1、Designation: D5791 95 (Reapproved 2012)1Standard Guide forUsing Probability Sampling Methods in Studies of Indoor AirQuality in Buildings1This standard is issued under the fixed designation D5791; the number immediately following the designation indicates the year oforiginal adoption or, in the case
2、 of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon () indicates an editorial change since the last revision or reapproval.1NOTEReapproved with editorial changes in April 2012.1. Scope1.1 This guide covers criteria for determin
3、ing when prob-ability sampling methods should be used to select locations forplacement of environmental monitoring equipment in a build-ing or to select a sample of building occupants for question-naire administration for a study of indoor air quality. Some ofthe basic probability sampling methods t
4、hat are applicable forthese types of studies are introduced.1.2 Probability sampling refers to statistical sampling meth-ods that select units for observation with known probabilities(including probabilities equal to one for a census) so thatstatistically defensible inferences are supported from the
5、sample to the entire population of units that had a positiveprobability of being selected into the sample.1.3 This guide describes those situations in which probabil-ity sampling methods are needed for a scientific study of theindoor air quality in a building. For those situations for whichprobabili
6、ty sampling methods are recommended, guidance isprovided on how to implement probability sampling methods,including obstacles that may arise. Examples of their applica-tion are provided for selected situations. Because some indoorair quality investigations may require application of complex,multista
7、ge, survey sampling procedures and because this stan-dard is a guide rather than a practice, the references inAppendix X1 are recommended for guidance on appropriateprobability sampling methods, rather than including exposi-tions of such methods in this guide.1.4 UnitsThe values stated in SI units a
8、re to be regardedas standard. No other units of measurement are included in thisstandard.2. Referenced Documents2.1 ASTM Standards:2D1356 Terminology Relating to Sampling and Analysis ofAtmospheres3. Terminology3.1 DefinitionsFor definitions of terms used in this guide,refer to Terminology D1356.3.2
9、 Definitions of Terms Specific to This Standard:3.2.1 censussurvey of all elements of the target popula-tion.3.2.2 cluster samplea sample in which the samplingframe is partitioned into disjoint subsets called clusters and asample of the clusters is selected.3.2.2.1 DiscussionData may be collected fo
10、r all units ineach sample cluster or, when a multistage sample is beingselected, the units within the sampled clusters may be furthersubsampled.3.2.3 compositing samplesphysically combining the ma-terial collected in two or more environmental samples.3.2.4 expected valuethe average value of a sample
11、 statis-tic over all possible samples that could be selected using aspecified sample selection procedure.3.2.5 multistage samplea sample selected in stages suchthat larger units are selected at the first stage, and smaller unitsare selected at each subsequent stage from within the unitsselected at t
12、he previous stage of sampling.3.2.5.1 DiscussionFor assessing the indoor air quality in apopulation of office buildings, individual buildings might beselected at the first stage of sampling, floors selected withinsample buildings at the second stage, and monitoring locations(for example, rooms or gr
13、id points) selected on sampled floorsat the third stage.3.2.6 population parametera characteristic based on orcalculated from all units in the target population.1This guide is under the jurisdiction of ASTM Committee D22 on Air Qualityand is the direct responsibility of Subcommittee D22.05 on Indoor
14、 Air.Current edition approved April 1, 2012. Published July 2012. Originallyapproved in 1995. Last previous edition approved in 2006 as D5791 - 95(2006).DOI: 10.1520/D5791-95R12E01.2For referenced ASTM standards, visit the ASTM website, www.astm.org, orcontact ASTM Customer Service at serviceastm.or
15、g. For Annual Book of ASTMStandards volume information, refer to the standards Document Summary page onthe ASTM website.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.3.2.6.1 DiscussionThe purpose of selecting a sample isusually to
16、estimate population parameters. Population param-eters cannot actually be calculated unless data are available forall units in the population.3.2.7 probability samplea sample for which every unit onthe sampling frame has a known, positive probability of beingselected into the sample.3.2.7.1 Discussi
17、onThe terms probability sampling andrandom sampling are sometimes used interchangeably.3.2.8 sampling framea list from which a sample is se-lected.3.2.8.1 DiscussionAn ideal sampling frame contains eachmember of the target population exactly once and contains nounits that are not members of the targ
