SPSS for Starters and 2nd Levelers

361 Pages·2015·8.883 MB·English

by  Ton J. Cleophas, Aeilko H. Zwinderman

Ton J. Cleophas · Aeilko H. Zwinderman
SPSS for
Starters and
2nd Levelers
Second Edition
SPSS for Starters and 2nd Levelers

Ton J. Cleophas (cid:129) Aeilko H. Zwinderman
SPSS for Starters
and 2nd Levelers
Second Edition
TonJ.Cleophas AeilkoH.Zwinderman
DepartmentMedicine DepartmentBiostatistics
AlbertSchweitzerHospital AcademicMedicalCenter
Dordrecht,TheNetherlands Amsterdam,TheNetherlands
EuropeanCollegePharmaceutical EuropeanCollegePharmaceutical
Medicine Medicine
Lyon,France Lyon,France
Additionalmaterial tothis bookcan bedownloaded from http://extras.springer.com
ISBN978-3-319-20599-1 ISBN978-3-319-20600-4 (eBook)
DOI10.1007/978-3-319-20600-4
LibraryofCongressControlNumber:2015943499
SpringerChamHeidelbergNewYorkDordrechtLondon
©SpringerInternationalPublishingSwitzerland2009,2016
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilarmethodologynowknownorhereafterdeveloped.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexempt
fromtherelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthis
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained
hereinorforanyerrorsoromissionsthatmayhavebeenmade.
Printedonacid-freepaper
Springer International Publishing AG Switzerland is part of Springer Science+Business Media
(www.springer.com)
Prefaces to the 1st edition
Part I
Thissmallbookaddressesdifferentkindsofdatafiles,ascommonlyencounteredin
clinical research and their data analysis on SPSS software. Some 15 years ago
serious statistical analyses were conducted by specialist statisticians using main-
frame computers. Nowadays, there is ready access to statistical computing using
personal computers or laptops, and this practice has changed boundaries between
basicstatisticalmethodsthatcanbeconvenientlycarriedoutonapocketcalculator
and more advanced statistical methods that can only be executed on a computer.
Clinical researchers currently perform basic statistics without professional help
from a statistician, including t-tests and chi-square tests. With the help of user-
friendlysoftware,thestepfromsuchbasicteststomorecomplextestshasbecome
smallerandmoreeasytotake.
It is our experience as masters’ and doctorate class teachers of the European
College of Pharmaceutical Medicine (EC Socrates Project, Lyon, France) that
students are eager to master adequate command of statistical software for that
purpose.However,doingso,albeiteasy,itstilltakes20–50stepsfromloggingin
to the final result, and all of these steps have to be learned in order for the
procedurestobesuccessful.
The current book has been made intentionally small, avoiding theoretical dis-
cussions andhighlightingtechnicaldetails.Thismeans thatthisbook isunableto
explainhowcertainstepsweremadeandwhycertainconclusionsweredrawn.For
that purpose additional study is required, and we recommend that the textbook
“Statistics Applied to Clinical Trials,” Springer 2009, Dordrecht, Netherlands, by
the same authors, be used for that purpose, because the current text is much
complementarytothetextofthetextbook.
We have to emphasize that automated data analysis carries a major risk of
fallacies. Computers cannot think and can only execute commands as given. As
an example, regression analysis usually applies independent and dependent
v
vi Prefacestothe1stedition
variables, often interpreted as causal factors and outcome factors. For example,
genderoragemaydeterminethetypeofoperationortypeofsurgeon.Thetypeof
surgeon does not determine the age and gender. Yet a software program does not
havedifficultytousenonsensedeterminants,andtheinvestigatorincharge ofthe
analysishastodecidewhatiscausedbywhat,becauseacomputercannotdothings
likethat,althoughtheyareessentialtotheanalysis.Thesameisbasicallytruewith
anystatisticaltestsassessingtheeffectsofcausalfactorsonhealthoutcomes.
At the completion of each test as described in this book, a brief clinical
interpretationofthemainresultsisgiveninordertocompensatefortheabundance
of technical information. The actual calculations made by the software are not
alwaysrequiredforunderstandingthetest,butsomeunderstandingmaybehelpful
andcanalsobefoundintheabovetextbook.Wehopethatthecurrentbookissmall
enoughforthosenotfondonstatisticsbutfondonstatisticallyprovenharddatain
order to start on SPSS, a software program with an excellent state of the art for
clinical dataanalysis.Moreover,itisverysatisfyingtoprovefromyourowndata
that your own prior hypothesis was true, and it is even more satisfying if you are
abletoproducetheveryproofyourself.
Lyon,France TonJ.Cleophas
December2009 AeilkoH.Zwinderman
Part II
The small book “SPSS for Starters” issued in 2010 presented 20 chapters of
cookbook-like step by step data analyses of clinical research and was written to
helpclinicalinvestigatorsandmedicalstudentsanalyzetheirdatawithoutthehelp
ofastatistician.Thebookserveditspurposewellenough,since13,000electronic
reprintswerebeingorderedwithin9monthsoftheedition.
Theabovebookreviewed,e.g.,methodsfor:
1. Continuousdata,liket-tests,nonparametrictests,andanalysisofvariance
2. Binarydata,likecrosstabs,McNemar’stests,andoddsratiotests
3. Regressiondata
4. Trendtesting
5. Clustereddata
6. Diagnostictestvalidation
Thecurrentbookisalogicalcontinuationandaddsfurthermethodsfundamental
toclinicaldataanalysis.
Itcontains,e.g.,methodsfor:
1. Multistageanalyses
2. Multivariateanalyses
3. Missingdata
Prefacestothe1stedition vii
4. Imperfectanddistributionfreedata
5. Comparingvaliditiesofdifferentdiagnostictests
6. Morecomplexregressionmodels
Althoughawealth ofcomputationallyintensive statisticalmethods iscurrently
available,theauthorshavetakenspecialcaretosticktorelativelysimplemethods,
becausetheyoftenprovidethebestpowerandfewesttypeIerrorsandareadequate
toanswermostclinicalresearchquestions.
Itistimeforcliniciansnottogetnervousanymorewithstatisticsandnottoleave
their data anymore to statisticians running them through SAS or SPSS to see if
significancescanbefound.Thisiscalleddatadredging.Statisticscandomorefor
youthanproduceahostofirrelevantp-values.Itisadiscipline attheinterface of
biologyandmathematics:mathematicsisusedtoanswersoundbiologicalhypoth-
eses.Wedohopethat“SPSSforStarters1and2”willbenefitthisprocess.
Two other publications from the same authors entitled Statistical Analysis of
Clinical Data on a Pocket Calculator 1 and 2 are rather complementary to the
above books and provide a more basic approach and better understanding of the
arithmetic.
Lyon,France TonJ.Cleophas
January2012 AeilkoH.Zwinderman

