Schutt Paginile 53 87
Schutt Paginile 53 87
Investrigating
WOELD
Univensity of Massachusetts, Boston
5th Edition
RUSSELL K. SCHUTT
/GSAGE Publications r .
\!7
Thor."nd Oaks
London
New Delhi
.%*i:.-..:,.
Social lmportance
Deductive Research Domestic Violence and the Research Ltrcle tnductive Research
Exploratory Research Victim Responses to Police lntervention Conflicts Among High School Studenx Descriptive Research Sociat Research Ethics Honesty and Openness The Uses of Science
Research on People
Scientific Relevance
Social Research Foundations Finding !nformation Searching the Literature Searching the Web Reviewing Research A Single-Article Review: Formal and lnformal Deterrents to Domestic
Violence An lntegrated Literature Review: When Does Arrest Matter?
or.resti" violence is a major problem in our society, with police responding'to between 2 million and 8 million complaints of assault by a spouse or lover yearly (Sherman 1992:6).
the accused arresting whether determine Minneapolis Police Department began an experiment to spouse abusers on the spot would deter repeat incidents. The study's results, which were widely publicized, indicated that arrest did have a deterrent effect. In part because of this, the Yy'hat can
53
r
I
mbstantial
remain obscure" nisms: Why does research article in solved issuesMany social
modeled after it, and the controversies arising from it will provide good examples for our systematic overview of the social research process. We will examine in this chapter different social research strategies. We will also consider in some detail the techniques required to begin the research process: formulating research questions, finding information, reviewing prior research, and writing a research proposal. Appendixes B, C, and D provide more details. I will use the Minneapolis experirnent and related research to illustrate the different research strategies and some of the related techniques. The chapter also introduces ethical guidelines that should be adhered to no matter what research strategy is used. By the chapter's end, you should be ready to formulate a research question, design a general strategy for answering this question, critique previous studies that addressed this question, and begin a proposal for additional research on the question. You can think of Chapters I and 2 as having introduced, respectively, the "why" and "what" of social research; Chapter 3 introduces the "how."
mcluded that
hbeling theorY cm
Sme social
sci
E[
rklter
A social research question is a question about the social world that you seek to answer
through the collection and analysis of firsthand, verifiable, empirical data. It is not a question about who did what to whom but a question about people in groups, about general social processes, or about tendencies in community change. What distinguishes Internet users from other persons? Does community policing reduce the crime rate? What influences the likelihood of spouse abuse? How do people react to social isolation? So many research questions are possible that it is more of a challenge to specify what does not qualify as a social research question than to specify what does. But that doesn't mean it is easy to specify a research question. In fact, formulating a good I research question can be surprisingly difficult. We can break the process into three stages: lidentifying one or more questions for study, refining the questions, and then evaluating the
l
Refining
lt is even
ri&
an i
rcasonably
-urt
the sociC
questlons.
qlestions may
The bestwa5r
Don't keep
Instad
you can look your list to the blps t0 i
Evaluating
In &e
thid
fu criteriafr
$te; social
Chapter 3
literature, most research articles highlight unresolved issues and end with suggestions additional research. For example, Lawrence Sherman and Douglas Smith, with their olleagues, concluded an article on some of the replications of the Minneapolis experiment m police responses to spouse abuse by suggesting that "deterrence may be effective for a dstantial segment of the offender population. . . . However, the underlying mechanisms rtmain obscure" (Sherman & Smith 1992:706). A new study could focus on these mechadsms: Why does the arrest of offenders deter some of them from future criminal acts? Any rcsearch article in a journal in your field is likely to have comments that point toward unremlved issues. Many social scientists find the source of their research questions in social theory. Some nsearchers spend much of their careers conducting research intended to refine an atswer to m research question that is critical for a particular social theory. For example, you may have mcluded that labeling theory can explain much social deviance, so you may ask whether
& fu
uhat is termed an RFP, a request for proposals. (Sometimes the acronym RFR is used, meanQ request for applications.) The six projects to test the conclusions of the Minneapolis I)omestic Violence Experiment were developed in response to such a call for proposals irom te National Institute of Justice. Or you may learn that the social workers in the homeless rhelter where you volunteer need help with a survey to learn about client needs, which
becomes the basis for another research question.
beling theory can explain how spouse abusers react to being arrested. Hnally, some research questions have very pragmatic 'sources. you may focus on a msearch question posed by someone else because it seems to be to your advantage to do so. Sme social scientists conduct research on specific questions posed by a funding source in
rith
*out
trey may address several research questions at once. Also, it might seem risky to focus on a research question that may lead to results discrepant with our own cherished assumptions *out the social world. The prospective commitment of time and effort for some research questions may seem overwhelming, resulting in a certain degree of paralysis. The best way to avoid these problems is to develop the research question one bit at a time. Don't keep hoping that the perfect research question will just spring forth from your pen. rnstead, develop a list of possible research questions as you go along. At the appropiiate time, you can look through this list for the research questions that appear more than once. Narrow your list to the most interesting, most workable candidates. Repeat this process as long as it
helps to improve your research questions.
staking a research project (and thereby a grant or a grade) on a single problem, and so
mruddd
xas
tud
6c hfiiq
&m 1t
shmld heo
socrAr
*m
rr
f
fr
hn'
do se frnd
mosril
ufr
rtuilar tryic- If fu nfch you rish mfi csrrch qu$imr.CI yn dsh to crffi
;3I
qicx
Social lmportance
Social research is not a simple undertaking, so it's hard tojustify the expenditure of effort and resources unless we focus on a substantive area that is important. Besides, you need to feel motivated to carry out the study. Nonetheless, "importance" is relative, so for a class assignment, student reactions to dormitory rules or something like that might be important enough.
dJoel
For most research undertakings, we should consider whether the research question is important to other people. Will an answer to the research question make a difference for society or for social relations? Again, the Minneapolis Domestic Violence Experiment is an
exemplary case. But the social sciences are not wanting for important research questions. The April 1984 issue of the American Sociological Review, which contained the first academic article on the Minneapolis experiment, also included articles reporting research on elections, school tracking, discrimination, work commitment, school grades, organizational change, and homicide. All of these articles addressed research questions about important social issues, and all raised new questions for additional research.
timingd,
-vour specific
m# rc tC
scholarlyjoun*r
nizations. and m
r,ourself.
Finding lnforns
Conductin_s
Scientific Relevance
Every research question should be grounded in the social science literature. Whether we
afu
of published
journdr
formulate a research question because we have been stimulated by an academic article or because we want to investigate a current social problem, we should first turn to the social science literature to find out what already has been learned about this question. You can be 1 I sure that some prior study is relevant to almost any research question you can think of. ' The Minneapolis experiment was built on a substantial body of contradictory theorizing about the impact of punishment on criminality (Sherman & Berk 1984). Deterrence theory
vour particular asEE that you need to il investigations: this I publication aftr tq
Chapter 3
of Social Research 57
predicted that arrest would deter individuals from repeat offenses; labeling theory predicted Sat arrest would make repeat offenses more likely. Only one prior experimental siudy of this islsue was about juveniles, and studies among adults had yielded inconsistent findings. clearty, the Minneapolis researchers had good reason for another study. Any ,", ."r"*"h question should be connected in this way to past research.
How do we find prior research on questions of interest? You may already know some of the relevant material from prior coursework or your independent reading, but that won,t be enough. When you are about to launch an investigation of a new research question, you must ryply a very different standard than when you are studying for a test or jusi seeking to ,.learn about domestic violence." You need to find reports of previous investigations thaisought to answer the same research question that you wish to answer, not just those that were about a similar topic' If there have been no prior studies of exactly the same research question on which you wish to focus, you should seek to find reports from investigations of very similar research questions. Once you have located reports from prior research similar to the research you wish to conduct, you may expand your search to include investigations about related topics or studies that used similar methods. Sometimes you'll find that someone else has already reviewed the literature on your research question in a special review article or book chapter. For example, Rosemary Chalk and Joel H. Garner (2001) published an excellent review of the research on arrest and domestic violence in the journal , New Directions for Evaluation. Most of the research articles you find will include a literature review. These reviews can help a lot, but they are no substitute for reviewing the literature yourself. No one but you can decide what is relevant for your research question and the research circumstances you will be facing-the setting you will study, the timing of your study, the new issues that you want to inciude in your stuay, and your specific methods. And you can't depend on any published research review for information on the most recent work. New research results about many questions appear continually in scholarlyjournals and books, in research reports from governrnent agenciis and other organizations, and on Web sites all over the world; you'll need to check f* r"w research like this yourself.
