Professional Documents
Culture Documents
Association For Library and Information Science Education (ALISE)
Association For Library and Information Science Education (ALISE)
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=alise.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Association for Library and Information Science Education (ALISE) is collaborating with JSTOR to digitize,
preserve and extend access to Journal of Education for Librarianship.
http://www.jstor.org
Teaching Descriptiveand
Inferential Statistics in
LibrarySchools
RONALD E. WYLLYS
3
JOURNALOF EDUCATION FOR LIBRARIANSHIP
meaninglessdifferencesas important.
Inferentialstatisticsfurnishestools by which to decide whether an
observeddifferenceis "significant," in a stricttechnicalsense: viz.,thatthe
differenceis very unlikelyto be due to chance. A differencethat is
statisticallysignificantmayor maynotbe of practicalimportance.On the
otherhand, a differencemayseem,to theuninitiated,large enough to be
importantand yetbe, in fact,non-significant because itcould easilyhave
been due to mere chance.
A fundamental practice in inferential statisticsis to formulate a
hypothesis- oftencalled a "null" hypothesis- thatamountsto saying,
"The observed phenomenon can reasonablybe attributedto chance, i.e.,
is not significant,"and then to examine the data in order to decide
whetherthishypothesiscan be rejected.Oftentheinvestigatorhopes tobe
able to rejectthe null hypothesisbecause he or she favorsan explanation
other than chance for the phenomenon.
The firstpurpose of thispaper is to indicatethatlibraryand informa-
tionscience already make considerableuse of statisticaltechniques,both
in researchand in practice.Of special interestis thecurrentlevelof use of
the techniquesof inferentialstatistics.From an on-goingstudyof trends
in theuse of statistics
in libraryand informationscience,data are available
on the use of statisticsin articlesinjournals in these fieldsduring 1975.
Before thedata are presented,some definitionsare necessary.Journals
oftencontain not onlyarticlesbut also editorials,news items,obituaries,
etc. In thisstudythebasiccriteriafordecidingwhatwas an articlewere( 1)
identificationas an articleby the editor in the table of contentsor (2)
presence of an author'sname along withthe titlein the tableof contents.
Regular columns(i.e., regularlyappearing features)were not takento be
articles; neitherwere editorials,news stories,and lettersto the editor.
Doubtfulitemswere counted as articlesiftheyhad at least one reference.
Inevitably,some decisionshad to be made ratherarbitrarily, butat leastall
such decisionswere made byjust one person,and the author made them
before examining an item to see whetherit contained statisticsor not.
An articlewas counted as using statistics ( 1) ifit mentionedquantitative
data for the purpose of comparison (e.g., of one institutionagainst
another,one situationagainstanother,or one timeagainstanother),or (2)
if the quantitativedata served to distinguishan institutionor situation
from others of the same general type - as in "LibraryA has 500,000
titles"(viz., not 5,000,000 or 5,000) - provided thatthe distinctionwas
essentialto the main themeof the article.Articlesthatincidentallymen-
tioned quantitativedata forbackgroundinformationwere not counted.
For example, the sentence,"At the beginningof 1974 therewere forty-
eightEDCs [European DocumentationCentres]in the United Kingdom,
mostlylocated withinuniversityor polytechniclibraries,"14appeared in
6
and Inferential
TeachingDescriptive Statistics
in LibrarySchools
an articlewhich,despiteitscontaining a wealthof information, was not
countedas.usingstatistics. Finally,vaguestatements suchas "about50 per
cent"or "over40,000readers"werenottreatedas uses of statistics.
For the purposesof thisstudy,an articlewas deemed to have used
inferential - as distinguished fromdescriptive - statistics
ifitemployed
suchtechniquesas: analysisof variance;chi-squaretestof association;
chi-squaretestof goodness-of-fit; confidenceintervalsforpopulation
parameters; F-testforthecomparison ofvariances;t-test ofthedifference
of means;or comparabletechniques,eitherparametric(i.e., based on
properties ofthenormal,or Gaussian,distribution) or non-parametric.15
Withperhapsa slightstretch ofthedefinition, an articlethatusedcorrela-
tionor regression wascountedas an instanceofinferential statistics
even
iftheauthorfailedto stateexplicitly a hypothesis aboutthepopulation
correlation coefficient.This does notmean,however,thatarticlescon-
tainingloosestatements aboutthe"correlation" oftwophenomenawere
counted.Foran articletobe includedintheinferential statistics
category,
theauthorhad to showclearlythathe or shewasreferring toa standard
correlation coefficient(e.g.,Pearson,Kendall,Spearman),notmerelya
subjectivelyjudged tendencytowardsimilarbehavior.
