106,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Geisteswissenschaft
- Sprache: Englisch
A concise, straightforward overview of research design and analysis, helping readers form a general basis for designing and conducting research The practice of designing and analyzing research continues to evolve with advances in technology that enable greater technical analysis of data--strengthening the ability of researchers to study the interventions and relationships of factors and assisting consumers of research to understand and evaluate research reports. Research Design and Analysis is an accessible, wide-ranging overview of how to design, conduct, analyze, interpret, and present research. This book helps those in the sciences conduct their own research without requiring expertise in statistics and related fields and enables informed reading of published research. Requiring no background in statistics, this book reviews the purpose, ethics, and rules of research, explains the fundamentals of research design and validity, and describes how to select and employ appropriate statistical techniques and reporting methods. Readers gain knowledge central to various research scenarios, from sifting through reports of meta-analyses and preparing a research paper for submission to a peer-reviewed journal to discussing, evaluating, and communicating research results. This book: * Provides end-to-end guidance on the entire research design and analysis process * Teaches readers how to both conduct their own research and evaluate the research of others * Offers a clear, concise introduction to fundamental topics ideal for both reference and general education functions * Presents information derived from the author's experience teaching the subject in real-world classroom settings * Includes a full array of learning tools including tables, examples, additional resource suggestions, complete references, and appendices that cover statistical analysis software and data sets Research Design and Analysis: A Primer for the Non-Statistician is a valuable source of information for students and trainees in medical and allied health professions, journalism, education, and those interested in reading and comprehending research literature.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 473
Veröffentlichungsjahr: 2019
Ähnliche
Table of Contents
Cover
List of Figures
List of Tables
Introduction
Section 1: The Purpose, Ethics, and Rules of Research
1 The Purpose and Ethics of Research
1.1 The Purpose and Risks of Research
1.2 History of Harm to Humans
1.3 Ethical Issues in the Social Sciences
1.4 History of Harm to Animal Subjects in Research
1.5 Ethics, Principles, and Guidelines
1.6 Statutes and Regulations Protecting Humans and Animals in Research
1.7 More About Informed Consent
1.8 The Importance of Freedom to Withdraw
1.9 Separation of Provider–Researcher Role
1.10 Undue Influence
1.11 Anonymity
1.12 Summary
Section 2: Basic Research Designs and Validity
2 Research Validity
2.1 Internal Validity
2.2 External Validity
2.3 Summary
3 Research Designs
3.1 The Lingo
3.2 Between‐Subjects Designs
3.3 Within‐Subjects Designs/Repeated Measures
3.4 Between–Within Subjects Designs (Mixed Factorial/Split‐Plot Designs)
3.5 Latin Square Designs
3.6 Nesting
3.7 Matching
3.8 Blocking
3.9 Nonexperimental Research
3.10 Case Studies
3.11 Summary
Section 3: The Nuts and Bolts of Data Analysis
4 Interpretation
4.1 Probability and Significance
4.2 The Null Hypothesis, Type I (
α
), and Type II (
β
) Errors
4.3 Power
4.4 Managing Error Variance to Improve Power
4.5 Power Analyses
4.6 Effect Size
4.7 Confidence Intervals and Precision
4.8 Summary
5 Parametric Statistical Techniques
5.1 A Little More Lingo
5.2
t
Tests
5.3 The NOVAs and Mixed Linear Model Analysis
5.4 Correlation and Regression
5.5 Logistic Regression
5.6 Discriminant Function Analysis
5.7 Multiple Comparisons
5.8 Summary
6 Nonparametric Statistical Techniques
6.1 Chi‐Square
6.2 Median Test
6.3 Phi Coefficient
6.4 Mann–Whitney
U
Test (Wilcoxon Rank Sum Test)
6.5 Sign Test and Wilcoxon Signed‐rank Test
6.6 Kruskal–Wallis Test
6.7 Rank‐Order Correlation
6.8 Summary
7 Meta‐Analytic Studies
7.1 The File Drawer Effect
7.2 Analyzing the Meta‐Analytic Data
7.3 How to Read and Interpret a Paper Reporting a Meta‐Analysis
7.4 Statistical Software Packages for Conducting Meta‐Analyses
7.5 Summary
Section 4: Reporting, Understanding, and Communicating Research Findings
8 Disseminating Your Research Findings
8.1 Preparing a Research Report
8.2 Presenting Your Findings at a Conference
8.3 Summary
9 Concluding Remarks
9.1 Why Is It Important to Understand Research Design and Analysis as a Consumer?
9.2 Research Ethics and Responsibilities of Journalists
9.3 Responsibilities of Researchers
9.4 Conclusion
Appendix A: Data Sets and Databases
A.1. Contents of a Data Set
A.2. Missing Data and Data Entry Errors
A.3. Database Programs
A.4. Database and Data Set Summary
Appendix B: Statistical Analysis Packages
B.1. SAS
, SPSS
, R, and Stata®
B.2. Sample Statistical Analyses Using SAS
B.3. Summary
Appendix C: Helpful Statistics Resources
C.1. Statistics Textbooks
C.2. Statistics Websites
Glossary
References
Index
End User License Agreement
List of Tables
Chapter 1
Table 1.1 Timeline of events and the evolution of research ethics. ...
Table 1.2 Sample informed consent form.
Table 1.3 Example of poor separation of investigator–clinician role...
Chapter 2
Table 2.1 Erroneous statements of causality.
Table 2.2 Erroneous statements of generalizability.
Chapter 3
Table 3.1 Single‐factor between‐subjects design with two levels of ...
Table 3.2 Single‐factor between‐subjects design with four levels of...
Table 3.3 Single‐factor between‐subjects design with three levels o...
Table 3.4 A 2 × 3 multifactorial between‐subjects design with two i...
Table 3.5 A within‐subjects design with three levels of one within‐...