18、et population. In practice,the sampling frame may miss some members of the targetpopulation (for example, new employees in a building) andinclude some individuals who are not members of the targetpopulation (for example, individuals who no longer work in thebuilding). However, no member of the popul
19、ation should belisted more than once on the sampling frame.3.2.9 simple random samplea sample of n elements se-lected from the sampling frame in such a way that all possiblesamples of n elements have the same chance of being selected.3.2.10 statistica sample-based estimate of a populationparameter.3
20、.2.11 stratified samplea sample in which the samplingframe is partitioned into disjoint subsets called strata, andsample units are selected independently from each stratum,possibly at different sampling rates.3.2.12 systematic samplea sample selected by choosingone of the first k elements on the sam
21、pling frame at random andthen including every kth element thereafter.3.2.13 target populationthe set of units or elements (forexample, people or locations in space and time) about which asample is designed to provide inferences.3.2.13.1 DiscussionThe target population is sometimesreferred to as the
22、population or universe of interest.3.2.14 unbiased estimatora statistic whose expectedvalue is equal to the population parameter that it is intended toestimate.4. Summary of Guide4.1 When the objectives of an investigation of indoor airquality include extending inferences from a sample of units toth
23、e larger population from which those units were selected,probability sampling methods must be used to select thesample units to be observed and measured. Examples include:4.1.1 Estimating the distributions of health and comfortsymptoms experienced by the employees in a particular build-ing during a
24、specific week.4.1.2 Estimating the distribution of hourly average concen-trations of specific substances in the breathing zone air in aparticular building during the working hours of a specificweek.4.1.3 Estimating the relationship between measures of en-vironmental conditions in a building and the
25、health or comfortsymptoms experienced by the occupants.4.1.4 Thus, the study objectives are always a key consider-ation for determining if probability sampling methods arenecessary. Potential objectives for indoor air studies that wouldrequire probability sampling methods are discussed explicitlyin
26、Section 6.4.2 Guidance is provided regarding the appropriate prob-ability sampling methods to address these and other goals thatrequire extending inferences from a sample to a specificpopulation. Those sampling methods require construction of asampling frame from which population elements can besele
27、cted. Examples include:4.2.1 A list of all offices or work stations in a building,4.2.2 A grid of potential monitoring locations that effec-tively covers the entire population of interest, and4.2.3 A list of all persons who work in a specific building.4.3 Since environmental concentrations usually v
28、ary con-tinuously in time, spatial frame units like those listed in 4.2often must be crossed with temporal units, such as seasons,weeks, days, or hours, to form sampling frame units (forexample, building-seasons, office-weeks, or person-days). Spe-cific issues that must be considered when constructi
29、ng thesetypes of sampling frames are discussed in Section 7.4.4 In addition to constructing sampling frames, a random-ization procedure is necessary so that units can be selectedfrom the frame with known probabilities. Some basic consid-erations for and methods of selecting probability samples forst
30、udies of indoor air quality are presented in Section 8.4.5 Finally, Section 9 discusses considerations for statisticalanalysis and reporting that are peculiar to data collected usingprobability sampling designs. Special statistical analysis meth-ods are necessary when the sampling design includes st
31、ratifi-cation, clustering, multistage sampling, or unequal probabili-ties of selection.5. Significance and Use5.1 Studies of indoor air problems are often iterative innature. A thorough engineering evaluation of a building (1-4)3is sometimes sufficient to identify likely causes of indoor airproblems
32、. When these investigations and subsequent remedialmeasures are not sufficient to solve a problem, more intensiveinvestigations may be necessary.5.2 This guide provides the basis for determining whenprobability sampling methods are needed to achieve statisti-cally defensible inferences regarding the
33、 goals of a study ofindoor air quality. The need for probability sampling methodsin a study of indoor air quality depends on the specificobjectives of the study. Such methods may be needed to selecta sample of people to be asked questions, examined medically,or monitored for personal exposures. They
