2178216
2178216
2178216
https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374
ebooknice.com
https://ebooknice.com/product/sat-ii-success-
math-1c-and-2c-2002-peterson-s-sat-ii-success-1722018
ebooknice.com
https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312
ebooknice.com
(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master the SAT
Subject Test: Math Levels 1 & 2) by Arco ISBN 9780768923049,
0768923042
https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-
arco-master-the-sat-subject-test-math-levels-1-2-2326094
ebooknice.com
(Ebook) Cambridge IGCSE and O Level History Workbook 2C - Depth Study:
the United States, 1919-41 2nd Edition by Benjamin Harrison ISBN
9781398375147, 9781398375048, 1398375144, 1398375047
https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044
ebooknice.com
https://ebooknice.com/product/research-methods-in-applied-settings-an-
integrated-approach-to-design-and-analysis-5907278
ebooknice.com
https://ebooknice.com/product/ibm-spss-for-introductory-statistics-
use-and-interpretation-4921952
ebooknice.com
https://ebooknice.com/product/biomaterials-in-clinical-practice-
advances-in-clinical-research-and-medical-devices-6790674
ebooknice.com
https://ebooknice.com/product/hepatology-a-clinical-textbook-10463530
ebooknice.com
Understanding and Evaluating Research in Applied and
Clinical Settings 1st Edition Et Al George A. Morgan
Digital Instant Download
Author(s): et al George A. Morgan
ISBN(s): 9780805853322, 0805853324
Edition: 1
File Details: PDF, 4.46 MB
Year: 2005
Language: english
UNDERSTANDING AND EVALUATING
RESEARCH IN APPLIED
AND CLINICAL SETTINGS
This page intentionally left blank
UNDERSTANDING AND
EVALUATING RESEARCH IN
APPLIED
AND CLINICAL SETTINGS
George A. Morgan
Jeffrey A. Gliner
Colorado State University
Robert J. Harmon
University of Colorado School of Medicine
In Collaboration With
Nancy L. Leech
University of Colorado at Denver
and Health Sciences Center
Jerry J. Vaske
Colorado State University
H62.M645 2005
300¢.72—dc22 2005046282
CIP
Preface ix
Acknowledgments xiii
I. INTRODUCTORY CHAPTERS
8 Measurement Reliability 44
v
vi CONTENTS
9 Measurement Validity 50
13 Quasi-Experimental Designs 81
15 Single-Subject Designs 95
References 344
ix
x PREFACE
terial so that the various chapters fit together logically. We realize that our
attempt to use a consistent framework and terminology are in some places
(e.g., research approaches and measurement) nontraditional. However, we
think that the framework and consistency pay off in terms of understanding
and evaluating research methods and data analysis.
As you can see from the contents, there are 38 chapters divided into
eight parts: Introductory Chapters; Variables and Their Measurement; Re-
search Approaches, Questions, and Designs; Introduction to the Evaluation
of Methods and Designs; Introduction to Inferential Statistics and Their In-
terpretation; Selection and Interpretation of Specific Statistics; Application
Chapters; and Evaluation of Research Validity.
Our approach is to relate statistics to research design so one can under-
stand why a particular statistical test was used and how to interpret the re-
sults. The discussion of data analysis (statistics) is based on preceding chap-
ters about research design. We show how the research approach and design
help determine the appropriate statistical analysis.
Our book is neither intended to be an in-depth approach to research de-
sign and data analysis, such as Kazdin (2003), nor to be a text on how to
conduct research; there already are numerous books in these areas. In-
stead, we have tried to make the approach conceptual and relatively brief,
so that the consumer does not need to read a long chapter to understand
why a particular statistic or research design was applied in a particular situa-
tion. Each chapter, except for the last two, is quite short and covers a cir-
cumscribed topic.
Although we have organized the chapters in a logical order, we also
wanted the chapters to “stand alone” as much as possible. This required
some repetition, or at least brief redefinition of terms at various points in
the text. Thus, the reader should have some flexibility to skip chapters or
read them in a different order.