Table 3.6 A 2 × 3 × 2 × 2 between–within subjects design with one d...
Table 3.7 Single‐case A‐B‐A‐B research design.
Chapter 4
Table 4.1 The null hypothesis, power, and errors.
Table 4.2 The relationship between error variance and the size of t...
Chapter 5
Table 5.1 Fictional illustration of mode, median, and mean.
Table 5.2 Data illustrated in the normal distribution in Figure 5.1...
Table 5.3 Data illustrated in the positively kurtotic distribution ...
Table 5.4 Data illustrated in the negatively kurtotic in Figure 5.3...
Table 5.5 Dependent
t
test using raw and scaled scores.
Table 5.6 Example of a
t
test for independent groups.
Table 5.7 Example of a
t
test for matched pairs.
Table 5.8 Sample study using ANOVA.
Table 5.9 Example of a multifactorial ANOVA.
Table 5.10 Example of a single‐Factor ANCOVA.
Table 5.11 Example of a multivariate analysis of variance study.
Table 5.12 Example of a repeated measures study.
Table 5.13 Sample study of a correlation between two variables.
Table 5.14 Sample study using multiple correlation.
Table 5.15 Multiple regression equation.
Table 5.16 Sample multiple regression as an extension of the study ...
Table 5.17 Example of a Logistic Regression Analysis.
Table 5.18 An example of a study using Discriminant Function Analys...
Chapter 6
Table 6.1 Hypothetical 2 × 2 contingency table for hand preference ...
Table 6.2 Sample Chi‐square table comparing groups.
Table 6.3 Formula and sample calculation for
χ
2
.
Table 6.4 Example of a study using the median test.
Table 6.5 Example of a study using Phi.
Table 6.6 Example of a study using a Mann–Whitney
U
test.
Table 6.7 Example of a Wilcoxon signed‐rank test.
Table 6.8 Example of a Kruskal–Wallis test.
Table 6.9 Example of a study with Spearman's rank‐order correlation...
Chapter 7
Table 7.1 Sample layout of a meta‐analytic table with standardized ...
Table 7.2 Sample layout of a meta‐analytic table with
r
.
Chapter 8
Table 8.1 Abstract of a sample study.
Table 8.2 Introduction section of a sample study.
Table 8.3 Sample methods section.
Table 8.4 Sample fictional results section.
Table 8.5 Discussion section of a sample research report.
1
Table A.1 Sample data set with SAS data step.
Table A.2 A sample data step.
Table A.3 Data from a within‐subjects study.
Table A.4 Larger data set with 1000 participants and 16 variables. ...
Table A.5 Steps for creating a basic Microsoft Access database with...
Table A.6 Data set created with Microsoft Access.
Table A.7 Text file exported from Access database.
Table A.8 SAS data step using pasted data exported from Access as a...
Table A.9 SAS Proc Print output using pasted data from text file.
2
Table B.1 Sample SAS program for conducting an analysis of variance.
Table B.2 Sample SAS Log from an analysis of variance.
Table B.4 Sample SAS results for an analysis of variance (selected portions).
Table B.5 Sample
t
test with confidence intervals run using SAS
.
Table B.6 Sample Chi‐square analysis run using SAS
.
Table B.7 Sample multiple regression analysis run using SAS.
List of Illustrations
Chapter 3
Figure 3.1 Sample Latin square design completion.
Figure 3.2 Sample double Latin square design.
Chapter 5
Figure 5.1 A roughly normal distribution.
Figure 5.2 Positively kurtotic distribution.
Figure 5.3 Negatively kurtotic distribution.
Figure 5.4 A positively skewed distribution.
Figure 5.5 A negatively skewed distribution.
Figure 5.6 A bimodal distribution.
Figure 5.7 Distributions with substantial overlap.
Figure 5.8 Distributions with little overlap.
Figure 5.9 Interaction between intervention and chronicity.
Figure 5.10 Significant main effects of both intervention type and ...
Figure 5.11 Significant interaction effect in the absence of main e...
Figure 5.12 Significant time × treatment group interaction in a rep...
Chapter 6
Figure 6.1 Illustration of a positively skewed distribution.
Chapter 7
Figure 7.1 A funnel plot of effect sizes.
Guide
Cover
Table of Contents
Begin Reading
Pages
xiii
xv
xvi
xvii
xix
xx
xxi
1
3
4
5
6
7
8
9
10
11
12
14
15
13
16
17
18
19
22
20
23
24
25
26
27
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
85
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
149
150
151
152
153
154
155
157
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
195
196
195
198
199
200
201
202
203
204
206
207
208
209
210
211
212
213
214
215
216
217
218
219
221
222
223
224
225
226
227
229
230
231
232
233
234
235
236
237
239
240
238
243
244
245
E1
Research Design and Analysis
A Primer for the Non-Statistician
Leslie D. Rosenstein
University of Texas Southwestern Medical Center
Copyright
This edition first published 2019
© 2019 John Wiley & Sons Inc.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.
The right of Leslie D. Rosenstein to be identified as the author of this work has been asserted in accordance with law.
Registered Office
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
Editorial Office
111 River Street, Hoboken, NJ 07030, USA
For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats.
Limit of Liability/Disclaimer of Warranty
While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.
Library of Congress Cataloging‐in‐Publication Data
Names: Rosenstein, Leslie D., author.Title: Research design and analysis : a primer for the non‐statistician /Leslie D. Rosenstein, UT Southwestern Medical Center.Description: Hoboken, NJ : Wiley, 2019. | Includes bibliographical references and index. |Identifiers: LCCN 2019003539 (print) | LCCN 2019017653 (ebook) | ISBN9781119563624 (Adobe PDF) | ISBN 9781119563617 (ePub) | ISBN 9781119563594(hardback)Subjects: LCSH: Medicine–Research–Methodology. | BISAC: SOCIAL SCIENCE /Sociology / General.Classification: LCC R850 (ebook) | LCC R850 .R67 2019 (print) | DDC610.72–dc23LC record available at https://lccn.loc.gov/2019003539
Cover image: © oxygen/Getty Images
Cover design by Wiley
Dedication
I am grateful for my loving and supportive family who has helped me overcome some hefty obstacles. I would not have been able to write this book without the support of Jean, Marv, Dana, Shari, Kevin, Cory, and Caleigh.