34、 may also be neededto select a sample of locations in space and time to bemonitored for environmental contaminants.5.3 This guide identifies several potential obstacles toproper implementation of probability sampling methods instudies of indoor air quality in buildings and presents proce-dures that
35、overcome those obstacles or at least minimize theirimpact.5.4 Although this guide specifically addresses samplingpeople or locations across time within a building, it also3The boldface numbers in parentheses refer to the list of references at the end ofthis guide.D5791 95 (2012)12provides important
36、guidance for studying populations of build-ings. The guidance in this document is fully applicable tosampling locations to determine environmental quality orsampling people to determine environmental effects withineach building in the sample selected from a larger populationof buildings.6. Study Obj
37、ectives That Require Probability SamplingMethods6.1 Inferences beyond the units actually observed in asample are not rigorously defensible unless the units observedare a probability sample selected from the population to whichinferences will be extended. Thus, probability sampling meth-ods are neede
38、d whenever inferences will be extended from theunits observed in a sample to a larger population. The need forsuch inferences depends directly on the objectives of the study.The study objectives may include characterizing a buildingsoccupants using a survey, or characterizing a buildings airquality
39、using environmental monitoring, or a combination ofboth.6.2 Occupant Survey:6.2.1 A sample of building occupants may be asked tocomplete a questionnaire or to submit to a physical examina-tion. If the intention is to make inferences from the sampleregarding the health and comfort symptoms of all the
40、 employ-ees of the building, a census of all building occupants or aprobability sample selected from them is required. The occu-pants would typically be asked about their health and comfortsymptoms for a specific period of time (for example, the daythat the survey is administered, the previous week,
41、 month, oryear, and so forth). Developing a valid and reliable question-naire is a complex process and is not directly addressed by thisguide (5).6.2.2 Specific study objectives that require inferences to apopulation of building occupants include the following:6.2.2.1 Estimate the distribution of he
42、alth and comfortsymptoms in a building either before beginning air qualitymeasurements, after implementing remedial measures, or as ameasure of the magnitude of a potential indoor air problem.6.2.2.2 Estimate the distribution of health and comfortsymptoms in a building with reported problems and in
43、anotherbuilding studied for comparison purposes.6.2.2.3 Estimate the relationship of health and comfortsymptoms with worker characteristics, such as age, sex, worklocation, or type of work performed.6.2.3 When inferences regarding the occupants of a buildingare needed, a census of all the building o
44、ccupants may benecessary. For example, a census of building occupants may beneeded to establish statistical differences in occupant comfortor health symptoms between different work areas (for example,floors) within a building. In other cases (for example, estimat-ing the relative frequency of compla
45、ints in a building with alarge number of workers), a probability sample may providesufficient precision at less cost.6.2.4 If the characteristics measured in a questionnaire aretemporally dependent (for example, comfort and health symp-toms on the day of questionnaire administration), a sample ofpeo
46、ple and time periods may be needed (for example, a sampleof person-days within a given week). Moreover, the surveymay need to be replicated across time (that is, repeated indifferent seasons).6.2.5 A successful occupant survey requires that a largeportion of the sample subjects complete the survey.
47、Forexample, the United States Office of Management and Budgetusually requires 75 % or more for federally funded surveys.Thus, the success of a survey may depend upon the burden itimposes, pre-survey publicity (for example, newsletters orunion endorsements), and follow-up of nonrespondents. Thesurvey
48、 should be conducted in such a manner that people aresufficiently motivated to participate but not unduly alarmedabout a potential air quality problem. Finally, residual nonre-sponse is inevitable, and survey data analysis procedures thatutilize weighting or imputation to compensate for nonresponsea
49、re recommended.6.3 Environmental Monitoring:6.3.1 Since air quality characteristics generally exhibit bothspatial and temporal variability, each air quality measurement(for example, temperature, humidity, or concentrations ofspecific substances) is generally representative of a specificlocation and time (or period of time). If the objective is to inferinformation about the distribution of the measured character-istics (for example, the mean or the range) for a targetpopulation of times and places, then probability sampling ofboth locations and times is requ