Another strength of this book is the inclusion within the data analysis
chapters of what we call “usable statistical information.” We explain why sta-
tistical significance testing, especially null hypothesis significance testing
(NHST), is often misleading and does not provide the information that is
of most use to the clinician. A statistically significant finding is often misin-
terpreted as indicating the strength of the outcome or relationship. Effects
sizes do describe the strength of relationships between variables. Therefore,
we provide information on effect sizes to accompany particular statistical
tests. Several different effect sizes, their relative strengths and weaknesses,
and how they should be interpreted are described.
Part VII includes three application chapters that should be helpful to the
clinician. Effect size computation and understanding are the foundations
for research synthesis and meta-analysis, which we describe in chapter 35.
Practitioners frequently use meta-analysis to provide evidence for therapeu-
PREFACE xi
This book is based on the Gliner and Morgan (2000) textbook, Research
Methods in Applied Settings (ISBN 0-8058-2992-X), also published by Law-
rence Erlbaum Associates, Inc. (www.erlbaum.com). We acknowledge the
use of several tables, figures, and some text from that book. Chapters 22, 32,
and 33 reprint some materials, including text, tables, and figures, from
Leech, Barrett, and Morgan (2005) SPSS for Intermediate Statistics (ISBN 0-
8058-4790-1) also published by Lawrence Erlbaum Associates, so we thank
our colleagues, Nancy Leech and Karen Barrett, for use of those materials.
We also appreciate and acknowledge that Lippincott, Williams, and
Wilkins (LWW), publisher of the Journal of the American Academy of Child and
Adolescent Psychiatry ( JAACAP), granted permission to republish edited ver-
sions of the 33 articles in the “Clinician’s Guide to Research Methods and
Statistics.” Special thanks to Cordelia Slaughter of LWW for facilitating this
permission and providing us with electronic copies of the articles to use as
the starting point for the book. We also appreciate the encouragement and
editorial assistance of Mina Dulcan, editor of JAACAP, who initially com-
missioned the Clinician’s Guide series and encouraged us to find a source
to republish these columns.
Next we acknowledge the help of our colleagues and collaborators, Hel-
ena Chmura Kraemer, Nancy Leech, and Jerry Vaske, who provided feed-
back on parts of this text and were coauthors on one or more of the original
columns. We especially appreciate the critiques and advice of Helen Krae-
mer, professor of biostatistics at the Stanford University Medical School,
who served as statistical consultant for the Clinician’s Guide series in the
JAACAP. We appreciate the advice and feedback of these collaborators, but
xiii
xiv ACKNOWLEDGMENTS
we, the authors, take responsibility for the content of the book and any er-
rors that might be identified.
We would like to thank several reviewers: Maura MacPhee, Mina Dulcan,
Kenneth Solberg, Karen Atler, Brian Isaacson, and Marty Drell for their
helpful reviews. Several colleagues, Robert Fetsch, Gail Gliner, Ray Yang,
Mark Kretovics, and Gene Gloeckner also provided helpful reviews of cer-
tain chapters. Students in our classes have provided useful suggestions for
improving the manuscript at several points in its development.
Our word processor, Catherine Lamana, was especially helpful in pro-
ducing the final manuscript. Nancy Plummer was essential to the develop-
ment of the manuscript for the JAACAP Clinician’s Guide series, and Don
Quick’s technical help was critical.
Finally, we want to thank our spouses, Hildy, Gail, and Darlene, for their
support during the writing of this book and over the years.
INTRODUCTORY CHAPTERS
This page intentionally left blank
Chapter 1
In this book, all quantitative research is divided into three main types that
we call experimental, non-experimental, and descriptive. The first two
types of research each have two approaches as follows:
Experimental Research
1. Randomized experimental approach (random assignment of participants
to groups with an active or manipulated independent variable1).
2. Quasi-experimental approach (an active independent variable but with-
out random assignment of participants to groups).
1
1A variable is a characteristic of a person or situation that has two or more values (it varies)
in a study. An active independent variable is one such as a treatment, workshop, or other interven-
tion that is given to one group of participants and withheld or given in another form to an-
other group. An attribute independent variable is one that is not given or withheld in the study. It
is a measure of a characteristic of the person or his/her situation.
3
4 CHAPTER 1
Nonexperimental Research
3. Comparative (in which a few groups, based on an attribute independ-
ent variable, are compared; e.g., males are compared to females).
