;::---- ...::.. V........, :....:.:.:.::t-...._:..........:- .. .. ...:. .......:

....."'"

....:.: ....................... ...........................:.::---::...............................................

5: ...%.......:... .....................-:..%-------:::..::-.i ;:.....-...' ':." ".1-01,11 :.......:....::::..:...K..:: -..:.:R..:::%,

....::.-:-::: .:.::::."-...--S--'.i.:...:?..-:.! .:-..;. -.:::: .............

%:..;..A. :.NI.:::%.::... :-: :1 .. .. :.:::....:N.:::.::.:...-:::

.........I:':::.:.....-::.:-:-.:-::::::::::--.:.: ...... .:-..........._..... .WMM..................-%-.... ...:::. .:

:................

"HeW&-- -.: ::::::- W- ..... I. .. A::-:.::.:...:-

...:........:.::.:,-":.-,%..........I:..:...:..:..:...-..-..-.:.PP.m:%.mi I..._..'.d0:.;;: ;i.MI..... :. -.....................

'-... :......:%.........:....-.-..--..--..-:::.:'.'..:..-........._

...............:.....:............::.::-::::::::::::::..:.:..:..: ..........m.mommom; ::.:::......:.......::-::-:.:.:..:..!.:%...:.........................----:--:...... :........:.::::::.::..::: .:. .. .%--.%:: ':-:.:.::..--

:....:..... .: ...._ :....:::..: ... -..-.....-%.:M. ...........,....N.:................... .:

.."... ..:.. .....'.- ....... ... -:....'..":'::':.::....".. :..:.v

._:...... ........".:

:: .:::::.",................,......::i:::::I.........' '"

......... ........ ..............:.:..... .4,