Conducting Semi-Structured Interviews

Abstract

Conducted conversationally with one respondent at a time, the semi-structured interview (SSI) employs a blend of closed- and open-ended questions, often accompanied by follow-up why or how questions. About one hour is considered a reasonable maximum length for SSIs in order to minimize fatigue for both interviewer and respondent. This chapter begins with a discussion on the disadvantages and advantages of SSIs. Despite the disadvantages and costs of SSIs, they offer some extraordinary benefits as well. Semi-structured interviews are superbly suited for a number of valuable tasks, particularly when more than a few of the open-ended questions require follow-up queries. The chapter presents some recommendations that can be considered when constructing an SSI guide. All in all, effectively conducted semi-structured interviews, even though labor intensive, should be worth the effort in terms of the insights and information gained. © 2015 by Kathryn E. Newcomer and Harry P. Hatry, and Joseph S. Wholey.

Full text

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
CHAPTER NINETEEN
CONDUCTING SEMI-STRUCTURED
INTERVIEWS
William C. Adams
How do we ask people for information? At one extreme is using a bat-
tery of identical, mostly closed-ended questions. These highly structured
surveys, typically with large samples, can be administered many ways (phone,
mail, Internet, in person; see Chapters Fourteen and Eighteen). At the oppo-
site extreme is the uid inquiry of focus groups (see Chapter Twenty). Com-
pared to surveys, a focus group engages far fewer people (an optimum of ten to
twelve per session) for a much longer period (up to two hours) with an elastic
agenda of open-ended questions that allow extended probing. Making up in
depth what they lack in breadth, focus groups enable the moderator not only
to pursue detailed inquiry into existing opinions but also to obtain reactions
to new ideas and conduct group brainstorming, if desired.
Another approach falls between standardized, mostly closed-ended surveys
of individuals and free form, open-ended sessions with groups. This intermedi-
ate method pulls elements from both but puts them into a distinctive package.
Curiously, this methodology does not have a consensus name. Lewis Dexter
(1970) called it elite interviewing, although that label may erroneously suggest
talking only with high-status respondents. Robert Merton (1956) termed it the
focused interview, although that phrase now risks confusion with focus group. Cul-
tural anthropologists speak more narrowly of the ethnographic interview. Sociol-
ogists sometimes refer to depth interviewing. Due to the approach’s many open-
ended questions, the term qualitative interviewing may also be used. However,
492

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 493
the name that appears to be currently garnering a majority of usage is not crisp,
clever, or inventive but it is simple and descriptive: the semi-structured interview.
That is the term used in the chapter. Let’s call it SSI for short.
Conducted conversationally with one respondent at a time, the SSI
employs a blend of closed- and open-ended questions, often accompanied by
follow-up why or how questions. The dialogue can meander around the topics
on the agenda—rather than adhering slavishly to verbatim questions as in a
standardized survey—and may delve into totally unforeseen issues. Relaxed,
engaging, in-person SSIs can be longer than telephone surveys, although they
seldom last as long as focus groups. About one hour is considered a reasonable
maximum length for SSIs in order to minimize fatigue for both interviewer and
respondent.
Disadvantages and Advantages of SSIs
Before going into more detail about semi-structured interviews, let’s rst con-
sider their disadvantages, in case you may then decide to skip the rest of this
chapter. SSIs are time-consuming, labor intensive, and require interviewer
sophistication. Interviewers need to be smart, sensitive, poised, and nimble,
as well as knowledgeable about the relevant substantive issues. The process of
preparing for the interviews, setting up the interviews, conducting the inter-
views, and analyzing the interviews is not nearly as quick and easy as you might
think. The time and effort required to do all of it right is considerable. SSIs
usually entail the arduous task of analyzing a huge volume of notes and some-
times many hours of transcripts.
Another drawback—unless you are just interviewing members of a small
group (such as the board of a nonprot organization or top program
administrators)—is that, without an enormous outlay of time and personnel,
SSIs are unlikely to encompass a large enough sample to yield precision of the
“plus or minus npercent” variety. Consequently, for many purposes, a stan-
dardized survey of six hundred clients would be superior to attempting six
hundred one-hour SSIs. For some other purposes, four focus groups with ten
people each would be much more efcient than conducting forty individual
SSIs. Yet, despite the disadvantages and costs of SSIs, they offer some extraor-
dinary benets as well.
Semi-structured interviews are superbly suited for a number of valu-
able tasks, particularly when more than a few of the open-ended questions
require follow-up queries. Especially consider employing SSIs in the following
situations:

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
494 Handbook of Practical Program Evaluation
rIf you need to ask probing, open-ended questions and want to know the
independent thoughts of each individual in a group
rIf you need to ask probing, open-ended questions on topics that your
respondents might not be candid about if sitting with peers in a focus group
rIf you need to conduct a formative program evaluation and want one-on-one
interviews with key program managers, staff, and front-line service providers
rIf you are examining uncharted territory with unknown but potential
momentous issues and your interviewers need maximum latitude to spot
useful leads and pursue them
In mixed methods research, SSIs can be useful as an adjunct to supplement
and add depth to other approaches. For example:
rIf you need to conduct some in-depth reconnaissance before designing a
large-scale survey, conguring a focus group agenda, or constructing an
overall research strategy
rIf, after drafting a standardized survey questionnaire, you discover that
important questions cannot be effectively addressed without more open-
ended questions and extended probing
rIf you want to explore “puzzles” that emerge (or remain) after you have
analyzed survey or even focus group ndings
The people who may be appropriate for SSIs can run the gamut of those
involved in the program being evaluated. For convenience, let’s put them into
three general groups:
1. Program recipients (or beneciaries, clients, customers, members, con-
stituents, or audience—preferred term will vary)
2. Interested parties (contributors, suppliers, any other stakeholders who are
neither direct recipients nor program administrators, plus others in prox-
imity who may be affected in collateral ways)
3. Administration(front-line service delivery people, other staff, top managers,
program board members, whether salaried or volunteer)
If one or more of the SSI situations listed previously applies to one or more
of these three general SSI-appropriate groups—and if you have adroit and well-
spoken interviewers available who can be adequately educated on the program
at hand—then semi-structured interviews would be the methodology of choice.

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 495
Designing and Conducting SSIs
Assuming that this methodology is your choice, let’s proceed to consider prac-
tical steps for designing and conducting semi-structured interviews: select-
ing and recruiting the respondents, drafting the questions and interview
guide, techniques for this type of interviewing, and analyzing the information
gathered.
Selecting Respondents and Arranging Interviews
Chapter Seventeen offers detailed advice on preparing for site visits, making
staff assignments, training eld teams, and carrying out other practical admin-
istration steps for collecting data in the eld. However, a few basic elements
important to SSIs should be mentioned here. Having identied at the outset
the target group or groups for SSIs, how do researchers then select respon-
dents from among the target group? If the group is a large one, researchers
ordinarily choose to interview a manageable random sample or a stratied ran-
dom sample (as dened in Chapter Fourteen). If the group is not so large and
resources permit, it may be possible to interview virtually everyone, such as all
key administrators and all program board members. Even if time and resources
do not allow conducting a large numbers of SSIs, it is still important to get the
perspectives of more than just a few people.
Respondents ought to have been identied and appointments set up
before interviewers arrive at the site. If staff members of an organization
are being interviewed, ordinarily, top managers will assist in setting up the
interviews, greatly simplifying the process. If a sample is being drawn from
a roster (stratied perhaps to include set numbers of managers, supervisors,
and various categories of staff members), the evaluators convey the names
of the chosen individuals to the managers, rather than letting the managers
personally pick which staff members are heard. Sometimes, researchers must
telephone those chosen to request and schedule each individual appointment.
Rather than making a “cold call,” researchers should send a short letter of
introduction in advance, noting the importance of the individual’s advice
and citing the project’s endorsement by the top administrator. This can add
legitimacy and save time that would otherwise have to be spent explaining
and justifying the research. That advance letter can pave the way for the
subsequent phone call to arrange the meeting.
Approaches vary when interviewing program beneciaries (rather
than program workers), depending on who the beneciaries are and their
relationship to the program. If possible, they should be chosen randomly