4. Associational (sometimes called correlational, in which two or more
usually continuous variables are related or associated for the same group of
participants). Again the independent variable is an attribute rather than ac-
tive.
Descriptive Research
5. Descriptive approach (answers descriptive questions using only descrip-
tive, not inferential, statistics).
Complex studies often use more than one of these approaches; for ex-
ample, “survey” studies may have descriptive as well as comparative and as-
sociational research questions. Of the hundreds of studies that we have
evaluated, all fit into one or more of these five categories based on the re-
search questions and data analysis. Chapter 11 presents this categorization
scheme in more detail.
There are several reasons to categorize research or, more accurately, re-
search questions, into one of the five approaches. First, we feel that ques-
tions of cause and effect can be answered best with well-controlled random-
ized experiments and to a lesser extent with the quasi-experimental
approach. Neither the comparative nor the associational approach is well
suited to deal with cause and effect, but we realize that some complex statis-
tics such as cross-lag panel correlations and structural equation modeling
may provide some evidence for causality from nonexperimental studies.
Furthermore, most meta-analyses (see chap. 35) separate studies by research
approach. For example, experimental and nonexperimental studies often
are not combined but instead make up separate meta-analyses. Some of
these meta-analyses, composed completely of nonexperimental or observa-
tional studies, may be weak on internal validity (see chap. 17). However, they
may be strong in external validity (chap. 18) and provide important informa-
tion for clinical practice.
Our classification of research approaches and Fig. 1.1 should help the
reader follow the research process from purpose through question/hy-
pothesis to data analysis. For example, the experimental, quasi-experi-
mental, and comparative approaches typically use what we call difference in-
ferential statistics such as the t test or analysis of variance, whereas the
associational approach uses associational inferential statistics such as correla-
tion and multiple regression. As discussed later, all parametric inferential
statistics are relational (special cases of canonical correlation), but we think
that it is helpful educationally to make the preceding distinction, which is
consistent with most basic statistics books.
FIG. 1.1. Schematic diagram of how the general type of statistic and research question corre-
spond to the purpose and approach of the study.
5
6 CHAPTER 1
SAMPLE STUDIES
Definition, Purposes,
and Dimensions of Research
What is research? Smith (1981) suggested that a synonym for research is dis-
ciplined inquiry which
must be conducted and reported so that its logical argument can be carefully
examined; it does not depend on surface plausibility or the eloquence, status,
or authority of its author; error is avoided; evidential test and verification are
valued; the dispassionate search for truth is valued over ideology. Every piece
of research or evaluation, whether naturalistic, experimental, survey, or his-
torical, must meet these standards to be considered disciplined. (p. 585)
PURPOSES OF RESEARCH
Research has two general purposes: (a) increasing knowledge within the
discipline and (b) increasing knowledge within oneself as a professional
consumer of research in order to evaluate and understand new develop-
9
10 CHAPTER 2
A second dimension for examining research is the setting. The term field
takes on many different meanings. Field could be a clinic, school, or home
setting. Laboratory implies a structured setting that is not where the sub-
jects or participants usually live, work, or obtain therapy. In the behavioral
DEFINITION, PURPOSES, DIMENSIONS OF RESEARCH 11
sciences, a laboratory often refers to a room with a video camera and micro-
phones (i.e., a somewhat unnatural setting). Clinical disciplines perform
research in both laboratory and field settings. Laboratory settings provide
better control over extraneous variables, but field settings provide a more
ecologically valid environment.
There is confusion in the social and health sciences about the difference
between quantitative and qualitative approaches to research. Much of the
confusion, as indicated in the last chapter, comes from equating these para-
digms with the way data are collected. Investigators often associate “subjec-
tive” methods of data collection, such as open-ended interviews, with the
qualitative approach and “objective” methods of data collection with the
quantitative approach. The type of data collection is not adequate to distin-
guish between the two paradigms.
PARADIGMS
What is a paradigm? The term, coined by Kuhn (1970), refers to the beliefs
shared by members of a specific scientific community. It is not strictly a
methodology but more of a philosophy that guides how the research is to
be conducted. A paradigm determines the type of questions that will be
asked and how they will be answered.