Finding lnformation
eonducting a thorough search of the research literature and then reviewing critically what you have found is an essential foundation for any research project. Fortunately, much ofthis inforrnation can be identified online, without leaving your desktop, and an incieasing number of published journal articles can be downloaded directly to your own computer ldepJnding on your particular access privileges). Butjust because there's a lot available online doesn't niean that you need to find it all. Keep in mind that your goal is to find reports of prior researchZ investigations; this means that you should focus on scholarlyjournals ihat choose articles fori publication after they have been reviewed by other sociai icientists-"refereed iournals.-( Newspaper and magazine articles just won't do, although you may find some th.r;i* i,";;;_ ' tant issues or even that summarize sociar science research investigations.
"Is inforai"Ot tf urti"les are judged relevant nor so nilrow that you miss important literature. rates reduce control social "Does informal broad. too probably mal social control effective?" is effective more control "Is social informal narrow. too of burglary in large cities?" is probably in reducing crime rates than policing?" provides about the right level of specificity' Identify appropriate bibliographic databases to search. Sociological Abstracts may meet *urylf yo* r""ar, but if you are studying a question about social factors in illness, you If your focus snouta also search it Medline,the database for searching the medical literature. Abstracts Psychological the online in a search include is on mental health, you'll also want to 1985, since of articles text full psycINFO, the contains also that or the version database, psycARTICiES. To find articles that refer to a previous publication, like Sherman and Berk's study of the police response to domestic violence, the Social Science Citation Index willbe "Google helpiul. In addition, the search engine Google now offers anyone with Web access journals) "Google Print" and selected of text full the Scholar' (which indexes and searches research by selected owned that are books of the text (which digitizes and searches the full libraries).lWhen this book went to press the Google Print project was on hold due to copyright concerns raised by some publishers, while the sealch engine and directory Yahoo was right law [Hafner 2005:C1]')
referChoose a search technology. For most purposes, an online bibliographic database that for unpopular searches However, you need. all be ences the publishedjournal literature will topics or very recent literature may require that you also search Web sites or bibliographies of relevant books.
by copystarting a similar venture that focused only on older books that are no longer covered
lrrror, a tentative list of search terms. Ltstthe parts and subparts of your research question [a.rd any related issues that you think are important: "informal social control," "policing," ,.influences on crime rates," and perhaps "community cohesion and crime." List the authors of relevant studies. Specify the most important journals that deal with your topic.
Chapter 3
ywr search.The sheer number of references you find can be a problem' Abstracts' Depending Efu*social capital" r"t;ir"d ittr93 citations it Sociological you search, of -"{ tu"l-t: tlT-l with and the purposes {o}. ffing conference papers
F;;engfist
@yeus.
heacrtarions Oott
-languagJpublications,_to journal articles ratherthan rnor" amcutt to acquire), and to materials published in oirit
i"i -"
search.Leam as you go. If your search yields qo.Tu"y iF_*, try using more genmore precisely. Ii you have not found much literature, terms &EGh complete until you search your consider d m- Whatever ,J;t ,"1 ,"*"h firr,, don't you find' A search articles many how 5*1id several diff-e; upplou"tt", and have seerl yielded 1,569 hits; 2005 11, September fiifuestic violence,, it siiiotogical Abstracts on of hits dropped number the terms, "effects" onlinn""o"Js" as requiied search 1xp eg
h3ffiL
down your w fulean search logic. lt,soften a good idea to narrow include more of the specific details contain combinations of woids or phrases that flffi likeAND allows you to do this' while
search by requiring that
d5wresearch question. Using the Booleanconnector - g the connector onlAto*r"you to find absffacts containing different words that J Uog. Exhibit 3.1 provides an example'
mean the
&es.tldwH,,
Slw@ro"rr&*..' &.itrffisa**e*c,.,
Ss-w &!J{#"*Mb..
{&dseew
*ddM.ek..
&wu
i&Mw.rd
: :
Eil H
lr.4
ll-r
ILI IC lLr
lr IG
IL lr ti lr Ic lrla
lh lr II
lr II
l5
,n
Use appropriate subject descriptors. Once you have found an article that you consider to be appropriate, take a look at the "descriptors" field in the citation (see Exhibit 3.2). You can then redo your search after requiring that the articles be classified with some or all of these descriptor terms.
Check the results. Read the titles and abstracts you have found and identify the articles that appear to be most relevant. Ifpossible, click on these article titles and generate a list oftheir {eferences. See if you find more articles that are relevant to your research question but that you have missed so far. You will be surprised (I always am) at how many important articles your initial online search missed. Read the articles. Now it is time to find the full text of the articles of interest. If you're lucky, some of the journals you need will be available to patrons of your library in online versions, and you'll be able to link to the full text just by clicking on a "full text" link. But many journa1s, specific issues of some journals, or both will be available only in print; in this case, you'll have to find them in your library or order a copy through interlibrary loan.
the time
1..ou
perhaps dissenarir smJct a coherent fr I,eSSOnS as YOU Cif, rheory and h1pofu tions, to explicarct
of prior research-
tributes to )'our
because
str
even more specific guidance. Arlene Fink's (2005) Conducting Research Literature Reviews: From the Internet to Paper is an excellent guide.
for
Don't thinli
You may be tempted to write up a "review" of the literature based on reading the abstracts or using only those articles available online, but you will be selling yourself short. Many
leave behind as
of, s
]ra
ticipated problerns i
Chapter 3
61
details about methods, findings, and theoretical implications will be found only in the dy of the article, and many important articles will not be available online. To understand, ,rrlipe; and really learn from previous research studies, you must read the important articles, Drnatter how you have to retrieve them. If you have done your job well, you will now have more than enough literature as Hground for your own research, unless it is on a very obscure topic (see Exhibit 3.3). (Of @rse, ultimately your search will be limited by the library holdings you have access to and lf the time jou have to order or find copies of journal articles, conference papers and FhaPs dissertations that you can't obtain online.) At this point, your main concern is to conItEt a coherent framework in which to develop your research question, drawing as many Lssons as you can from previous research. You may use the literature to identify a useful ftory and hypotheses to be reexamined, to find inadequately sfudied specific research quesfus, to explicate the disputes about your research question, to summarize the major findings dpnor research, and to suggest appropriate methods of investigation. Be sure to take notes on each article you read, organizing your notes into standard sections: ftory, methods; findings, conclusions. In any case, write the literature review so that it contibutes to your study in some concrete way; don't feel compelled to discuss an article just hecause you have read it. Bejudicious. You are conducting only one study of one issue; it will only obscure the value of your study if you try to relate it to every tangential point in related
research.
mial
Don't think of searching the literature as a one-time-only venture-something that you have behind as you move on to your real research. You may encounter new questions or unanticipated problems as you conduct your research or as you burrow deeper into the literature.