I also want to acknowledge my students at UT Southwestern Medical Center for their patience and interest. Their reactions during class discussions, though hard to read at times, helped guide me in organizing and formulating the chapters of this primer. I especially want to thank Dr. Mallory Jacobs who inspired me to try to write something succinct and user‐friendly for busy physicians who want to be good consumers of clinical research.
List of Figures
Figure 3.1
Sample Latin square design completion 58
Figure 3.2
Sample double Latin square design 59
Figure 5.1
A roughly normal distribution 86
Figure 5.2
Positively kurtotic distribution 87
Figure 5.3
Negatively kurtotic distribution 88
Figure 5.4
A positively skewed distribution 89
Figure 5.5
A negatively skewed distribution 90
Figure 5.6
A bimodal distribution 90
Figure 5.7
Distributions with substantial overlap 92
Figure 5.8
Distributions with little overlap 92
Figure 5.9
Interaction between intervention and chronicity 105
Figure 5.10
Significant main effects of both intervention type and chronicity in the absence of an interaction 106
Figure 5.11
Significant interaction effect in the absence of main effects of gender or hand preference 107
Figure 5.12
Significant time × treatment group interaction in a repeated measures study 115
Figure 6.1
Illustration of a positively skewed distribution 141
Figure 7.1
A funnel plot of effect sizes 151
List of Tables
Table 1.1
Timeline of events and the evolution of research ethics
13
Table 1.2
Sample informed consent form
19
Table 1.3
Example of poor separation of investigator–clinician role
23
Table 2.1
Erroneous statements of causality
38
Table 2.2
Erroneous statements of generalizability
43
Table 3.1
Single-factor between-subjects design with two levels of the independent variable
49
Table 3.2
Single-factor between-subjects design with four levels of the independent variable
50
Table 3.3
Single-factor between-subjects design with three levels of the independent variable and three dependent variables
51
Table 3.4
A 2 × 3 multifactorial between-subjects design with two independent variables
51
Table 3.5
A within-subjects design with three levels of one within-subjects independent variable
53
Table 3.6
A 2 × 3 × 2 × 2 between–within subjects design with one dependent variable
56
Table 3.7
Single-case A-B-A-B research design
63
Table 4.1
The null hypothesis, power, and errors
69
Table 4.2
The relationship between error variance and the size of the
F
statistic
70
Table 5.1
Fictional illustration of mode, median, and mean
80
Table 5.2
Data illustrated in the normal distribution in Figure 5.1
87
Table 5.3
Data illustrated in the positively kurtotic distribution in Figure 5.2
88
Table 5.4
Data illustrated in the negatively kurtotic in Figure 5.3
89
Table 5.5
Dependent
t
test using raw and scaled scores
96
Table 5.6
Example of a
t
test for independent groups
100
Table 5.7
Example of a
t
test for matched pairs
100
Table 5.8
Sample study using ANOVA
103
Table 5.9
Example of a multifactorial ANOVA
108
Table 5.10
Example of a single-Factor ANCOVA
111
Table 5.11
Example of a multivariate analysis of variance study
113
Table 5.12
Example of a repeated measures study
118
Table 5.13
Sample study of a correlation between two variables
122
Table 5.14
Sample study using multiple correlation
123
Table 5.15
Multiple regression equation
123
Table 5.16
Sample multiple regression as an extension of the study in Table 5.14
125
Table 5.17
Example of a Logistic Regression Analysis
127
Table 5.18
An example of a study using Discriminant Function Analysis
129
Table 6.1
Hypothetical 2 × 2 contingency table for hand preference and gender
134
Table 6.2
Sample Chi-square table comparing groups
135
Table 6.3
Formula and sample calculation for
X
2
135
Table 6.4
Example of a study using the median test
138
Table 6.5
Example of a study using Phi
139
Table 6.6
Example of a study using a Mann–Whitney U test
140
Table 6.7
Example of a Wilcoxon signed-rank test
143
Table 6.8
Example of a Kruskal–Wallis test
145
Table 6.9
Example of a study with Spearman's rank-order correlation
147
Table 7.1
Sample layout of a meta-analytic table with standardized mean differences
154
Table 7.2
Sample layout of a meta-analytic table with r
154
Table 8.1
Abstract of a sample study
160
Table 8.2
Introduction section of a sample study
161
Table 8.3
Sample methods section
163
Table 8.4
Sample fictional results section
165
Table 8.5
Discussion section of a sample research report
166
Table A.1
Sample data set with SAS data step
180
Table A.2
A sample data step
181
Table A.3
Data from a within-subjects study
183
Table A.4
Larger data set with 1000 participants and 16 variables
184
Table A.5
Steps for creating a basic Microsoft Access database with forms
188
Table A.6
Data set created with Microsoft Access
189
Table A.7
Text file exported from Access database
190
Table A.8
SAS data step using pasted data exported from Access as a text file
190
Table A.9
SAS Proc Print output using pasted data from text file
191
Table B.1
Sample SAS program for conducting an analysis of variance
198
Table B.2
Sample SAS Log from an analysis of variance
199
Table B.3
Sample SAS output data for an analysis of variance
200
Table B.4
Sample SAS results for an analysis of variance (selected portions)
201
Table B.5
Sample
t
test with confidence intervals run using SAS
205
Table B.6
Sample Chi-square analysis run using SAS
207
Table B.7
Sample multiple regression analysis run using SAS
210
Introduction
In this book, I set out to provide a hopefully, pain-free overview of research methods, design, and analysis. The intended audiences include those in the sciences who wish to conduct their own research without investing several semesters completing coursework in statistics and related fields, as well as those in the sciences, clinical fields, education, and the media who wish to read published research in an informed manner. In the former case, this manuscript will provide a general basis for designing and conducting research, though with the assistance of a statistical consultant. In the latter case, I hope this primer will provide a basis for reading, understanding, and critically evaluating research reports.