In thestudyallarticlesappearingin 1975in 36journalsinthefieldsof
library and information sciencewereexamined.Thesejournalswereall
thenationaland international scholarlyjournalsin thefieldsof library
and information sciencethat(1) were availablein the libraryof the
University ofTexas atAustinand (2) wereinEnglishorFrench.Regional
American journals(e.g.,statelibrary associationjournals)andjournalsin
otherlanguageswereexcluded.
Thejournalsexaminedconstitute nearlyall ofthepopulationofschol-
arlyjournals in Englishin libraryand information science,plus two
journalsinFrench.One ofthejournals,Scientific andTechnical Information
Processing, consistsof translationsof "selectedmajor articles"from
Nauchno-Tekhnicheshaya Informatsiya,Seriya1, and thusrepresents someof
theliterature inRussian.It seemsfairtoconsiderthejournalsinthestudy
as constitutinga substantialportionofall thescholarly journalliterature
inlibrary andinformation sciencefor1975,anda verysubstantial portion
oftheEnglish-language subsetofthatliterature. The year1975wasused
becauseit was the mostrecentyearforwhich,owingto delaysin the
bindery, thejournalliterature was conveniently accessible.
The basicdatafromthestudyappearin Table 1. The totalnumberof
articlesexaminedwas 1,157,ofwhich522,or 45.1 percentmadeat least
one use ofstatisticsthatwasconsideredworthcountingaccordingtothe
criteriaexplained earlier.Every journalthatwasexaminedturnedoutto
haveat leastone articlewithstatistics. Some34, or 2.9 percentofall the
articles,made at leastone count-worthy use of inferential an
statistics,
7
JOURNAL OF EDUCATION FOR LIBRARIANSHIP
Collegeand ResearchLibraries 44 23 4
Documentation etBibliothèques 25 10 -
DrexelLibraryQuarterly 30 6 -
I FLA Journal 12 2 -
InformationProcessingand
Management 23 12 1
International
Forumon Information
and Documentation 5 1 -
International
LibraryReview 54 36 -
JournalofAcademicLibrarianship 20 8 -
Journalof theAmericanSociety
Science
for Information 45 26 8
JournalofDocumentation 22 16 2
JournalofEducationfor
Librarianship 18 8 1
JournalofLibrarianship 21 13 4
JournalofLibraryAutomation 14 11 -
JournalofLibraryHistory 24 8 -
Law LibraryJournal 24 11 -
LibraryAssociation
Record 60 21 -
LibraryHistory 3 1 -
LibraryJournal 53 20 -
LibraryQuarterly 23 12 3
LibraryResourcesand Technical
Services 32 13 1
LibraryTrends 55 36 -
Libri 27 14 1
MicroformReview 10 6 -
8
and Inferential
TeachingDescriptive in LibrarySchools
Statistics
Researchin Librarianship} 4 4 1
RQ 26 7 -
SchoolMedia Quarterly 22 7 3
and TechnicalInformation
Scientific
Processing 48 23 1
SLJ¡SchoolLibraryJournal 32 3 -
SpecialLibraries 67 30 1
UnescoBulletinfor Libraries 34 18 -
WilsonLibraryBulletin 42 13 -
Totals 1,157 522 34
*Forease ofreading,"0" is replacedby"- ".
tOnlyissues28 and 29 wereavailableforthestudy.