There currently are two major paradigms within the social and health
sciences. One paradigm, often referred to as the quantitative paradigm, has
been the dominant one and is usually associated with the so-called “scien-
tific method.” Qualitative theorists (e.g., Lincoln & Guba, 1985) use the
term logical positivist to identify the dominant paradigm. Even though the
term is not totally appropriate, we use positivist when referring to the quan-
titative paradigm, in part to distinguish it from quantitative data collection
and analysis. The other paradigm, referred to by Lincoln and Guba (1985)
14
A TALE OF TWO PARADIGMS 15
Is there one reality or are there multiple realities? This issue is the most divi-
sive because seemingly there are either many realities (constructivist) or
one (positivist); both positions cannot exist at the same time. We think that
part of the issue comes from the lack of an agreed-on definition of reality.
Phillips (1992) pointed out that there is confusion because different peo-
ple and different cultures have different views about what is real (which
seems undeniable). In addition, there is the issue of whether or not we can
know which view is correct, or even whether there is a correct view. Because
we might not be able to reach agreement, it does not follow that there is
more than one “reality.”
Quantitative researchers, recognizing that participants have different
perspectives or points of view, report those as variability. In addition, quan-
titative studies often examine factors that are related to and perhaps cause
different perceptions. That a person states a belief does not necessarily
mean that it is “real,” even for that person. Sometimes people lie or are con-
fused about their beliefs. It seems to us that it is important to acknowledge
that people have different perceptions, but it is best to investigate why per-
ceptions seem to be different and to be cautious about assuming that what
one says is what one really believes.
The positivist is usually more confident that bias can be overcome, but both
paradigms need to be sensitive to this issue.
book is that such words should be used with caution and then probably only
after performing a study with a tight, randomized experimental design.
HISTORICAL OVERVIEW
Although there have been ethical problems regarding the treatment of hu-
man subjects throughout history, it is common to begin with the Nazi “re-
search” atrocities of 1933–1945. Nazi research atrocities refer to experi-
ments conducted by respected German doctors on concentration camp
inmates that led to their mutilation or death. Surprisingly, before 1933 Ger-
many had advanced moral and legal regulations concerning consent and
special protections for vulnerable subjects. As a result of the trial of 23 of
these doctors, the Nuremberg code was prescribed by an international
court in 1947 (see Shuster, 1997). Although it is tempting to think that
these deeds could be blamed on prison guards, or rogue scientists, the evi-
dence indicates otherwise (e.g., Pross, 1992).
Examples of ethical problems in American research are not uncom-
mon. Mentally impaired children from the Willowbrook State School in
New York were given live hepatitis A virus. Some say that their parents
were not adequately informed and were even coerced into volunteering
their children for the study. A second example was the Tuskegee study of
the natural course of syphilis, which took place between 1932 and 1972.
The study involved more than 400 impoverished African American men
in Alabama who were studied but not treated, even though antibiotics
were available and commonly used to treat syphilis for more than 25 years
of the study. In the African American community, the long-term effects of
this study include mistrust and suspicion of medical research and of doc-
tors in general.
18
ETHICAL PRINCIPLES IN HUMAN RESEARCH 19
Comprehension. The participants should have the legal capacity and the
ability to understand the information and risks involved so that they can
make an informed decision. Some participants (e.g., children) are not le-
gally qualified to make decisions of consent for themselves, so others must
make the decision for them. This is usually the parent or guardian, but the
child also must assent to the procedure. Comprehension also may be im-
paired in mentally retarded or emotionally disabled persons. To the extent
possible, these persons should be allowed to assent or not, but a third party
(e.g., the legal guardian) should be chosen to act in their best interest.
Voluntariness. The third aspect of informed consent means that the par-
ticipant freely, without threat or undue inducement, has decided to partici-
pate in the study. There should not be any element of deceit, constraint, or
coercion. Persons in authority can elicit unjustifiable obedience from chil-
dren and even from well-educated adults. Also voluntariness is reduced
when the research offers financial or other inducements that the potential
participants would find hard to refuse. Sieber (1992) listed a number of as-
pects of the consent process that should be considered. Rapport should be
achieved, not only because participants are more likely to cooperate, but
because it can strengthen the ecological validity of the study. It is important
that the researcher not rush through this aspect of the study or give the im-
pression that consent is unnecessary. Developing trust and understanding
personal and cultural situations is important, especially for community-
based research done in cultures that are different from the researcher’s.