Qualiry
k adequacY d
librarian to
Antici@c
and go rm57 longer exi*Onc size Web sitesresult, Yo Yahoo)
Exhibit 3.4 illustrates the first problem that you may encounter when searching the Web: the sheer quantity of resources that are available. It is a much bigger problem than when
searching bibliographic databases. On the Web, less is usually more. Limit your inspection of Web sites to the first few pages that turn up in your list (they're ranked by relevance). See what those first pages contain and then try to narow your search by including some additional terms. Putting quotation marks around a phrase that you want to search will also help to limit
oBe
departmeil
sites-
your search-searching for "informal social control" on Google (on September 11,2005)
produced 31,100 sites, compared to the roughly 15,500,000 sites retrieved when I omitted the quotes, so Google searched "informal" and"social" and"controll' Remember the following warnings when you conduct searches on the Web:
Bm
o
o
Evaluitu
Some
Clartf! your goals. Before you begin the search, jot down the terms that you think you need to search for as well as a statement of what you want to accomplish with your search. This will help to ensure that you have a sense of what to look for and what to
ignore.
from th Avoid
on the
Chapter 3
Quality is not guaranteed. Anyorre can post almost anything, so the accuracy
and
adequacy of the information you find are always suspect. There's no journal editor or librarian to evaluate quality and relevance. Anticipate change. Web sites that are not maintained by stable organizations can come and go very quickly. Any search will result in attempts to link to some URLs that no Ionger exist. One size does not fit all. Different search engines use different procedures for indexing Web sites. Some attempt to be all-inclusive, whereaS others aim to be selective. As a
result, you can.get different results from different search engines (such as Google or Yahoo) even though you are searching for the same terms. Be concerned about generalizability. You might be tempted to characterize police department policies by summarizing the documents you find at police department Web sites. But how many police departments a-re there? How many have posted their policies on the Web? Are these policies representative of all police departments? To answer all these questions, you would have to conduct a research projecijust on the Web sites
themselves.
Evaluate the sites. There's a lot of stuff out there, so how do you know what,s good? Some Web sites contain excellent advice and pointers on how to differentiate the good from the bad. I have included one such site in Appendix H. AvoidWeb addiction. Another danger of the enormous amount of information available on the Web is that one search will lead to another and to another and so on. There are
ofasinglcr&
rcseilch.f-di!t
rcyis in APltfl
clrtriil
usodI
Reviewing Research
Effective review of the prior research you find is an essential step in building the foundation for new research. You must assess carefully the quality of each research study, consider the implications of each article for your own plans, and expand your thinking about your research question to take account of new perspectives and alternative arguments. It is though reviewing the literature and using it to extend and sharpen your own ideas and methods that _z!ou become a part of the social science community. Instead of being just one individual \ studying an issue that interests you, you are building on an ever-growing body ofknorvledge \ that is being constructed by the entire community of scholars. The research information you find on various Web sites comes in a wide range of formats and represents a variety of sources. Caveat emptor (buyer beware) is the watchword when you search the Web; following review guidelines like those I have listed will minimize, but not eliminate, the risk of being led astray. By contrast, the published scholarly journal literature that you find in databases llke Sociological Abstracts and Psychological Abstracrs follows a much more standard format and has been subject to a careful review process. There is some variability in the contents of these databases-some journals publish book reviews, comments on prior articles, dissertation abstracts, book reviews, and conference papers. However, most literature you will find on a research topic in these databases represents peer-reviewed articles reporting analyses of data collected in a research project. These are the sources on which you should focus, This section concentrates on the procedures you should use for reviewing these articles. These procedures also can be applied to reviews of research monographsbooks that provide much more information from a research project than that which can be contained in a journal article. Reviewing the literature is really a two-stage process. In the first stage, you must assess r--qach article separately. This assessment should follow a standard format like that represented i Ui, tt "Questions to Ask About a Research Article" in Appendix B. However, you should kkeep " in mind that you can't adequately understand a research study if you just treat it as a series of discrete steps, involving a marriage of convenience among separate techniques. Any research project is an integrated whole, so you must be concerned with how each component of the research design influenced the others-for example, how the measurement approach might have affected the causal validity of the researcher's conclusions and how the sampling strategy might have altered the quality of measures. The second stage of the review process is to assess the implications of the entire set of articles (and other materials) for the relevant aspects of your research question and procedures and then to write an in{egrated review that highlights these implications. Although you can find literature reviews that consist simply of assessments of one published article after another-that never get beyond stage one in the review process-your understanding of the
rrydk-Ar#I**
tude4 bf
Peqt
rs
eF,xperidl
.'s
ffi-*ird-q
frm's(192)E
nAl**ir"tt-4
:rchmfoft{ Theismd ItrEe@g girm 6e qr tosd
asihe
rigbdd
Daiil:
,.Ib
Resr;ardt
dhr
than
rd
inftrmal
ad.f4l TkehYPdfi
rcsearct
t7l-Thl{
Chapter 3
+ The
of Social Research
65
p'me and the quality of your own work will be much improved if you make the effort to mh an integrated review.
In the next two sections, I will show how you might answer many of the questions in fpcndix B as I review a research article about domestic violence. I will then show how the
,-&* of a single article can be used within an integrated review of the body of prior research r &is research question. Because at this early point in the text you won't be familiar with all - mminology used in the article review, you might want to read through the more elaborate
rticle review in Appendix C later in the
course.
and
Antony pate and Edwin Hamilton at the national Police Foundation designed one of the dies funded by the U.S. Department of Justice to replicate the Minneapolis Domestic ffience Experiment. In this section, we will examine the article that resulted from that replirxbn, which was published in the American Sociological Review (Pate & Hamilton 1992).
tb lb d
numbers in brackets refer to the article review questions in Appendix B. Research Question.
Lke Sherman and Berk's (1984) original Minneapolis study, Pate spouse assault experiment sought to test the deterrent (1992) Metro-Dade Hamilton's but with an additional focus on the role of inforcases, violence domestic cffwt of arrest in r social control tll. The purpose of the study was explanatory because the goal was to crplain variation in the propensity to commit spouse abuse [2]. Deterrence theory provided & theoretical framework for the study, but this framework was broadened to include the poposition by Williams and Hawkins (1986) that informal sanctions like stigma and the loss d valued relationships augment the effect of formal sanctions like arrest [4]. Pate and Eamilton's (lgg2) literature review referred, appropriately, to the original Sherman and Berk findings, and to t1984) research, to the other studies that attempted to replicate the original control social [3]. research on informal There is no explicit discussion of qthical guidelines in the article, although reference is made b a more complete unpublished repo\t [6]. Clearly, important ethical issues had to be considere4 given the experimental intervention in the police response to serious assaults, but the dherence to standard criminal justice procedures suggests attention to the welfare of victims as well as the rights of suspects. We will consider these issues in more detail later in this chapter.
The Research Design. Developed as a follow-up to the original Minneapolis experiment, the
Metro-Dade experiment exemplified the guidelines for scientific research that were presented in Chapter 2 t5l. It was designed systematically, with careful attention to specification of terms and clarification of assumptions, and focused on the possibility of different outcomes rather than certainty about one preferred outcome. The major concepts in the study, formal and informal deterrence, were defined clearly t9l and then measured wilh straightforward indicators-arrest or nonarrest for formal deterrence and marital status and employment status for informal deterrence. However, the specific measurement procedures for marital and employment status were noi discussed, and no attempt was made to determine whether they
captured adequately the concept of informal social control [9, 10]. Three hypotheses were stated and also related to the larger theoretical framework and prior research t7l. The stqdy design focused on the behavior ofindividuals [13] and collected data
Alirelft
I
-trrfus-J rqi-
rfstheffi
tq#rAt*
lh es{ of fre m 5m seprre mi*
rucuchIt6-l$7t.
fnfii
ftcr
hat {
a Suhsequent Assault
b,
l-
Hove you
=20 (!
o o
E15 (,
o o 3 {n arv
.cl
(E E
2- Is
the
reseortt
"classic"
smfu
3. Ha'e you dt
kt
t E
= Eo-
o o o
Chapter 3
of Social Research
fr
h -est- Although the use of a population of actual spouse assault cases precluded the use drry sophisticated measures of informal social control, the experimental disign of the study rdfu researchers' ability to interpret the results in the contexi of several other comparable
as exceptionally worthwhile. It is not hard to underwhy these studies continue to stimulate further research and ongoing policy discussions.
, h
ffyour
The goal of the second stage of the literature review process is to integrate the results separate article reviews and develop an overall assessment of the implications of
The integrated literature review should accomplish three goals:
iirrcsearch.