For health care providers who wish to read studies and make treatment recommendations to their patients based on study outcomes, I hope this book will be a good reference tool. Research publications can sometimes be full of nuances and jargon that are only meaningful to the trained researcher. Without a clear understanding of research design, validity, and interpretation, the results reported in publications can be misunderstood and applied improperly. Sometimes, the research may be poorly conducted or poorly reported, and a basic knowledge of research design and interpretation can be particularly useful in judging when that is the case. At other times, the research is well done, but difficult to understand without a basic knowledge of research methods.
Professionals working in the media are well aware of their great responsibility in reporting research findings to the public. The media has a special role in providing information to the public while avoiding harm as outlined in the Professional Journalists’ Code of Ethics (Society of Professional Journalists, 2014). That code also mandates that journalists are responsible for the accuracy of their reporting, including verifying the information before it is released.
Carrie Figdor (2017) points out the difficulty presented to journalists in their role of reporting and providing information that is accurate when the material is the product of scientific endeavors. Journalists cannot necessarily rely on authors of scientific reports to provide accurate and valid information, and this quandary has become exponentially worse with the evolution of mass communication tools. Non-peer-reviewed research reports are more readily available to the masses. Moreover, journalists cannot necessarily count on peer-reviewed journals to publish only sound research. Most do, but journalists must be careful, yet, to review and understand the research design as presented along with the results and conclusions.
Journalists must take care, for instance, to not translate a conclusion of an association between two events or variables into a claim of causality. Oftentimes, the correct language to that effect is included in a research publication, but it is incumbent on the journalist to read and understand such language. Otherwise, there is a real and great risk that the public will be misinformed and harmed as a result. In Chapter 9, I discuss this in more detail with respect to specific instances of marked harm being perpetrated unintentionally (e.g. the unsubstantiated fear of the measles vaccine, misinformation about the true risks of chronic traumatic encephalopathy, and misinterpretation of the Women's Health Initiative findings).
The chapters of this book are laid out into four major sections. In Section 1, I briefly review the purpose of research as well as ethics and rules guiding research involving human participants and animal subjects. In Section 2, I walk you through basic research designs and validity. In Section 3, I provide a cursory review of statistical techniques, just enough to make you conversant with your statistical consultant or to be able to comprehend the jargon you find in many research documents. I have also included a chapter on meta-analytic studies. The goal of that chapter is to help you in sifting through reports of meta-analyses, though I also provide some direction in case you ever consider conducting your own meta-analytic study. In the fourth section, I review the how-tos of disseminating research findings, including reporting and presenting research results. I discuss how to prepare a research paper for submission to a peer-reviewed journal. I also talk about the concept of poster presentations and how to submit research more quickly for presentation at a conference.
In Section 4, I also present my concluding remarks. There, I repeat what I emphasize throughout this primer; that is, research and research findings are only as good as the research design. Most importantly, it is crucial to avoid making statements of claims of causality between two conditions, or variables, when the research design does not permit drawing such conclusions with any degree of confidence. Accurate interpretation of research findings is of critical importance. This does not just apply to the authors of the original research but also to others who report about and share research findings and claims more broadly. In particular, I hope to underline the importance and responsibility carried by journalists and others who discuss research claims. Sadly, when research claims are reported and shared with the public without a critical eye or with misstatements about causality, harm may ensue.
Finally, I have prepared appendices with tools for those who are planning to conduct their own research. These contain information about data sets, databases, statistical software programs, and resources for those who want to learn more about inferential statistics. I have additionally included a glossary of many of the terms included in this primer; in the glossary, terms are alphabetized for quick lookup.
Section 1The Purpose, Ethics, and Rules of Research
1The Purpose and Ethics of Research
1.1 The Purpose and Risks of Research
Why do we do research? There are many reasons: to answer a question, to advance understanding of a topic, to evaluate interventions, to predict behavior, to understand differences between groups, and so forth. When we conduct research, we usually start with an inquiry based on theory. We then develop hypotheses. Hypotheses are testable questions or predictions, which are ideally based on theory or pre‐existing knowledge about a topic.
Is there ever a time when research should not be conducted? Yes! Logistically, some research ideas may not have benefits that outweigh the costs of conducting the research. These cost considerations include fiscal costs, time, and effort. But there are also ethical considerations in determining whether research should be conducted and how it is conducted.
In terms of the whether, one might ask if there are any potential harms of the research. Consideration of the issue of potential harm typically refers to the harm that may be incurred by the participants or subjects in the process of conducting the research, but harm can also theoretically result from the findings or the knowledge gained by the research. For instance, what if you want to know whether a necessary, life‐saving treatment causes long‐term cognitive impairments? One might argue that there is no point in “proving” the adverse effects of the treatment if the treatment is required for survival and there are no alternative options. On the other hand, perhaps patients deserve to be fully informed of the potential side effects before deciding whether to pursue treatment versus opting for fate. Perhaps, too, an understanding of negative side effects could lead to the development of strategies and interventions to minimize or reverse them.
As another example, many researchers may be interested in knowing the negative consequences of a certain type and severity of injury. If one stops there in the research, that is, if the research concludes following documentation of the negative impacts of an event or injury, then the research has possibly not improved the human state. However, if used as a starting point, the findings from a line of research investigating those consequences could provide very important information, particularly if it opens the door to further investigations into the mechanisms of those consequences as well as future interventions to address them. In other words, if the next step is taken toward understanding the mechanisms that trigger those sequelae, that understanding can then be used to theorize about and then test potential therapeutic strategies.