Preface to 2nd edition
Over 100,000 copies of various chapters of the first edition of SPSS for Starters
(Parts I (2010) and II (2012)) have been sold, and many readers have commented
andgiventheirrecommendationsforimprovements.
Inthis2ndedition,allthechaptershavebeencorrectedfortextualandarithmetic
errors,andtheycontainupdatedversionsofthebackgroundinformation,scientific
question information, examples, and conclusions sections. In “notes section”,
updated references helpful to a better understanding of the brief descriptions in
thecurrenttextaregiven.
Instead of the, previously published, two-20-chapter Springer briefs, one for
simpleandoneforcomplexdata,this2ndeditionisproducedasasingle60-chapter
textbook.
The,previouslyused,ratherarbitraryclassificationhasbeenreplacedwiththree
parts,accordingtothemostbasicdifferencesindatafilecharacteristics:
1. Continuousoutcomedata(36chapters)
2. Binaryoutcomedata(18chapters)
3. Survivalandlongitudinaldata(6chapters)
The latter classification should be helpful to investigators for choosing the
appropriateclassofmethodsfortheirdata.
Eachchapternowstartswithaschematicoverviewofthestatisticalmodeltobe
reviewed,includingtypesofdata(mainlycontinuousorbinary(yes,no))andtypes
ofvariables(mainlyoutcomeandpredictorvariables).
Entire data tables of the examples are available through the Internet and are
redundant to the current text. Therefore, the first 10 rows of each data table have
nowbeenprintedonly.
However, relevant details about the data have been inserted for improved
readability.
ix