'|fr
(l) Summarize (2) critique prior research, and (3) present pertinent conclusions (Hart trIB:18G187).I'll discuss each of these goals in turn.
research,
Utbrinology research article about mandatory arrest policies in domestic violence cases with ft they term a "provocative" question: What is the point of making it a crime for men to ralt theii female partners and ex-partners? They then review the different theories and rlEnrting research that has justified different police policies: the "victim choice,' position, lh ipo-anest" position, and the "victim empoweflnent,, position. Finally, they rwiew the reach on the "controlling behaviors" of men that frames the specific research question on nfch they focus: how victims view the valuerof criminal justice interventions in their own \ res (Hoyle & Sanders 2000:15).
Ask yourself three questions about your summary of the literature:
'%uize p:rior research. Your summary of prior research must focus on the particular rcrch questions that you will address, but you also may need to provide some more genrd background' Carolyn Hoyle and Andrew Sanders (2000:14) begin their B ritish Journil of
l-
Have you been selective? If there have been more than a few prior investigations of your research question, you will need to narrow your focus to the most relevant and highest-quality studies. Don't cite a large number of prior.articles 'Just because they
are there."
L
I
Is the research up-to-date? Be sure to include the most recent research, not just the
"classic" studies-
lz! I
ruow you used direct quotes sparingly? To focus your literature review, you need to express the key points from prior research in your own words. Use direct quotes only when they are essential for making an important point (pyrczak 2005:51-59).
pior research. Evaluate the strengths and weaknesses of the prior research. In to all the points you develop as you answer the "article review questions" in Appendix B- you should also select articles.for review that reflect work published in peer-reviewed.lournals 41d written by credible authols who have been funded by reputable sources. Consider the ftillowing questions as you decide how much weight to give each article: l. How was the report reviewed prior to its publication or release? Articles published in rademic journals go through a rigorous review process, usually involving criticism "-"fut and revision. Top "refereed" journals may accept only I07o of submitted articles, so they can
Critique
uHiti6l
SOCIAL RE
areserchrpaffi
reodr to coosider{
3. Who funded and sponsored the research? Major federal funding agencies and private foundations fund only research proposals that have been evaluated carefully and ranked highly by a panel of experts. They also often monitor closely the progress of the research. This does not guarantee that every such project repoft is good, but it goes a long way toward ensuring some worthwhile products. On the other hand, research that is funded by organizations that have a preference for a particular outcome should be given particularly close scrutiny (Locke, Silverman, & Spirduso 1998:37_/.4).
Present pefiinent conclusions. Don't leave the reader guessing about the implications of the prior research for your own investigation. Present the conclusions you draw from the research you have reviewed. As you do so, follow several simple guidelines:
FtrBss ot (bfirxr$El il&rmrirell- reserrh oniluting the.iaq-* d 3-6r ThLs in&xtive r A,* 1ou'll sce- a re$El
smninqsltlfe
o Distinguish o
clearly your own opinion of prior research from conclusions of the authors of the articles you have reviewed. Make it clear when your own approach is based on the theoretical framework you are
using rather than on the results of prior research.
6t
fl-alrraring
ftEayi
project. Don't empha1 . ,q.nrowledge the potential limitations of any empirical research size problems in prior research that you can't avoid either (Pyrczak 2005:53-56). ?o Explain how the unanswered questions raised by prior research or the limitations of
methods used in prior research make it important for you to conduct your own investi-
isrdr
gation (Fink 2005:190-192). good example of how to conclude an integrated literature review is provided by an article based on the replication in Milwaukee of the Minneapolis Domestic Violence Experiment. For this article, Raymond Paternoster, Robert Brame, Ronet Bachman, and Lawrence Sherman (1991) sought to determine whether police officers' use of fair procedures when arresting assault suspects would lessen the rate of subsequent domestic violence. Paternoster et al. (1997) conclude that there has been a major gap in the prior literature: "Even at the end of some seven experiments and millions of dollars, then, there is a great deal of ambiguity surrounding the question of how arrest impacts future spouse assault" (p. rcD. Specifically, they note that each ofthe seven experiments focused on the effect ofarrest itself, but ignored the possibility that "particular kinds of police procedure might inhibit the recurrence of spouse assault" (p. 165). So Paternoster and his colleagues (1991) ground their new analysis in additional literature on procedural justice and conclude that their new analysis will be "the first study to examine the effect of fairness judgments regarding a punitive criminal sanction (arrest) on serious criminal behavior (assaulting one's partner)" (p. l'72).
Exhibit3S
E
With
re
a research question formulated and a review of the pertinent literature taking shape, we ready to consider the process of conducting our research. 'When we conduct social research, we are attempting to connect theory with empirical
dm-the
bvstarting
evidence we obtain from the social world. Researchers may make this connection *iE? q9gg1-11r"*,ry 4I4I!191r t_eltiqg s:ggre.of its igrpli-c_g-ti9qs with data. This is the
d"ata
@"*ir,;itismost.'ofi.-q-tls*.".+eg-.919$inguantititivemethots'
f,ftffi,atively, researchers may develop a connection GffiAri soCilt ttreory ahil
bf
firSt
odlecting the data and then developing a theory that explains patterns in the data (see Exhibit 3-6)- This inductive research process is more often the strategy used in qualitative.methods<r
As you'll see, a research project can draw on both deductive and inductive
strategies. l
ftggrylelps social scientists to know what to implieations of6"ir-iiudltE, for otliei*reSearifilBuifaiig 1 {u$y-.an{.to spee-ify.the. flenaluating theory is therefore one of the most important objectives of socidi siienie. Most social research is guided by some theory. a\hough the theory may be only partially fueloped in a particular study or may even be u)rrecognized by the researcher. When msearchers are involved in conducting a research project or engrossed in writing a research mport, they may easily lose sight of the larger picture. It is easier to focus on accumulating a clarifying particular findings rather than consider how the study's findings fit into a more general understanding of the social world.
hok !o111
ftely to occur
,Ti*"r;:l[':'lu:ffi;
n
l;r
rli
i[
commit another offense so they can avoid further punishment (Lempert & Sanders 1986:86-87). Deterrence theory leads to the prediction that arresting spouse abusers will
lessen their likelihood of reoffending.
Labeling theory distinguishes between primary deviance, the acts of individuals that lead to public sanction, and secondal:y deviance, the deviance that occurs in response to public sanction (Hagan 1994:33). Arrest or some other public sanction for misdeeds labels the offender as deviant in the eyes of others. Once the offender is labeled, others will treat the offender as a deviant, and he or she is then more likely to act in a way that is consistent with the deviant label. Ironically, the act of punishment stimulates more of the very behavior that it was intended to eliminate. This theory suggests that persons arrested for domestic assault are more likely to reoffend than those who are not punished, which is the reverse of the deterrence theory prediction. Theorizing about the logic behind punishment can also help us draw connections to more general theories about social processes. Deterrence theory reflects the assumptions of ratio' nal choice theory, which assumes that people's behavior is shaped by practical calculations: People break the law if the \nefits of doing so exceed the costs. If crime is a rational choice for some people, then increasiirg the certainty or severity of punishment for crime should shift the cost-benefit balance away from criminal behavior. Labeling theory is rooted h symbolic interactionism, which focuses on the symbolic meanings that people give to behavior (Hagan,
it increase
1994:40).Instead of assuming that some forms of behavior are deviant in and of themselves (Scull 1988:678), symbolic interactionists view deviance as a consequence ofthe application of rules and sanctions to an "offender" (Becker 1963:9). Exhibit 3.7 summarizes how these general theories relate to the question of whether or not to arrest spouse abusers. Specific predictions about the effect of arrest are deduced ftom
general theories about punishment. Does either deterrence theory or labeling theory make sense to you as an explanation for the impact of punishment? Do they seem consistent with your observations of social life? As you have already seen, Patemoster et al. (1997) concluded that something was missing in these theories. Procedural justice theory explains law-abidingness as resulting from a sense of duty or morality and predicts that people will obey the law from a sense of obligation that flows from seeing legal authorities as moral and legitimate (Tyler 1990). From this perspective, individuals who are arrested will be less likelyto reoffend if they are treated faidy, irrespective of the outcome of their case, because fair treatment will enhance their view of legal authorities as moral and legitimate. Procedural justice theory expands our view of the punishment process by focusing attention on how authorities treat subjects rather than just on what decisions they make. Are you now less certain about the likely effect of arrest for intimate partner violence? Will arrest decrease abuse because abusers do not wish to suffer from legal sanctions again? Will
fre impli
keep thinking
Explanatory
The process involves movi
ized with a
Deduclirc
As Exhibit
general We call the
hypothesisftt
a relationship
Chapter 3
Symbolic interactionism
People give symbolic meanings to objects, behaviors, and
assumption
other people.