65.4 - -
LibraryTrends
BulletinoftheMedicalLibrary
Association 63.8 2.1 3.3
JournalofLibrarianship 61.9 19.0 30.8
Review 60.0 - -
Microform
Journalof theAmericanSociety
Science
for Information 57.8 17.8 30.8
and Technical
Scientific
InformationProcessing 47.9 2.1 4.3
Law Library 45.8 - -
Journal
SpecialLibraries 44.8 1.5 3.3
JournalofEducationfor
Librarianship 44 .4 5.6 12.5
LibraryResourcesand Technical
Services 40.6 3.1 7.7
10
and Inferential
TeachingDescriptive in LibrarySchools
Statistics
Statistics
table 3. JournalswithArticlesUsingInferential
Percentof
statistical Numberof
articles Numberof articles
thatused articles using
inferential using inferential
Journal statistics statistics statistics
SchoolMedia Quarterly 42.9 7 3
JournaloftheAmericanSociety
Science
for Information 30.8 26 8
JournalofLibrarianship 30.8 13 4
LibraryQuarterly 25.0 12 3
AslibProceedings 8.3 12 1
InformationProcessingand
Management 8.3 12 1
LibraryResourcesand Technical
Services 7.7 13 1
Libri 7.1 14 1
and Technical
Scientific
InformationProcessing 4.3 23 1
AustralianLibraryJournal 3.7 27 1
BulletinoftheMedicalLibrary
Association 3.3 30 1
SpecialLibraries 3.3 30 1
Whathappenswhenone examinesthesedatabythemethodsofinfer-
ential statistics?Using the Kolmogorov-Smirnov two-sampletestof
goodness-of-fit,21 one findsthatthe observedvalue,0.24, of the test
is muchlessthanthecriticalvalue,0.48,which(withsamplesof
statistics
14 and 18, respectively) mustbe exceeded for rejectionof the null
hypothesis at the 5 per cent The generalnull
level of significance.
hypothesis of theKolmogorov-Smirnov testis,Ho: The twopopulations
fromwhichthesamplesweredrawnhavethesamedistribution function.
In thisexamplethe null hypothesis maybe particularizedthus:The
variousreasonsforusingthelibrary are snared,inthesameproportions,
by directors and non-directors.
14
Teaching Descriptiveand InferentialStatisticsin LibrarySchools
In short,when one examines the data of the example fromthe view-
point of inferentialstatistics,one realizes that the apparent difference
betweendirectorsand non-directorsin theirreasons forusing thelibrary
is probablyjust the kind of differencethatarises by chance alone out of
typicalvagariesof the process of sampling.The evidence simplydoes not
justifythereportedconclusionthatdirectorsand non-directorsgenerally
differin theirreasons for using the library.
One need not be versed in the Kolmogorov-Smirnovtestto check the
conclusion the article draws from Table 5. Though the Kolmogorov-
Smirnov procedure is preferable for dealing with Table 5, the better-
knownchi-squaretestof associationcould be used. An appropriatewayof
using it entailsignoringthe "workin quiet place" categoryand merging
the "literaturesearch" and "obtaincitation"categoriesto avoid expected
frequenciesless thanone.22The resultis a chi-squarescore of 4.54 with2
degrees of freedom, well under the criticalvalue of 5.99 needed for
rejection,at the 5 per cent level,of the null hypothesisof no association
between being a directoror non-directorand reasons for using the li-
brary.
Both the chi-square and the Kolmogorov-Smirnovprocedures show
thatthearticle's"mostimportantfinding"is notsupportedbyitsdata. The
pointis thatitsauthorfailedto recognizethe possibility thathis data were
inconclusive and to test for that possibility.Now, all students in an
elementarycourse in inferentialstatistics willworkat leasta fewproblems
using thechi-squaretestof association.The author would argue thatany
such studentexperiencesenough of the vagariesof smallsamples to have
asked himselfor herselfwhetherthe frequenciesin Table 5 reallysup-
ported the conclusiondrawn in thisexample, and thatany such student
would thus have avoided the error.
Even an elementaryunderstandingof theconceptsof inferentialstatis-
ticswould have saved theauthorofthisexample fromstatingin print- as
"themostimportantfinding"of his study- a conclusionwhichhis own
data belie. But more important,such an understandingwould have saved
his readers fromacquiringas factan unsubstantiated,and quite possibly
false,conclusion.
A final example. While the author was writingthis paper, a former
studentdropped in to visit.During the conversationhe casually men-
tioned thathe had just found a potentiallyembarrassingmisuseof statis-
ticsin a requestfora grantcurrentlybeing draftedin thelibrarywherehe
now works.Because he knewenough about statistics, he notonlykepthis
institutionfrombeing embarrassedby the misuse but also quite possibly
preventedthe proposal itselffrombeing considered unfavorablyby the
prospectivefundingagencybecause of the error.In short,his knowledge
of statisticsmay well mean thousands of dollars to his libraryin this
15
JOURNALOF EDUCATION FOR LIBRARIANSHIP
instancealone.