The second set of ethical considerations referred to earlier pertains to the how research is conducted. If not carefully planned, research, particularly at the data‐collection stage, can cause harm to the participating humans or animal subjects. Even when experimentation is carefully planned, harm may be inevitable. Harm can occur physically and/or psychologically through injury, misinformation, or misunderstanding.
At times, scientists may face a dilemma in asking whether their research will have more harm than benefit. There may be no clearly right or wrong answer to that question, but it is one that scientists should consider before pursuing a line of research. Ultimately, a researcher wants to be aware of all possible outcomes and impacts that their research may have, not just in the process of conducting the research but also in the process of sharing their results.
Regarding the potential harm to human participants and animal subjects in the process of experimentation, society is replete with examples of individuals and animals being injured physically and/or psychologically by scientific inquiries. This history and the rules and guidelines that have stemmed from attempts to prevent future harm are reviewed next.
1.2 History of Harm to Humans
I am hesitant to introduce the darker side of research in this primer that is, otherwise, intended to be a light overview of research methods. However, understanding harm inflicted on innocent individuals in the name of research throughout recent history puts into perspective the institutional research requirements most of us face when undertaking a new investigation. At times, working through the process of an Institutional Review Board (IRB; which I will discuss in Section 1.6) can be tedious and even frustrating. However, IRB members and other regulators are not trying to make research life difficult. While it does at times seem like we have to overcome irrelevant obstacles, these entities are tasked with the noble and critical goal of preventing a repeating of history. The following pages, therefore, are intended to provide perspective by reminding us of what has come before.
Cruelty inflicted on humans in experimentation actually dates back at least several centuries. In recorded history, convicted criminals were used in studies of human anatomy as far back as the fourth to third century BCE (see Franco (2013), for a review and discussion). Specifically, ancient Greek scientists performed vivisections and dissections on convicted criminals.
Probably the most egregious and well‐known modern examples of unethical, horrific research experiments occurred during the reign and terror of the Nazi regime. Weindling, von Villiez, Loewenau, and Farron (2015) provide a systematic accounting of the atrocities inflicted on tens of thousands of adult and child victims from at least 24 countries during that period. There were a particularly large number of victims from Poland, including both Jewish and Catholic citizens. Many, many cases occurred in the context of imprisonment such as in concentration camps, while other cases occurred under coercion and without consent within psychiatric facilities. Many victims died or were killed in the course of experimentation, with some murders occurring in order to study bodies and body parts. Many other victims lived, but were left with serious, often life‐altering injuries.
As outlined by Weindling et al. (2015), the Nazi experimentation began as an undertaking in eugenics. There, the goal of the Nazis was to wipe the land of individuals from different groups, including those of the Jewish faith, individuals identifying as Roma or Sinti, individuals of “mixed race,” individuals with mental illness, and others. Experimentation grew from X‐ray sterilization to also include studies in which victims were infected with diseases in order to test new drugs. Forced infection included diseases such as tetanus, typhoid, and typhus. Twins sometimes served as comparison controls for their victim sibling in these drug studies, and they were also used in gruesome experiments in which perpetrators attempted to conjoin twins surgically. There were experiments assessing the impact of exposure to high altitude/low pressures and/or extreme freezing temperatures. Anthropometric “data” were collected, including in the form of a “Jewish skeleton collection.” In all, with just what is now known through others’ tenacious inspection of records and interviews, there were a great number of atrocities in the name of experimentation committed against innocent, nonconsenting human beings during the reign of the Nazi terror.
At the same time that the Nazis were inflicting atrocities on humans as research guinea pigs, the Japanese Imperial Army and the Japanese Imperial Guard were, likewise, inflicting great harm on prisoners of war in the name of scientific investigation. Those atrocities have only come to be widely known in more recent years, however. Two major components of the wartime experiments occurred at Unit 731 and other prisoner‐of‐war camps in Japanese‐occupied China, and in an American prisoner‐of‐war camp in Japan.
Unit 731 was a prisoner‐of‐war camp in Manchuria where Chinese inmates, both soldiers and locals, were subjected to horrific experiments as described by Herbert Kikoy (2018) and others. As reported by Kikoy, similar experiments were conducted on Russian and other allied war prisoners at nearby camps. The perpetrators horrendously conducted live vivisections and other biological investigations. These included studies of biological weapons for warfare in which bombs were set off to test the ability to spread airborne infection of gangrene and other bacteria. Other similarities with the Nazi experiments were studies evaluating the impact of exposure to extreme temperatures or decompression. The Japanese Imperial scientists led by a surgeon also tested the effects of X‐ray bombardment, starvation, and sleep deprivation in humans. In addition to the live vivisections, they boiled humans alive, and subjected others to centrifuges resulting in unimaginable terror, pain, and death.
In addition to the atrocities performed on the prisoners at Unit 731 and the nearby war camps, the Japanese Imperial research extended to studies of the plague involving the Chinese populace. The perpetrators bred rats that could be infested with fleas infected with the plague. The goal, again, was to develop biological weapons. Those experiments resulted in several outbreaks of the plague according to Kikoy (2018).
Another component of the Imperialists' atrocious research program occurred in Japan at the medical school of Kyushu Imperial University. One experiment involved injecting an anesthetized prisoner with seawater to test whether it could serve as a substitute for sterile saline solution (McCurry, 2015). According to McCurry, based on review of testimony used during war tribunals, organs were removed from prisoners to assess the impact of surgery on organ systems. One prisoner suffered through the experience of having his skull drilled through to test a surgical intervention for the treatment of epilepsy. Corpses of deceased prisoners of war were also preserved in formaldehyde for medical students to use in their studies. These and other details are discussed in a Japanese book written by a physician who was a young medical student at Kyushu Imperial University during this period; his accounts are reviewed, in English, by McCurry (2015) and others (e.g. O'Flynn, 2015).