ffi Y
Criminological component
Deterrence theory: People break the law if the benefits ofdoing so outweigh the costs. Labeling theory: People label offenders as deviant, promoting further deviance.
domestic
assault)
again.
i increase abuse because abusers feel stigmatized by being arrested and thus are more likely h act like criminals? Or will arrest reduce abusd only if the abusers feel they have been
rcarcd fairly by the legal authorities? By suggesting such questions, social theory makes us
mere sensitive to the possibilities and so helps us to design better research and draw out ft implications of our results. Before, during, and after a research investigation, we need to teep thinking theoreticallY.
nch
kplanatory Research
The process of conducting research designed to test explanations for social phenomena fuvolves moving from theory to data and then back to theory- This process can be characterized with a research circle (Exhibit 3.8).
Deductive Research
general theoretical premise and then tested with data that have been collected for this purpose.
We call the specific expectation deduced from the more general theory a hypothesis. It is the hypothesis that researchers actually test, not the complete theory itself. A hypothesis proposes a relationship between two or more yariables---characteristics or properties that can vary.
u"ff5illil""
(it
,1
Ll
HYPothesis
Variation in one variable is proposed to predict, influence, or cause variation in the other variable. The proposed influence is the independent variable; its effect or consequence is the dependent variable. After the researchers formulate one or more hypotheses and develop research procedures, they collect data with which to test the hypothesis.
Hypothesis A tentative sta(ement abodt empirical reality, involving between two or more variables.
a
relationship
Example of a hypothesis.' The higher the poverty rate in a community, the higher the percentage of community residents who are homeless.
Variable
attributes).
Example of a variable: The degree of honesty in verbal statements' Independent vurinble A variable that is hypothesized to cause, or lead to, vaflation in another variable. Example of an independent variable.' Poverty rate.
Depend.ent variable A variable that is hypothesized under the influence oi, anothtr variable.
Chapter 3
IF-THEN
Hypothesis
The greater the use dlnternet, the greater the strength
Ihe risk of
poperty theft
decreases as
income
increases.
If
years
of
IF years of
education
decrease, THEN
education
decrease, income
decreases.
income decreases.
Political
conservatism
increases
.
with IF
THEN property
crime is higher
-compared to crime in suburban
in suburban
or rural areas.
or rural areas.
Hypotheses can be worded in several different ways, and identifying the independent and
lilendent variables is sometimes difficult. When in doubt, try to rephrase the hypothesis as i ff-then" statement: 'f the independent variable increases (or decreases), then the depen-
itet
variable increases (or decreases)." Exhibit 3.9 presents several hypotheses with their L&pendent and dependent variables and their "if-then" equivalents. Exhibit 3.9 demonstrates another feature of hypotheses: direction of association. When
Esearchers hypothesize that one variable increases as the other variable increases, the direcfu of association is positive (Hypotheses 1 and 4). When one variable decreases as the other wiable decreases, the direction of association is also positive (Hyryrthesis 3). But when one wiable increases as the other decreases, or vice versa, the direction of association is nega6rr, or inverse (Hypothesis 2). Hypothesis 5 is a special case, in which the independent raiable is qualitative: It cannot be said to increase or decrease. In this case, the concept of &ection of association does not apply, and the hypothesis simply'states that one category of 6e independent variable is associated with higher values on the dependent variable.
expectation.
In both explanatory and evaluative research, the statement ofexpectations for the findings and the design of the research to test these expectations strengthen the confidence we can place in the test. The deductive researcher shows her hand or states her expectations in advance and then designs a fair test of those expectations. Then, "the chips fall where they may"-in other words, the researcher accepts the resulting data as a more or less objective picture of reality.
frfr mr
S.liI
Dmtfr,rmqh6il
xpsds- to se nt* imyail after arrest d B!, $e dme rd ilf,s apP6rt0r- la t mmh'anhtr- Inissri ir*lems aEry llm*la+he poffi hbofl Srman and [rsf,er fte uigirdl dEr rhan lflbGf I l*o qr*im- IE k as fre restd hllon-up dM m mmd fre rescdr Seifi l99ll
$omptfliilfl.
dtte
sircle
nifr,d
rd
Chapter 3
75
r"\
Theory
\/
\ o^o1
Measures for 330 domestic assault cases
: Data
whether professional counseling helped and whether the length of time spent after arrest mattered at all. By the time resurts were reported from five of the cities in the new study, a problem of the cities_omaha, Nebraska; Charlotte, Norrh Carotina; and LT..o"T:11":i lT,:: wisconsin-researchers '.ffiwaukee, were finding rong-term increases
iliBinal complainant' Some of the replications also examined different aspects of the arrest to see
[Bs'
L.fril
'rhida-the 'ffi[,|::::,Tj,:::^:I predicted deterent liiyororalo. effecrs seemed be^occi,.ring rsrr""r#l"r;;ti;;y; -eman and his colleagues had now traversed to the research Jr"l" t*i"" in an attempt to rsrer the original research question, fust in Minneapolis and then in
slrings,
:rser
i k
JHlow-up data from seleral cities to try to explain arr" ait"r"p"ri r"*ur, thereby starting mund the research circre once again (Berk et ar. r992;pate &Hamirton r99z;Sherman
Smith I992).
Butthan leading to more confidence in deterrence theory the research results were calling into question. Deterrence theory now seemed inadequate to explain empiricar rearity, at as the researchers had measured this reality. So the researche.s began to rcanalyze the
&
lnductive Research
data, which are 6en used to develop (induce) a general explanation (a theory) to account fbr the data. one Tay to think of this process is in terms of the research circle: Rather than starting
at the top of the circle with a theory the inductive researcher starts at the bottom of the circle with data aod then develops the theory. Another way to think of this process is represented in Exhibit i'I1' In deductive research, reasoning from specific premises resultsin a conctusion that a
Deductive
Premise
Premise
1: 2:
All
Conclusion:
Inductive
Evidence 1: Evidence 2: Evidence 3: Conclusion:
.Ioe
will recidivate,
loe, an unemployed spouse abuse4 recidivated. Harold, an unemployed spouse abuser, recidivated.
All
I ffi,
ffircreta;'
Srrse:
theory
is
supported, while
in
patterns results in a generalization about some social process. Inductive reasoning enters into deductive research when we find unexpected patterns in the data we have collected for testing a hypothesis. We may call these patterns serendipitous findings or anomalous findings. Whether we begin by doing inductive research or add an inductive element later, the result of the inductive process can be new insights and provocative questions. But the adequacy of an explanation formulated after the fact is necessarily less certain than an explanation presented prior to the collection of data. Every phenomenon can always be explained in some way. Inductive explanations are thus more trustworthy if they are tested subsequently with deductive research'
rEts.ppor6
\ \
took an inductive turn when Sherman and the other researchers began trying to make sense of the differing patterns in the data collected in the different cities. Could systematic differences in the samples or in the implementation of arrest policies explain the differing outcomes? Or was the problem an inadequacy in the theoretical basis of their research? Was deterrence theory really the best way to explain the patterns in the data they were collecting? As you learned in my review of the Pate and Hamilton (1992) study, the researchers had found that individuals who were married and employed were deterred from repeat offenses by arrest, but individuals who were unmarried and unemployed were actually more likely to commit repeat offenses if they were arrested. What could explain this empirical pattern? The researchers turned to control theory, which predicts that having a "stake in conformity"
fru
deterrene
freo,r1. sgaUol tt
fu !r!