The problemsof makingmistakesand missingopportunitiesto probe
more deeply into phenomena constitutethe main argumentforteaching
inferentialstatisticsin libraryschools. One should not overlookthe addi-
tional argument of enabling students to understand the professional
literaturebetter.
But can libraryschool studentslearn inferentialstatistics? The present
is
paper arguing thatthe answer is "Yes," and thisanswer is supportedby
experience in teaching inferentialstatisticsin the Graduate School of
LibraryScience, Universityof Texas at Austin.
For the past six years the course, Research in Library Science, has
included an introductionto descriptiveand inferentialstatistics.Other
parts of the three semesterhour course are an introductionto scientific
research,trainingin the preparation of research proposals, trainingin
surveymethods, and brief treatmentsof other areas of library-related
research, viz., content analvsis, stylostatistics,historicalresearch (with
emphasis on cliometrics,the newly developing area of quantitative
methods in historicalresearch),and analyticbibliography.
This course appears quite similar in content to the quantitative-
methods course at Loughborough Universitythat Brittain23outlines,
except that it places less emphasis on surveymethods and much more
emphasis on statistics.These differencesin emphasis stemnot onlyfrom
thecase forteachinginferentialstatisticspresentedin thispaper, but also
fromthebeliefthatstudents(and professionals)can learn surveymethods
on theirown muchmoreeasilythantheycan learn statistics on theirown.
To guide the autodidact in surveymethods,there is an abundance of
good textbooks.The appendix to this paper offerssome suggestions.
The would-be autodidact in statistics,however,faces greaterdifficul-
ties. There is simplyno satisfactorytextbookin English24on statistics
designed specificallyfor librariansand informationscientists,although
H. Goldhor's book25on researchprovidesa usefulintroduction.A recent
efforttowardsuch a text26turnedout, unfortunately, to be "repletewith
errors"27thatmake itunsuitableforthestatisticalnovice.For textbooksin
statistics, theTexas course has relied on Goldhor'sbook and on introduc-
tory texts for studentsin psychologyand sociology(for details, see the
appendix).
The intrinsicdifficulties of statisticsand thelack of suitabletextsaimed
at librariansand informationscientistswould sufficeas reasons for em-
phasizing statisticsover surveymethods in libraryschool courses in re-
searchor quantitativemethods.But thereis stillanotherreason: It is now
possible to teach students computer-assistedstatistics.Computer pro-
gramscan handle thetediousarithmeticthatused to burden studentsand
statisticians.The student,freed fromarithmetictoil,can concentrateon
16
and Inferential
TeachingDescriptive in LibrarySchools
Statistics
theconceptsofstatistics insteadofthecalculations. Furthermore, compu-
ter programscan easilyhandle problemslarge enough and realistic
enoughtoprovidethestudentadditionalincentives forlearning.Practice
withsuchprograms helpsstudents tobecomeawareofhoweasyitreallyis
nowadaysto getanswersto statistical problems.Clearly,itwillordinarily
be mucheasier (foradministrative, ratherthan technical,reasons)to
provide instruction in computer-assisted to organizedclassesin
statistics
universitiesthanto autodidacts.
The Texas coursein researchinlibrary scienceintegrates thelearning
of statisticalconceptswithpracticein usingseveralstatistical program
packages,28 includingBMD,29OMNITAB II,30and SPSS.31Barelyhave
thestudents learnedwhatthearithmetic meanisbeforetheylearnhowto
makeOMNITAB calculatemeansforthem.Bytheend ofthecoursethe
studentsare usingthecomputerto performregression analysesand to
drawdatapointsand trendlineson a computer plotter. Alongthewaythe
students learntheconceptsofinferential and someofitsprinci-
statistics
pal techniques(analysisof variance,chi-square,confidenceintervals,
correlation, regression, and t-test)byworking library-oriented problems
withcomputerassistance.
The totalcostofcomputertimeand suppliesfortheproblemsis about
$15 per student.For thismodestexpenditurebytheuniversity thestu-
dentsacquirenotonlya knowledge ofinferentialstatisticsbutalsoa good
deal ofpracticein usingcomputers and inbatch-mode.
interactively The
practiceiseffective inovercoming thequalmsaboutusingcomputers with
whichmanystudentsbeginthecourse.Furthermore, thestatistical
pro-
grampackagesused in the courseare widelyavailable.The students'
attentionis called to the factthatmanyof themwillbe workingfor
academicinstitutions, municipalgovernments, schoolsystems, and com-
panies with computersystems on which these or comparablestatistical
programpackages are or can be installed.