The research also should be relevant to the concerns of the research popu-
lation and explained in those terms.
The issue of who should provide the consent is easy when the potential
participant is an adult who has the capacity to consent. The issue is less
clear with children. We should not automatically assume that parental or
guardian consent is sufficient, although in most cases it should be. In some
situations there may be a conflict of interest. For example, poor parents of-
fered large payment for their children’s participation might not have the
interest of the child foremost. How is consent obtained? IRBs require a for-
mal signed consent form, except in certain situations specified in the fed-
eral regulations. A signed consent form may be omitted when adult subjects
ETHICAL PRINCIPLES IN HUMAN RESEARCH 21
who have the legal capacity to consent can easily refuse by discontinuing a
phone call with an interviewer or by not returning the survey that was re-
ceived in the mail. It is important, however, that the interviewer or ques-
tionnaire cover letter describe the purpose of the research and any risks in-
volved and state that participation is voluntary. Returning the survey or
answering the questions is the subject’s way of implying consent.
Privacy
Probably the most important concern about research ethics is that the indi-
viduals not be harmed by serving as participants in the study. Risk refers
both to the probability of harm and to the magnitude and type of harm.
There are many possible harms and benefits that need to be taken into ac-
count. Psychological and physical pain or injuries are the most often dis-
cussed, but other risks, such as legal, economic, or social (e.g., embarrass-
ment, stigmatization, or invasion of one’s privacy), should be considered.
Although it is rare to attempt to quantify the risks and benefits of a par-
ticular research study, there should be a systematic assessment of these fac-
tors. The Belmont Report states that the assessment of whether the re-
search is justifiable should reflect at least five considerations:
There are certain ethical principles that researchers should keep in mind
when writing papers and reviewing the work of others. These ethical princi-
ples are no less real than those involving the protection of human subjects.
The requirements for manuscripts submitted to biomedical journals (Inter-
national Committee of Medical Journal Editors, 1997) and the American
Psychological Association (2001) publication manual provide advice and
discussion to supplement most of the issues raised in this chapter.
DISSEMINATION OF RESULTS
The research process is not complete until the results are disseminated to
the public and to interested researchers. Although oral presentations and
publications in semipopular magazines have their place, publishing in ref-
ereed journals is key to the progress of science. Considerable detail should
be provided about the procedures and data analyses so that the research-
er’s work is available for scrutiny by the scholarly community. Refereed
publications also are used to evaluate performance of the researcher, and
they are an important aspect of tenure and promotion at a university. Be-
cause there is considerable pressure, especially for young faculty, to pro-
duce refereed publications, a number of potential ethical problems arise.
24
ETHICAL ISSUES OF PUBLISHING AND REVIEWING 25
Most grant proposals and proposed journal articles are reviewed by peers
who must be careful not to use the ideas of the original authors until they
are published and then give credit. Editors and reviewers must not quote
ETHICAL ISSUES OF PUBLISHING AND REVIEWING 27
proposals they have reviewed unless given explicit permission by the au-
thor.
The process of reviewing requires a good deal of trust and integrity by
the reviewers for the process to work fairly and not be exploitative. Prob-
lems related to fairness of reviews are relatively common, and most funding
agencies and journals have specific policies to deal with them. Usually re-
viewers’ identities are not revealed to the authors, on the assumption that
this will make reviews more candid and negative reviews less open to repri-
sal. On the other hand, others have argued that reviews might be more re-
sponsible and balanced if the identity of the reviewer was known. In fact, in
small fields, applicants can often guess the identity of the reviewer.
Masked, formerly called blind, review occurs when the author’s identity is
not given to the reviewer. This type of review is common for manuscripts,
but it is unusual for grant proposals. The argument for anonymous or
masked review is that it gives a better chance to a new scholar because the
work is judged solely on its merits rather than on the status of the authors.
Again, in small fields, it may not be possible to disguise the manuscripts of
well-known researchers.