This last indncft ricier picture of, t hcle come closr I ns that their
Emplo-'-ment
initi{
s&il[t
* mrurt- So ma1'be ml
llEnts: the-v also
rinlence. The rcd& is more costl]' to r n'oman. judges wil rBerk et al. 1993t
(resulting from inclusion in social networks at work or in the community) decreases a person's likelihood of committing crimes (Toby 1957). The implication is that people who are employed and married are more likely to be deterred by the threat of arrest than those without such stakes in conformity. And this is indeed what the data revealed. Now, the researchers had traversed the research circle almost three times, a process perhaps better described as a spiral (see Exhibit 3.12).The first two times, the researchers had traversed the research circle in a deductive, hypothesis-testing way. They started with theory and then deduced and tested hypotheses. The third time they were more inductive: They started with empirical generalizations from the data they had already obtained and then tumed to a new theory to account for the unexpected patterns in the data. At this point, they believed
Exploratory Rer
Qualitative rs observing said a explanation for*E
Chapter 3
of
Social
Research 77
h deterrence theory made correct predictiong; given certain conditions, and that ¬her ttry, control theory might specify what these conditions were.
This last inductive step in their research made for a more complex but also gdnceptually picture of the impact of arrest on domestic violence. The researchers seemed to hve come closer to understanding how to inhibit domestic violence. But they cautioned rs that their initial question-the research problem-was still not completely answered. Employment status and marital status do not solely measure the strength of social attach[cnts; they also are related to how much people earn and the social standing of victims in ourt. So maybe social ties are not really what make arrest an effective deterrent to domestic vftrlence. The real deterrent may be cost-benefit calculations ("If I have a higher income. jail b more costly to me") or perceptions about the abtions of authorities C'If I am a married woman, judges will treat my complaint more seriously"). Additional research was needed (Berk et al.1992).
rtrr
Exploratory Research
Qualitative research is often exploratory and, hence, inductive: The researchers begin by observing social interaction or interviewing social actors in depth and then developing an explanation for what has been found. The researchers often ask questions like "What is going
Uds eocse5i
rcELefu de!
ItcsrcA.
'lr
I:ri
XL-
lF-rd h fr
iprffi
fid
I Iihemfq
dmurrydrrc
tufl
dd?*r.!tfl
1. How do youths represent everyday conflict in their stories? 2. What decision-making and reasoning processes do youths produce in their
about conflict?
stories
&-ShYaqrn
rxrlrrdil rdsd rf
dmfrb
3. How do young people represent various means for handling peer conflict? 4. How is violence portrayed in their narratives? 5. How are adults situated in youth conflict narratives? (Monill et al. 2000:531)
Each of these examples illustrates how the research questions that serve as starting points for qualitative data analyses do not simply emerge from the setting studied but are shaped by the investigator. As Harry Wolcott (1995) explains,
jr*&lrilr
flc'+
focrfl- rG
H*nh"u rcd c{Ecs d
Chapter 3
lcseach questionl is not embedded within the lives of those whom we study,
ffi*. h
waiting to be discovered. Quite the opposite: We instigate the problems we invesThre is no point in simply sitting by, passively waiting to see what a setting is m'tell' us or hoping a problem will "emerge." (p. 156)
for guiding qualitative data ffi h6s on the importance of the research question as a tool The research question process. analytic the of nature iterative the obscure m rnootA not data collection and of processes the throughout multiply or expand, h-g", narro%
developed inductively from qualitative research can feel authentic because have tried to see the UlLne kard what people have to say "in their own words" and we will be richer and research qualitative from derived Explanations seeit." mrld..as they are likely to be but they research, quantitative in are often they fuely textured than people studied in this the that assume cannot We area. a limited from fL*", cases -ae to ours to similar explanations develop will researchers other that like others or a test of a set up initially do not we Because or heard. observed sse of what was and along come cannot researcher another rules, specific to some is according
fuaations
Research
learned in Chapter 1 that some social research is purely descriptive. Such research fu rct involve connecting theory and data, but it is still a part of the research circle-it t-E- with data and proceeds only to the stage of making empirical generalizations based on lhdata (refer to Exhibit 3.8,P.12). mlid description is important in its own right-in fact, it is a necessary component of all hsigations. Much important research for the government and public and private otganiza,f:r is primarily descriptive: How many poor people live in this community? Is the health of rrcdaeiy improving? How frequently do convicted criminals return to crime? Simply put, research process and an essential 6pd, description of data is the cornerstone of the scientific world. qor"ot for understanding the social Gooa aescriptive research can also stimulate more ambitious deductive and inductive rmch. The Minneapolis Domestic Violence Experiment was motivated, in part, by a growLg body of descriptive research indicating that spouse abuse is very common: 572,000 cases d-women victimiied by a violent paflner each year; 1.5 million women (and 500,000 men) qiring medical attention each year due to a domestic assault (Buzawa & Buzawa [g96:l-3). You may recall from Chapter 1 that much of the research on the Internet and social detions also has been descriptive; however, by identifying the prevalence of Internet use and rment levels of social isolation, this research has helped to establish priorities for both public
;. k
research.
I
(h
of
for social
research
is not complete until it includes an understanding for the protection of human subjects. Every scientific
rolcte-amqf
rthoHuPin
fdnepolicyc
aesr policies
hpdcdaq il
nigl
ame
ahsers
{Frafngwas l5licity is
od
&al
. Saial
difficuhb
pubticnY
ion; however, distinguishing between unintentional error and deliberate fraud can be very difficult. For example, a 1963 report of the U.S. Senate's Subcommittee on Problems of the Aged and Aging concluded that a study of elderly persons' health needs, publicized by the American Medical Association, was a "supposedly objective, scientific, academic study" that was really a "pseudo-scientific half-effort" (Cail 196'7 18-79). The researchers were accused of having an upper-class bias in the design of their sample and of using some questions that
underestimated elders'health needs. Yet the researchers were convinced that they had adhered to scientific guidelines. It is not ciear in this case, nor in many others, whether the research was designed to favor
ndfre
rmwebqrE'
and
ermflict vith
dvace |fiiilltr
particular findings or to adapt to unavoidable constraints. Error, committed without any intent io defraud, is inevitable. Social scientists who do not do their best to minimize error skirt the boundaries of fraud, but the discovery oferrors in a study should not, in itself, be taken as an indication of dishonestY. Openness about research procedures and results goes hand in hand with honesty in research design. Openness is also essential ifresearchers are to learn from the work ofothersIn spite ofthis need for openness, some researchers may hesitate to disclose their procedures or risults to prevent others from building on their ideas and taking some of the credit. You rnay have heard of the long legal battle between a U.S. researcher and a French researcher about how credit for discovering the AIDS virus should be allocated. Although such public disputes are unusual, concerns with priority of discovery are common. Scientists are like other people in their desire to be first. Enforcing standards ofhonesty and encouraging openness about research are the best solutions to these problems'
6il) low-incm
ftir
ir
ad
&e l95tls"
Respectfu
The
fue tramled
f.ederal Policy
smial scienoe
Chapter 3
Research
81
on partial information was preferable to waiting until the information was more Although we now know that the original finding of a deterrent effect of arrest did in each replication, Sherman (1992) later suggested that implementing mandatory plicies might have prevented some subsequent cases of spouse abuse (pp. 150-153). ; in the Omaha follow-up study, arrest warrants reduced repeat offenses among &tsers who had already left the scene when police arrived. Horrvever, this Omaha ras not publicized, so it could not be used to improve police policies. So how.much is warranted and at what point in the research should it occur? irl scientists who conduct research on behalf of specific organizations may face addidifficulties when the organization, instead of the researcher, controls the final report publicity it receives. If organizational leaders decide that particular research results the researcher's desire to have findings used appropriately and reported fully with contractual obligations. Researchers can often anticipate such dilemmas tn and resolve them when the contract for research is negotiated:or simply decline a research opportunity altogether. But often, such problems come up only after a has been drafted, or the problems are ignored by a researcher who needs to have ajob ds to maintain particular personal relationships. These possibilities cannot be avoided but because of them, it is always important to acknowledge the source of research in reports and to consider carefully the sources of funding for research reports
by
others.
hearch
on People
'r fumal procedures for the protection of human subjects in research grew out of some ,libly publicized abuses of human subjects. A defining event occurred in 1946, when the
Uhrmberg War Crime Trials exposed horrific medical experiments conducted by Nazi docand others in the name of "scienee." In another example, many Americans were shocked learn in the 1970s that researchers funded by the U.S. Public Health Service had infected lffi low-income African-American men with syphilis in the 1930s without their knowledge d then studied the "natural" course of the illness. Many participants.were not informed of frir illness and denied treatment antll 1972, even though a cure (pgriicillin) was developed
itk
1950s.