Lestanyonethinkthatstudentsemergefromthecourseable to work
problemsonlywiththeaid of a computer,
statistical it shouldbe added
thatmanualmethodsarenotneglected.However,arithmetic forthesake
of arithmetic is avoided,and studentsare encouragedto use personal
electronic calculators. Quite inexpensivemodels(under$30) nowoffer
directcalculations ofmeanand standarddeviation. ForstudentuseTexas
provides three electronic calculators - Hewlett-Packard models67 and
45 and a CommodoremodelS-61("Statistician") - and thestudents are
requiredto learnto use all three.
Experienceindicatesthata coursein researchor quantitative methods
oughtto comeearly,ratherthanlate,in theeducationof thefledgling
librarian or information AtTexas theresearchmethodscourse
scientist.
used to be scheduledforthestudent'sterminalsemester, as one of the
17
JOURNALOF EDUCATION FOR LIBRARIANSHIP
APPENDIX
Recommended
Workson SurveyMethods
Berdie, D. R., and Anderson,J. F.: Questionnaires: Designand Use. Metuchen, New Jersey,Scarecrow,
1974.
Hyman, H. H., etal.: Interviewingin Social Research.Chicago, Universityof Chicago Press, 1954.
Line, M. B.: LibrarySurveys.London, Clive Bingley, 1967.
Moser, C. A.: SurveyMethodsin Social Investigation. London, Heinemann, 1958.
Oppenheim, A. N.: Questionnaire Designand Attitude Measurement.New York, Basic Books, 1966.
Tauber, M. F., and Stephens, I. R., eds.: LibrarySurveys.New York, Columbia UniversityPress, 1967.
Wheeler, M.: Lies, Damn Lies, and Statistics:The ManipulationofPublic Opinionin America.New York,
1976.
Liveright,
18
and Inferential
TeachingDescriptive in LibrarySchools
Statistics
Comments on Textbooks in Statistics
Beginning withthespring semester ofacademicyear1977-78,thetextbooks instatistics
forthecourse,
Researchin Library Science,at theUniversityof Texas at Austin,are:
Wallis,W. A., and Roberts,H. V.: Statistics:
A NewApproach. Glencoe,Illinois,FreePress,1956.
and
Bruning,J.L., and Kintz,B. L.: Computational
Handbook ofStatistics.
2nded. Glenview,Illinois,Scott,
Foresman,1977.
The previously usedtextbook forthecoursewas:
Hardyck, C. D., and Petrinovich,L. F.: Introduction
to Statistics
for theBehavioralSciences.2nd ed.
Philadelphia,W. B. Saunders,1976.
The approachin thisworkis sound;however, froman
ofthesecondeditionsuffers
thefirstprinting
excessivenumberoftypographical errors(thisauthorcan providea listof errata).
A textin use earlierforthecoursehasjust beenpublishedin a newedition:
Young, R. K., and Veldman, D. J.: Introductory for theBehavioralSciences.3rd ed. New York,
Statistics
Holt,Rinehart
and Winston,1977.
Changesintheneweditionappearto haveincreasedthebook'ssuitability
forlibrary
schoolstudents.
and Notes
References
1. Palmer, D. C, ed.: Planningfor a NationwideSystem
ofLibraryStatistics.
Washington,D.C., U.S.
NationalCenterforEducational 1970.
Statistics,
2. Childers,T. A.: Statistics
That DescribeLibrariesand LibraryService.In: Voigt,M. J.,ed.:
AdvancesinLibrarianship.
NewYork,AcademicPress,1975.Vol. 5, pp. 107-122.
3. Anders,M. E.: Statistical
Information as a BasisforCooperative
Planning.
Library 24:229-
Trends,
244,Oct. 1975.
4. Burns,R. W.,Jr.:AnEmpirical Rationale fortheAccumulationofStatistical
Information.
Library
Resourcesand TechnicalServices,18:253-258, Summer, 1974.