Once an article or book is published, a different kind of review takes
place, not just in published book reviews but also in literature reviews and
meta-analyses, in which the reviewers exclude studies judged not to be of
high quality. Or the reviewer may decide to weight studies in terms of their
merit, so some count more than others. Although these practices are a nec-
essary part of the scientific process, they provide the opportunity for poten-
tial abuse and, at the least, hurt feelings.
CONFLICTS OF INTEREST
"Niin vain — ehkä siitä olisi ollut apua"', vastasi hän päätään
pudistaen. "Mutta hän on poissa näkyvistä nyt! Hm! — ehkäpä hän
oli se mies, jota he tulivat tapaamaan tänne eilen illalla? Kun kerran
me olemme tai te olette nähnyt hänet täällä päin, niin ehkä hän asuu
täällä. Kuulkaapas, antakaa Birkemille täydellinen kuvaus hänestä."
Kuvailin hänet Birkemille niin tarkkaan kuin osasin ja sellaisena
kuin muistin hänet Clarence-laiturilta, ja sitten Jifferdene ja minä
jätimme hänet ja lähdimme pois. Ja hetken aikaa olin vähällä kertoa
kumppanilleni Neamoresta, sillä minä olin varma siitä, että Neamore
oli ollut Hollimentin kanssa Warrington-hotellissa edellisenä iltana ja
mennyt hänen kanssaan Delaware-tielle. Mutta muutin mieltä —
kumpi sitten olisi ollut parempi — ja päätin olla mainitsematta
mitään, kunnes olin nähnyt lady Renardsmeren tai kunnes jotakin
uutta oli tullut ilmi. Olin tuskallisen hämmästynyt — Holliment,
Quartervayne, Neamore, lady Renardsmere muodostivat
kummallisen yhdistelmän.
"Hän oli ulkona viime yönä joka tapauksessa, lyönpä vaikka vetoa
siitä!" mutisi Jifferdene. "Panen miljoonan yhtä vastaan siitä, että
hän se Hollimentin surmasi."
Quartervayne
"No vaikka mihin — pois täältä niin pitkälle kuin tietä riittää!" hän
vastasi. "Laittakaa niin, että meri on hänen ja teidän välillänne —
ainakin tällä haavaa. Minä lähden — jopa heti huomenaamulla.
Flushingiin tai Middelburgiin — minulla on kumppaneita
kummassakin paikassa. Tulkaa tekin mukaan — olette nokkela nuori
mies, ja minä pidän teistä huolta. Rahoista ei ole kysymys — minulla
on paljon — aivan yllin kyllin. Tulkaa — voitte ruveta apulaisekseni
siellä. Minä voin hoitaa liikettäni Middelburgista käsin tai Flushingista
tai mistä ulkomaisesta paikasta hyvänsä yhtä hyvin kuin
Portsmouthista tai Lontoosta. Mitä kannattaa jäädä tänne, jossa voi
saada tuon kinun verisen puukon kurkkuunsa — ennemmin tai
myöhemmin. En minä ainakaan jää — pois se!"
"En voi sanoa", hän vastasi. "En ollut milloinkaan kyllin lähellä.
Hän oli siististi puettu mies — eurooppalaiset vaatteet, nähkääs —
paremmin puettu kuin useimmat Hollimentin vieraat siinä yömajassa.
Mutta — kinu!"
"Te ette tunne noita kavereita, nuori mies", väitti hän. "Minä tiesin,
että tuo kinu, saadessaan selville, että Holliment oli myynyt hänen
tavaransa, kostaisi hänelle samoin kuin jokaiselle, joka oli joissakin
suhteissa häneen. Ja hän tiesi, että minä olin Hollimentin
liikekumppani."
"Riippuu siitä, mitä noiden tavarain joukossa oli, herra
Quartervayne", huomautin minä ivallisesti. "Te ainakin tiedätte!
Mutta jatkakaa."
"Ei ole sen enempää kertomista", hän sanoi. "Minä livistin — siksi
päiväksi ainakin. Lähetin tuon kirjelipun Hollimentille teidän
välityksellänne. Lopun tunnette. Kinu lähti hakemaan Hollimentia —
niinkuin tiedätte. Tiedätte myös, mitä sitten tapahtui."