Sadly, harmful experimentation inflicted on unknowing or unwilling individuals has also occurred outside of the context of war. In the United States, one of the most infamous and shameful of these experiments, originally known as the “Tuskegee Study of Untreated Syphilis in the Negro Male,” was carried out by the United States Public Health Service. That study concluded in 1972 and involved monitoring the natural course of syphilis even after penicillin became the accepted treatment of choice in the 1940s (Centers for Disease Control and Prevention, 2017c).
As chronicled by the National Center for Bioethics in Research and Health Care (Tuskegee University, 2018), 600 men, including 399 who were found to have syphilis and 201 who served as a control group, were recruited into the study beginning in 1932. In exchange for their participation, they were offered, “medical exams, rides to and from the clinics, meals on examination days, free treatment for minor ailments and guarantees that provisions would be made after their deaths in terms of burial stipends paid to their survivors.” The purpose of the study was to learn about the natural course of the disease. When the participants were initially recruited, there was not a proven treatment for syphilis. However, penicillin became the standard treatment in 1947 or earlier. In spite of this, treatment with the antibiotic was withheld in order to continue the study and the endeavor to learn more about the disease progression. The study was not halted until its existence was made public in a news story in 1972. Through a series of reviews and investigations instigated by the Assistant Secretary for the United States Health and Scientific Affairs, it was determined that the participants had never been adequately informed through a procedure called informed consent. Related, they were never warned of the risks of the disease to themselves, their sexual partners, or their children conceived after infection. They were also reportedly never given the choice to withdraw from the study in order to receive treatment once it became available. Ultimately, the study participants and their families were awarded a sum of more than $9 million as part of a class‐action law suit, but the harm that had been inflicted could not be undone.
There are many examples throughout history of marginalized individuals being used in research without informed consent and with adverse outcomes for those individuals. The Beecher report of 1966 reviewed several unethical or potentially unethical studies funded by the government that caused or may have caused harm to participants (Beecher, 1966). Dr. Beecher conducted his review with a reported sense of urgency due to increased requirements at the time to test new treatments prior to implementation as well as due to the increasing availability of research funds providing more incentive to engage in such research. He also noted that physicians were under pressure to succeed as investigators when seeking promotion in academic settings.
The Beecher report noted an increasing presence of ethical errors, and Dr. Beecher addressed several types of these errors. He reviewed studies with ethical errors in various areas including: known effective treatment withheld (such as in the treatment of rheumatic fever); physiologic studies, such as of drugs or toxins with known deleterious effects, or of potentially harmful surgical procedures; studies to improve the understanding of disease; and a study using X‐ray exposure in newborn infants to determine whether reflux occurs in the normal bladder. Dr. Beecher emphasized the need for informed consent as well as the need for a responsible and compassionate investigator to be present.
One study that gained some notoriety following its inclusion in the Beecher report was the so‐called Willowbrook study. However, subsequent publications have suggested its inclusion was somewhat controversial (see Robinson and Unruh (2008), for a discussion of this and a review of the study). The study involved children with intellectual disability living at the Willowbrook State School. There, new infection (post admission) with a mild form of hepatitis was prevalent. Children were enrolled in a study of hepatitis and its prevention with the informed consent of their parents. In one phase of the study, children treated with gamma globulin were found to have protection from the hepatitis strain in comparison to an untreated comparison group (controls). A subsequent phase of the research was conducted to test the hypothesis that infecting individuals with the virus after injecting them with gamma globulin would induce passive–active immunity against the disease. The hope was that future cases of hepatitis at the institution (and elsewhere) could be reduced. In fact, there was reportedly an 80–85% reduction in incidence of hepatitis among the children and employees at Willowbrook.
As noted earlier, Beecher included the Willowbrook study as one of several unethical or questionably ethical studies. Notably, though, as reviewed by Robinson and Unruh (2008), the ethical issues involved had been carefully considered and methodically reviewed by the lead investigator, Dr. Saul Krugman, and his colleagues. For instance, informed consent had been obtained from the participants' parents, the study protocol and environment had been designed to present the least possible risk to participants, the investigators had considered the pre‐existing risk to the children that was already present (i.e. the prevalence of the infection at the facility), they had carefully reviewed previous cases at the facility and learned that the cases were mild with no deaths, they had considered the potential benefits to the participants themselves in addition to other children facing the same pre‐existing risks, and they had obtained an independent review of the study design from other experts in the field.
In spite of arguments made by various authors regarding the sound ethics of the Willowbrook study, concern about its ethics have continued to be raised as a point of contention in the conducting of research with vulnerable populations such as children and those with intellectual disability. Some authors caution, though, that concerns raised by studies such as the Willowbrook experiments may be over‐applied and actually result in unnecessarily avoiding research among particular groups (e.g. women of childbearing age and children). This avoidance may lead to a lack of research with particular groups and, instead, an applying of research findings from one group to another when those original findings may not generalize across those groups. For instance, as pointed out by Robinson and Unruh (2008) in their review of the Willowbrook study and the fallout of the Beecher report, excluding women and children from research can and has led to assumptions about whether and how women (and children) benefit from the products of research in which those groups have been excluded due to ethical concerns. In other words, as pointed out by the authors, the unique medical issues of these populations are sometimes ignored with differential responses of the groups to standard care remaining unknown (“invisible”). The authors conclude in stating that it would be a “mistake to continue to allow the experiments at Willowbrook to cast a restrictive ethical pall over the participation of vulnerable children in medical research.” The valid issues raised by Robinson and Unruh must, of course, be balanced against the risks to vulnerable parties. As part of that balance, the potential benefit of the research is of paramount importance in the moral and ethical equation.