5. Drake,M. A.: Forecasting
AcademicLibrary Growth. andResearch
College Libraries,
37:53-59,Jan.
1976.
6. Leech,S.: YouCan'tAddApplesandOranges.Kentucky Association
Library 39,no.4: 24-27,
Bulletin,
specialissue,1975.
7. Williams,W. H.: HowtoProveYourLibrary's WorthtoItsCommunity. Wyoming Library
Roundup,
31, no. 2:7-8,Tune,1976.
8. See,forexample:Slanker,B. O.: Developing
LIBGIS withStateParticipation.
In: Miele,M.,ed.:
AnnualofLibrary
Bowker andBookTradeInformation.
19thed. New York,R. R. Bowker,1974, pp. 225-228.
9. Leech,ref.6.
10. Purdy,G. F.: Professional
Overview.In: Palmer,ref.1, pp. 25-29.
11. Anders,ref.3.
J.M.: TeachingQuantitative
12. Brittain, MethodsinLibrary Schools.
Journal 9: 108-
ofLibrarianship,
119,April,1977.
13. Simpson,I. S.: BaskStatistics
forLibrarians.
London,CliveBingley,1975.
14. Hopkins.M.: The Bibliographical Resourcesof EuropeanDocumentation Centres.
Journal of
7:84-99,April,1975.
Librarianship,
15. Concisedescriptionsof manyparametric and non-parametric of
techniques inferential
statistics
maybe foundin: Roscoe,J.T.: Fundamental theBehavioralSciences.2nd ed. New York,
ResearchStatisticsfor
Holt,Rinehart and Winston, 1975.
16. Vickery, B. C: AcademicResearchin Library and InformationStudies.
JournalofLibrarianship,
7: 153-160,July,1975.
17. Brittain,ret.12.
18. Burns,ref.4.
19. Blankenship, W.C: Head Librarians:HowManyMen?HowManyWomen?College andResearch
28:41-48,Jan.1967.
Libraries,
20. Fatcheric,J. P.: Surveyof Usersofa Medium-Sized TechnicalLibrary. 66:245-
SpecialLibraries,
251,May/June, 1975.
21. See, forexample:Conover,W. I.: Practical Statistics.
Nonparametric NewYork,Wiley,1971.
22. Fora discussion ofwhatminimum valueofexpectedfrequenciesshouldbe usedinthechi-square
procedure, seepp.235and241of:Snedecor, G. W.,andCochran,W.G.:Statistical
Methods.6thed.Ames,
Iowa,Iowa StateUniversity Press,1967.
19
JOURNALOF EDUCATION FOR LIBRARIANSHIP
ref.12.
23. Brittain,
24. A recentworkin Germandeservesspecialmention:Stock,K. F.: Grundlagen undPraxisder
PullachbeiMünchen,
Bibltotheksstatistik, VerlagDokumentation, handbook
1974.A reference ratherthan
overview
thisworkcontainsan excellent
a textbook, and inferential
of descriptive discusses
statistics,
and providesmanyexamplesofanalysesof library
numerouspossibleapplications, statistics.
25. Goldhor, H.: An Introduction Researchin Librarianship.Urbana, Graduate School of
to Scientific
LibraryScience,University of Illinois,1972.
26. Simpson,ref.13.
27. Norton,H. W.: Reviewof BasicStatistics byI. S. Simpson.Newsletter
forLibrarians, onLibrary
no. 18:8-10,Dec. 1976.
Research,
28. Wyllys,R. E.: InstructionalUse ofStatistical
ProgramPackages:BMD, IMP,OMNITAB II, and
Scienceand Statistics:
SPSS. In: Hogben, D., and Fife,D. W., eds.: Computer on the
TenthAnnualSymposium
pp. 265-270.(NationalBureau of StandardsSpecialPublication
Interface, D.C.,
503.) Washington,
NationalBureauof Standards,1978.
29. Dixon,W. J., ed.: BMDP: Biomedicai ComputerPrograms.Berkeley,California, of
University
California
Press,1975.
30. Hogben,D., Peavy,S. T., andVarner,R. N.:OMNITABII User's Manual.(NBSTechnical
Reference
Note552.) Washington, D.C., NationalBureauof Standards,1971.
Packagefor theSocial Sciences.2nd ed. New York, McGraw-Hill,
31. Nie, N. H., etal. : SPSS: Statistical
1975.
20