"Vai niin, nuori mies, jos siitä on kysymys", sanoi hän, äänessä
jotakin teennäistä myöntyväisyyttä, "olin kyllä! En lähtiessä — hän
otti minut mukaan vähän matkan päästä, Coshamin puolelta —
kaupungin loppupäässä."
"Minä kai olin nukuksissa?" sanoin. "Sikeässä unessa auton
sisällä."
"No niin, asian laita oli niin, että teidät oli nukutettu", vastasi hän
kylmästi. "Muistatte kai, että hän antoi teille juomaa — kylmää
vastaan — juuri ennen lähtöä? No niin, — siihen oli sekoitettu
jotakin, ymmärrättehän? Aivan vaaratonta kuitenkin — teille ei
koitunut siitä mitään vahinkoa."
"En kerro heille — mitään!" selitti hän. "Ja miten voisin erottaa
kiinalaista toisesta? Kaksi kiinalaista muistuttaa toisiaan kuin kaksi
hernettä — keltaiset naamat, vinot silmät — en voi tuntea erikseen
kinua ja kanua. En!"
"Sanon, etten sitä tiedä!" vakuutti hän. "En tiedä siitä mitään!"
Juna lähtee
"Kuuluuko mitään?"
"Hyvä on", vastasin. "Siis puoli yksi tai pian sen jälkeen täällä."
"Tunsitteko hänet?"
"En — en muista milloinkaan nähneeni häntä ennen — en täällä
enkä muualla."
"Siinäkö kaikki?"
"Siinä."
"Ei", sanoi hän. "Nuo kolme miestä lähtivät tiehensä heti sen
jälkeen kun olin nähnyt lady Renardsmeren kirjoittavan
maksuosoituksen; hän odotti täällä, kunnes hänen autonsa tuli häntä
noutamaan kello kolmelta."
Lähdin pois ja tiesin nyt, että kuka hyvänsä Neamore olikaan, oli
hän esittänyt Hollimentin ja Quartervaynen lady Renardsmerelle Ritz-
hotellissa ja kaiken todennäköisyyden mukaan tämä oli ostanut
noilta kolmelta sen esineen jonka Holliment oli varastanut Chuh
Siniltä ja jonka Chuh Sin oli ryöstänyt herraltansa, sfinksimäiseltä
herra Chengiltä. Mitähän se oli? Ja miksi, vieläkin sanoakseni, ei
herra Cheng halunnut kertoa meille?
Palasin omaan hotelliini aprikoiden näitä salaperäisiä asioita — ja
puoli yhdeltä asteli Jifferdene sisään ilmeisesti pahantuulisena.
"Sama mies! Lähti tiehensä — viime yönä", vastasi hän. "Ja vain
hyvin vähän aikaa sen jälkeen kun me olimme häntä tavanneet.
Lähti Pariisin yöjunalla, senkin tulen vietävä! — koko
tämänaamuinen puuhani tai suurin osa siitä, on mennyt hukkaan."
"No niin", hän vastasi. "En ollenkaan epäile sitä, että Hollimentin
murhasi se kinu, jonka herra Cheng tahtoo saada käsiinsä — enkä
sitäkään epäile, että kinu varasti jotakin herra Chengiltä ja että
Holliment vuorostaan ryösti häneltä. Luulen, että kinu käänsi
Hollimentin nurin narin, niin sanoakseni, etsiessään tuota saalista.
Tuntuuko se teistä todenmukaiselta?"
"Laskin heidät sisään — itse", vastasi hän. "Toinen oli lyhyt paksu
herra; toinen oli nuorempi — luulisinpä hänen olleen juutalaisen. He
kysyivät herra Carria ja minä vein heidät hänen arkihuoneeseensa.
He viipyivät siellä ehkä kymmenen minuuttia — sitten kaikki kolme
lähtivät yhdessä ulos."
"Yksinkö?"
"Yksin!"
"Sanoiko hän perästäpäin teille mitään sen johdosta, että oli ollut
ulkona niin myöhäiseen?"
"Ei sanonut! Hän tuli aivan hiljaa sisään. Hän oli hyvin hiljainen,
hyvin kunnollinen mies."