In this discussion, I have reviewed instances of physical harm being intentionally or unintentionally perpetrated against research participants. Not all harm in research occurs in the physical realm, however. Researchers must also be aware of the potential risk of psychological harm that can occur as a result of research inquiries. Some of the potential psychological risks are reviewed next.
1.3 Ethical Issues in the Social Sciences
Different types of psychological harm can arise in a number of research contexts. In modern times, concerns have been raised about research in the social sciences involving deception. The use of deception is somewhat controversial, and studies of this type may leave participants psychologically affected.
One well‐known study involving deception was the Milgram Experiment on obedience to authority figures (Milgram, 1963). In that study, participants were instructed to press a lever in order to administer shocks to another individual who was in a different room. The participant was told that the purpose of the study was to assess the impact of punishment on learning. In reality, the individual receiving “shocks” was a confederate who was not actually being harmed, but that was not known by the participants. The actual research participants who complied with the instructions were led to believe that they were actually inflicting suffering on another human being. Notably, they were pressured to do so. Based on Milgram’s own description in his report, it is clear that a number of participants experienced distress during the actual experiment. Another concern to be considered, though, is the psychological burden that participants may have experienced upon realizing what they were capable of.
This latter issue is also a concern raised by opponents of bystander apathy studies. Many argue that studies of bystander apathy have the potential to leave participants feeling ashamed. At times, these studies may involve staged observational studies in which informed consent is not first obtained or studies in which deception is employed (i.e. participants do not know the true nature or purpose of the experiment). Some might even consider the popular ABC television show What Would You Do? (Quiñones, 2008–Present) to have the potential to harm unwitting participants upon learning that they had acted apathetically instead of intervening on the behalf of a confederate victim.
A good example of a bystander apathy study was an experiment by Darley and Latané (1968). Theirs was an experiment evaluating the phenomenon of diffusion of responsibility and its impact on people's willingness to intervene on behalf of other individuals. Participants in the study were told that they were taking part in a discussion about personal problems associated with college life, but that the conversations were being held via an intercom system to protect privacy. During the “discussion,” a confederate faked a seizure. The researchers recorded the time it took for the participant to seek help for the “victim” in need. The purpose was to determine if that speed was influenced by the number of other “discussants” the participant believed could also hear the pleas of the victim. Shortly after each participant's responses were recorded, they were informed of the true nature of what had just occurred.
Empirical evidence for the potential psychological harm of bystander apathy studies and other research involving deception has been equivocal (Schwartz & Gottlieb, 1981). Regardless of whether evidence supports concerns about potential harm, though, there are steps that can be taken to reduce such harm and discomfort for the participants.
Pascual‐Leone, Singh, and Scoboria (2010) review practical issues and evidence regarding ethical considerations when using deception. A primary factor addresses whether the benefits of the research outweigh the potential distress for the participants. When there is the potential for distress, researchers should provide debriefing and support following the experiment, and the respective IRB should evaluate the relative merits of the use of deception and whether debriefing should occur (Cheng‐Tek Tai, 2012). There may be times when using deception is necessary and the research answers questions that may eventually improve understanding of human behavior in a way that has positive outcomes for society as a whole. In terms of debriefing, there may be instances in which the process may not be possible, including when there is the risk of subject pool contamination, when debriefing is not practicable, and when there is risk that the debriefing may be more harmful than the deception (Sommers & Miller, 2013).
So far, I have reviewed issues related to physical harm and psychological distress imposed on human research participants. Animals are also frequently utilized in research for a variety of reasons, and they can also suffer. In the following section, I review issues related to the welfare of animal subjects in research.
1.4 History of Harm to Animal Subjects in Research
Many of us can look back with fondness on memories of Elle Woods, J.D. fighting to save animals from unjust harm in research conducted solely for the purpose of developing cosmetics. While Ms. Woods was a fictional movie character with a propensity for all things pink, the issues highlighted in the movie were real. Animals can and do experience pain as has been demonstrated through scientific investigations, including studies showing that certain animals can be taught to avoid painful stimuli or even to seek pain‐mitigating stimuli (for a review and discussion, see National Research Council (2009)). When used in research, then, animals may potentially suffer.
In fact, animals are frequently used in research, and this is due to several factors. These factors include their relatively short life cycle for evaluating effects of treatments over their lifespan, the ease of breeding and controlling genetic lineage, and the ease of reproducing large numbers of potential animal research subjects. Animals are also not free to decline participation in research making it even easier to acquire large sample sizes. Due to these factors and the potential for harm, there are now rules in place to protect the rights and welfare of animals (for instance, see National Institutes of Health (2015)).
In the story depicted in Legally Blonde 2, Ms. Woods addresses the rights of animals, and the story line focuses on the use of animals to evaluate cosmetics. In contrast to this scenario, much research is conducted using particular animals to study human diseases and response to treatments, as well as to develop vaccinations to prevent new cases of diseases. There are historical instances of lifesaving treatments emerging from such research, including the creation of vaccines against rabies, polio, diphtheria, and tetanus. Lifesaving antibiotics have also been developed thanks to animal research and testing, and organ transplant techniques have evolved with the aid of animal research. Drugs to treat various ailments have been developed, and surgical techniques have been improved through animal research. These and more medical advances are reviewed in a document published by the National Academy of Sciences (1991) along with a discussion of animal welfare issues and regulations to protect animal subjects in research.
Unfortunately, there have been incidents involving the inhumane treatment of animals, including in medical research. As noted by Franco (2013), examples of animal cruelty in the name of science date back several centuries with ancient Greek physicians conducting vivisections on animals for anatomical research as long ago as the sixth to fifth century BCE. As reviewed by Franco, studies of animal anatomy and physiology continued through the centuries, with the presence of competing philosophies regarding the sentience of animals.