Egregious violations of human rights like these resulted, in the United States, in the crcation of a National Commission for the Protection of Human Subjects of Biomedical and Eehavioral Research. The Commission's 1979 "Belmont Report" (Department of Health, Education, and Welfare 1979) established three basic ethical principles:
t Beneficence: Minimizing possible harms and maximizing benefits; o .Iustice: Distributing benefits and risks of research fairly.
Respectfor persons: Treating persons as autonomous agents and protecting those with diminished autonomy;
The Department of Health and Human Services and the Food and Drug Administration Ihen ffanslated these principles into specific regulations that were adopted in 1991 as the Federal Policy for the Protection of Human Subjects. This policy has shaped the course of social science research ever since its inception, and you will have to take it into account as
rilq qf
tudl5mdtrdl
m lllrry{
mmrify."-f
io
Ficbm
to
Can
p*tfuil*fr
Eil
prirm;i
d
cil
by frEir
e:rlEr ilE-
Hr.
inlixmsd
fums
pic b
rmk tl-rsm
l993din
As simple as these guidelines may seem, they are difficult to interpret in specific cases and harder yet to define in a way agreeable to all social scientists. For example, how should "no harm to subjects" be interpreted? Does it mean that subjects should not be at all harmed psychologically as well as physically? That they should feel no anxiety or distress whatever during the study or only after their involvement ends? Should the possibiliry of any harm, no matter how remote, deter research? Consider the question of possible harm to the subjects of a well-known prison simulation study (Haney, Banks, & Zimbardo 1973). The study was designed to investigate the impact of social position on behavior-specifically, the impact of being either a guard or a prisoner in a prison, a "total institution." The researchers selected apparently stable and mature young male volunteers and asked them to sign a contract to work for 2 weeks as a guard or a prisoner in a simulated prison. Within the first two days after the prisoners were incarcerated by the "guards" in a makeshift basement prison, the prisoners began to be passive and disorganized, while the guards became verbally and physically aggressive. Five "prisoners" were soon released for depression, uncontrollable crying, fits of rage, and, in one case, a
re m soilE GIfiG, f &iefd $fi*Lt fr r:rnhs ma1" siII he'c Fil his sE dxl rLa+href-s rl9-Tlll i &oominacnYd h&oom and qid{ Lcrisitedthe hmd
rhrmphre)"s
Hlul
be conducted-a
fullf
ior-trmEil
;gnhmomatic rash; on the sixth day, the researchers terminated the experiment. Through ahllssions in special postexperiment encounter sessions, feelings of stress among the pa(ic-
[m who played the role of prisoner seemed to be relieved; follow-up during the next year ffiaed no lasting negative effects on the participants and some benefits in the form of Uffi insight. :. fruld you ban such experiments because of the potential for harm to subjects? Does the ftt Sat the experiment yielded significant insights into the effect of a situation on human
t&vior-insights
that could be used to improve prisons-make any difference (Reynolds you believe that this benefit outweighed the foreseeable risks? Do ttfl!L133-139)? of informed consent is also more difficult to define than it first appears. requirement The must be given by persons who are competent to consent, have coninformed, consent frte fully informed about the research, and have comprehended what they voluntarily, are ffd (Reynolds 1979). The researcher's actions and body language must help to told Lrrc been consent is voluntary. Children cannot legally give consent his verbal assurance that o[lEy they must have the opportunity to give or withhold their prticipate in research, but r participate to which their legal guardians have consented (Sieber in research r3ruz, to prisoners give informed consent? Can students who are asked to participate in il!II2)- Can participants professor? in covert experiments? These situations by their Can rrearch extra care. ryirc frrlly informed consent may also reduce participation in research and, because signing onsent forms prior to participation may change p-articipants' responses, produce biased cmlts (Larson 1993:114). Experimental researchers whose research design requires some _, qpc of subject deception try to get around this problem by withholding some informatiorf Hqe the experiment begins but then debriefing subjects at the end. In the debriefing, the
q:sions (Sieber 1992:394I). However, even though debriefing can be viewed as a substire, in Some cases, for securing fully informed consent prior to the experiment, if the
subjects disclose the nature of the experiment to other participants, subsequent rsrlts may still be contaminated (Adair, Dushenko, & Lindsay 1985). For his study of homosexual behavior in public bathr\ms, Tearoom Trade, Laud &mFhreys (1970) decided that truly informed consent would be impossible to obtain. hsilead, he first served as a lookout-a "watch queen"-for men who were entering a public Lrtroom in a city park with the intention of having sex. In a number of cases, he then left the ffioom and copied the license plate numbers of the cars driven by the men. One year later, b visited the homes of the men and interviewed them as part of a larger study of social issues. lfumphreys changed his appearance so that the men did not recognize him. His conclusions, rcmingly, were benign-the men engaged in what were viewed as deviant acts were, for the nost part, married, suburban men whose families were unaware of their sexual practices. But debate has continued ever since the publication of this research about Humphreys's failure to cll the men what he was really doing in the bathroom or why he had come to their homes for fu interview. If you were to serve on your university's IRB, would you allow this research to
be conducted?
&iefed
Well-intentioned researchers may also fail to foresee all the potential problems. In the pison simulation, all the participants signed consent forms, but how could they have heen fully informed in advance? The researchers themselves did not realize that the study
SOCIAL TESE
ftreouc
es
sgErdt srhE&ErIrutI
ed
=q
Maintaining confidentiality is another key ethical obligation; a statement should be included in the informed consent agreement about how each subject's privacy will be protected (Sieber 1992). Procedures such as locking records and creating special identifying codes must be created to minimize the risk of access by unauthorized persons. Howeveq statements about confidentiality should be realistic: Laws allow research records to be subpoenaed and may require reporting child abuse; a researcher may feel compelled to release information if a health- or life-threatening situation arises and participants need to be alerted. Also, the standard of confidentiality does not apply to observation in public places and information available in public records. The potential of withholding a beneficial treatment from some subjects also is cause for ethical concern. The Sherman and Berk experiment required the random assignment of subjects to treatment conditions and thus had the potential of causing harm to the victims of domestic violence whose batterers were not arrested. The justification for the study design, however, is quite persuasive: The researchers didn't know prior to the experiment which response to a domestic violence complaint would be most likely to deter future incidents (Sherman 1992).The experiment provided clear evidence about the value of arrest, so it can be argued that the benefits outweighed the risks. The extent to which ethical issues are a problem for researchers and their subjects varies dramatically with the type ofresearch design. Survey research, in particular, creates few ethical problems. In fact, researchers from Michigan's Institute for Survey Research interviewed a representative national sample of adults and found ihat 687o of those who had participated in a survey were somewhat or very interested in'participating in another; the more times respondents had been interviewed, the more willing they were to participate again. Presumably they would have felt differently if they had been treated unethically (Reynolds 1919:56-51). On the other hand, some experimental studies in the social sciences that have put people in uncomfortable or embarrassing situations have generated vociferous complaints and years of debate about ethics (Reynolds 1979; Sjoberg 1967). The evaluation of ethical issues in a research project should be based on a realistic assessment ofthe overall potential for harm and benefit to research subjects rather than an apparent inconsistency between any particular aspect of a research plan and a specific ethical guideline. For example, full disclosure of "what is really going on" in an experimental study is unnecessary if subjects are unlikely to be harmed. Nevertheless, researchers should make every effort to foresee all possible risks and to weigh the possible benefits of
repqmal
ism.ThE
hs&fl
lm promi*.