In addition to potential physical pain to animals in the study of animal physiology, anatomy, and response to treatment, many individuals and groups have raised concerns about harm and suffering that occur in other types of research, including behavioral research. Groups have long objected to studies in which stress is imposed on animals to evaluate their responses in the study of depression and anxiety, for example. There have also been outcries against studies in which very young animals are separated from their mothers to assess the impact of separation and/or isolation. These are serious issues that researchers need to consider and address.
As noted earlier, there have been many benefits of having animals involved in research. In addition to the benefits for humans, animals do potentially benefit from the research, including that animals are sometimes treated with similar drugs, antibiotics, and vaccines as humans, and surgical techniques and developed pain‐control measures can sometimes be applied to animals. However, some authors have noted that there are times when findings from animal research do not translate well to humans, and this may result in harm to humans and/or premature abandonment of potential treatments due to misleading findings from animal studies (Akhtar, 2015). Where there are benefits to employing animal testing, though, these benefits must still be balanced against the cost to the animals in terms of pain, suffering, and lack of freedom. Fortunately, there are steps that can be taken to minimize the suffering of animals, and there have been regulations developed to address this (to be discussed further in Section 1.6).
1.4.1 Summary
In the preceding pages, I have reviewed some dark history in the world of science and research involving humans as well as animals. In some of these cases, most of us would not consider the willful acts of malevolence to fall into the realm of research as they are more accurately characterized as torture and extreme cruelty. Some of the examples reviewed and others prevalent in our history, though, can be characterized as research with good intentions but sometimes ill results for the participants. It is because researchers can sometimes have good intentions while failing to be aware of or mindful of the needs of their participants that we need guidelines, rules, and regulations to protect potential victims. In the following pages, I summarize some of the major milestones in the evolution of research ethics, guidelines, and rules. In Table 1.1, I summarize the timeline of some of these milestones along with some of the more widely recognized occurrences of research harm.
Table 1.1 Timeline of events and the evolution of research ethics.
Year(s)
Event
Sixth to third centuries BCE
Vivisection and dissection of live animals are performed by ancient Greek scientists, with vivisection of convicted (human) criminals in the latter part of that period
1876
Britain's Cruelty to Animals Act – an extension of an earlier version of the Act – is established to protect animals from painful experimentation
1932–1972
Tuskegee Experiment
1935–1945
Unit 731 of the Japanese Imperial Army experiments on Chinese and Russian Prisoners of War; Experiments on American POWs at Kyushu University
≤1939–1945
Nazi Medical Experiments are perpetrated against innocent victims
1947
Nuremberg Trial occurs, and the Nuremberg Code is established
1951
Ethical Standards for Psychology, Section 4 – Ethical Standards in Research is prepared by the American Psychological Association Committee on Ethical Standards for Psychology (Cook et al.,
1951
)
1963
Milgram Experiment on obedience is reported
1964
World Medical Association adopts the Declaration of Helsinki
1966
The Beecher report is published
1966
American Medical Association adopts Ethics Guidelines for Clinical Investigation (American Medical Association,
1966a
,
b
)
1966
The Animal Welfare Act is signed into law in the United States, creating animal welfare regulations (United States Department of Agriculture,
1966
)
1974
National Research Act is signed into law in the United States, creating the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research
1979
The
Belmont Report
:
Ethical Principles and Guidelines for the Protection of Human Subjects of Research
is produced, and the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research is established
1981
United States Department of Health and Human Services signs a revised Code of Federal Regulations for the Protection of Human Subjects, which is later adopted in 1991 by 16 federal agencies involved in the conduct, support, or regulation of research involving human subjects; this group includes the National Institutes of Health (Rice,
2008
)
1985
The Health Research Extension Act of 1985, Public Law 99–158, “Animals and Research” is established as law mandating the development of Public Health Service Policy on Humane Care and Use of Laboratory Animals
1994
NIH Guidelines on the Inclusion of Women and Minorities as Subjects in Clinical Research are established
2009
Presidential Commission for the Study of Bioethical Issues is created
1.5 Ethics, Principles, and Guidelines
As noted earlier, there have been debates about the humane treatment of animals and humans in research for centuries. In the mid‐1800s and early 1900s, physicians and scientists were calling for the ethical consideration of research participants and even informed consent (see Jones, Grady, and Lederer (2016), for a discussion). Historical instances of unethical research, especially those during World War II, have led to the development of formalized codes of ethics, guidelines, and rules regulating scientists in the use of human participants and animal subjects. The over‐riding goals of these dicta are to protect humans and animals from harm, and to ensure that there is benefit without undue suffering of participants or maleficence on the part of investigators.
The Nuremberg code stemmed from the Nuremberg trials, where individuals in the military as well as scientists were convicted of war crimes. These crimes included the atrocities perpetrated by the Nazis against prisoners in their concentration camps (Fischer, 2006). The code was a culmination of evolving views regarding ethical research practices, and summarized as “Permissible Medical Experiments” in the documentation of the Nuremberg Military Tribunals (Germany {Territory under Allied occupations, 1945–1955; U.S. Zone}, 1949). Codified was the essentiality of voluntary consent of human participants (“subjects”). Other aspects of the code included the justification of the research based on potential outcome versus human cost, minimization of harm or injury, the ability of the participants to withdraw from the research, and quality assurance.
In 1964, the World Medical Association adopted the Declaration of Helsinki. The Declaration outlined principles for ethical research, and has been amended several times (World Medical Association, 1964–2013). Many of the principles parallel the Nuremberg code, but with some additions. Among the additions is the admonition that under‐represented groups should be given access to research participation. Another principle states that individuals harmed by research must be compensated and treated.
Following the harrowing revelations of the Tuskegee experiment, the US government took steps to prevent future such atrocities. The National Research Act was signed into law in 1974. That Act included the creation of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (see Centers for Disease Control and Prevention (2017a