An
irrnrfurrqrd
I ,*{rerf,irsslgid
ryrmmmd
rEsemd
hbi
* Yor,'illdsotrdn
&bmss'utdefr
lf .vou rEsEiltL F rirale fcr fudft;l H1'm roPgsebilr t*ff- he sltr fit i irr Qm a-taing L5fl m he ctrrect rDasqs I
A reseach Pmpd pi|fi s4" of fie rtr
the research against these risks. They should consult with individuals with different perspectives to develop a realistic risk/benefit assessment, and they should try to maximize the benefits to, as well as minimize the risks for, subjects of the research (Sieber
1992:75-108). Ultimately, these decisions about ethical procedures are notjust up to you, as a researcher, to make. Your university's IRB sets the human subjects protection standards for your institution and may even require that you submit your research proposal to them for review.
Fiounaircmasd rifr a uonP of ffi m propmeA fr a;l s4 and6eproM Don't ueglecr F &c@ l-ourPmpml
Chapter 3
+ The
Process
I
l}
proposal graleful for those people or groups who require you to write a formal research you constructive rF*UO as that seems), and be even more grateful for those who give mr&".t. Whether your proposal is written for a professor, a thesis committee, an otganiza-
lf6 rcking
proposal practical advice, or a government agency that funds basic research, the ltru fuce you to set out a problem statement and a research plan. So even in circumstances for feedback' ft a proposal is not reqoired, you should prepare one and present it to others fu *riting your ideas down will help you to see how they can be improved, and almost any ffinfiack will help you to refine your plans' Proposal" exercises that will Each chapter in this book inciudes "Developing a Research an overview of the rmile you through the process of proposal writing. This section presents introduction to these special end-of-chapter ;*; of proposal w.lilng that also serves as an a wrap-up discussion of the entire proposalcontains text the in ;Eises. itre tast ct aptei
process. fllqrration - in=ry research proposal should have at least five sections. The following list is adapted ful-ocke, Spirduso, & Silverman (2000:8-34):
it is that An introductory statement of the research problem, in which you clarify what you are interested in studYing.
.
o
on what has A literature review,in which you explain how your problem and plans build already been reported in the literature on this topic' of A meihodological plan, detalling just how you will respond to the particular mix opportunities and constraints you face' how you will An ethics statement, identifying human subjects issues in the research and
respond to them in an ethical fashion' and presentA itatement of limitations, reviewing weaknesses of theqroposed research ing plans for minimiZing their consequences'
You
within
&
p.opori will be reviewed competitively, it must present a compelling research problem mionale for funding. Ii is not possible to oYerstate the importance of the
If your
t*
a hypothyou propose to study (see the first section of this chapter). If you propose to test focusto avoid want You alternatives' plausible are GdL be sure that it is one for which there ..boring hypothesis"-one that has no credible alternatives, even though it is likely fu on a m be correct (Dawes 1995:93). from a A research proposal also can be strengthened considerably by presenting results proposed the administering involved have might This ,tro1 study of ihe research question. a preliminary version of the proposed experiment conducting sample, a small to ryffiionnaire like ii& u group of students, or making observations over a limited period of time in a settingpilot the in used methods the of presentation &a proposed for a qualitative study. Careful ,strdi *d the problems that were encountered will impress anyone who reviews the proposal' to Ooo't n"gt""t procedures for the protection of human subjects. Even before you begin for requires IRB your university's procedure dn"elop youi p.oposal, you should find out what
WORLD
the review of student research proposals. Follow those procedures carefully, even if they require that you submit your proposal for an IRB review. No matter what your university's specific requirements a-re, if your research involves human subjects, you will need to include in your proposal a detailed statement that describes how you will adhere to these requirements.
You have learned in this chapter how to formulate a research question, review relevant literature, consider ethical issues, and identify some possible research limitations, so you are now ready to begin proposing new research. If you plan to do so, you can use the proposal exercises at the end of each of the subsequent chapters to incorporate more systematically the research elements discussed in those chapters. By the book's end, in Chapter 15, you will have attained a much firmer grasp of the various research components. At that point, we will retum to the process of proposal writing.
,Ifr
rffi
rN[-l G[ilr
E[
CONCLUSIONS
Selecting a worthy research question does not guarantee a worthwhile research project. The simplicity of the research circle presented in this chapter belies the complexity of the social research process. In the following chapters, I will focus on particular aspects of the research process. Chapter 4 examines the interrelated processes of conceptualization and measurement, arguably the most important part of research. Measurement validity is the foundation for the other two aspects of validity. Chapter 5 reviews the meaning of generalizability and the sampling strategies that help us to achieve this goal. Chapter 6 introduces causal validity and illustrates different methods for achieving it. Most of the remaining chapters then introduce different approaches to data collection--experiments; surveys, participant observation and intensive interviewing, evaluation research, comparative historical research, secondary data analysis and content analysis-that help us, in different ways, to achieve results that
are valid. Of course, our answers to research questions will never be complete or entirely certain. Thus, when we complete a research project, we should point out how the research could be extended and evaluate the confidence we have in our conclusions. Recall how the elaboration of knowledge about deterrence of domestic violence required sensitivity to research difficulties, careful weighing of the evidence, and identification of unanswered questions by several research teams. Ethical issues also should be considered when evaluating research proposals and completed research studies. As the preceding examples show, ethical issues in social research are no less complex than the other issues that researchers confront. And it is inexcusable to jump into research on people without any attention to ethical considerations. Owning a large social science toolkit is no guarantee of making the right decisions about which tools to use and how to use them in the investigation of particular research problems, but you are now forewarned about, and thus hopefully forearmed against, some of the problems that social scientists face in their work. I hope that you will return often to this chapter as you read the subsequent chapters, when you criticizethe research literature, and when you design your own research projects. To be conscientious, thoughtful, and responsible-this is the mandate of every social scientist. If you formulate a feasible research problem, ask the right questions in advance, try to adhere to the research guidelines, and steer clear ofthe most common difficulties, you will be well along the road to fulfilling this mandate.
Chapter 3
a The
Process
TCY TE RMS
ilmalous
findings
Ettriveresearch Esrdentvariable
haion
of association
generalization
edrical Ective
Iffpttesis
fideFadent variable
research
IIICHLIGHTS
r, .
Research questions should be feasible (within the time and resources available), socially imporand scientifically relevant..
r Building social theory is a major objective of social science research. Investigate relevant theories }bre starting social research projects, and draw out the theoretical implications of research frndings. - o The type of reasoning in most research can be described as primarily deductive or inductive.
ltmarch
based on deductive reasoning proceeds from general ideas, deduces specific expectations from
be ideas, and then tests the ideas with empirical data. Research based on inductive reasoning begins nilt specific data and then develops general ideas or theories to explain patterns in the data. . It mly be possible to explain unanticipated research findings after the fact, but such explanations
hrrc less credibility than those that have been tested with data collected for the purpose of the study.
o The scientific process can be represented as circular, with a path from theory to hypotheses, to lre, and then to empirical generalizations. Research investigations mq begin at different points along t research circle and traverse different portions of it. Deductiv" ."r"-bh begins at the'piint of theory
Lfuctive research begins with data but ends with theory, and descriptive research begins with data and
its generalizability in other situations. An research stemming from a particular research question should include a series of orvlies that, collectively, traverse the research circle multiple times.
in
C-ontemporary ethical standards also require that social research cause no harm to subjects; that particintion be voluntary, as expressed in informed consent; that researchers fully disclose their identity; that bcnefits to subjects outweigh any foreseeable risks; and that anonymity or confidentiality be maintained fu participants unless it is voluntarily and explicitly waived.
DrscussroN QUEsrroNs
Find a research journal article that is cited in another source. Compare the cited source to what was said about it in the original article. Was the discussion of it accurate? How well did the authors of both articles summarize their work in their abstracts? What important points would you have missed
you had relied only on the abstracts?
l.
if
2. Classify five
in previous exercises or in
other
courses, as primarily inductive or deductive. Did you notice any inductive components in the primarily deductive projects? How much descriptive research was involved? Did the findings have any