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
496 Handbook of Practical Program Evaluation
(incorporating stratication when certain subgroups are targeted) to elimi-
nate the biasing effect of convenience samples. Sometimes the program staff
is best positioned to set up the interviews, or at least make the introductions.
Other times researchers may obtain a list from which to sample and contact
the potential respondents.
Prospective respondents will probably want to know how much of their pre-
cious time you covet, and that can be the trickiest single issue in obtaining the
interview. Proposing too long a period can prompt an outright refusal. Con-
versely, if an unrealistically short period is requested, respondents may depart
after the allotted time, even if key agenda items are far from nished; the inter-
viewer also risks appearing to have been deceptive or foolish or both. So here
are a few ideas to consider. Pretesting the interview should yield a rough idea
of how long the questions will take. You can initially mention that time (“It
shouldn’t take much longer than . . ..”). Then, if things are going well but
slowly during the actual interview, ask permission for “just a few more ques-
tions” to nish the core questions and perhaps cover some of the second tier
of topics. A late afternoon session may have the advantage of not running up
against another meeting. Regardless of the time and place, the most important
element—aside from respondents’ actually consenting to be interviewed—is
the content of those interviews. The development of appropriate and well-
crafted interview guides is essential.
Drafting Questions and the Interview Guide
Questionnaire is not the best term for the compilation of SSI questions, because
that word connotes a xed instrument to be read verbatim, rather than the
exible, interactive approach of SSI questions. Instead, you must create the
agenda for the interview guide, the outline of planned topics, and questions to
be addressed, arrayed in their tentative order. Of course, if SSIs are going to be
conducted with different groups, a guide will have to be tailored to each group.
Consider the following recommendations when constructing an SSI guide:
1. At the outset, be sure to budget enough time to carefully draft, edit, pretest,
and polish the interview questions and guide, allowing time for several
iterations and feedback from colleagues. If possible, pilot tests with a few
intended respondents (or people similar to them) can be the nal step in
rening the guide.
2. Don’t try to cram too many issues into the agenda, but if the list of potential
topics is long, decide in advance which ones are critical and which ones are

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 497
optional. Once the top priorities are clear, put those questions in bold. Clas-
sify the second and third tiers of questions to be raised if time allows, but
decide in advance which questions are vital and which ones are lagniappes.
In theory, SSIs can be somewhat lengthy, but that does not mean that busy
respondents are going to want to dedicate hours to talk. As part of rigorously
editing the draft questions, be sure to omit questions asking for simple facts
that can be retrieved from an organization’s website, published material, or
available records (unless perhaps you want to assess someone’s understand-
ing of those facts).
3. Don’t forget that closed-ended questions can be ideal gateways to open-
ended probing. For example, after asking, “In your judgment, was this
program change a major improvement, minor improvement, or not an
improvement?” the interviewer could follow up by asking, “Why is that?”
or “Why do you feel that way?” and continue with additional probing as
needed. Compared to using a broad start (such as, “What did you think
about this program change?”), the beauty of incorporating a closed-ended
query rst is that it dramatically streamlines the summary analysis to have
some rm quantitative points of reference (for example, “Ten of the twelve
board members called the change a ‘major improvement’ and cited these
reasons. . . .”).
4. When potential respondents do not speak English, careful translations and
bilingual interviewers will be required. Even if the respondents do speak
English, don’t assume it’s exactly the same language spoken by university-
trained researchers. Communication can fail when we blithely assume that
everybody shares our vocabulary, acronyms, and lingo. In designing an SSI,
use the everyday words of the target groups, while taking care not to talk
down to them as well. It may be useful to make adjustments to question
wording after the rst round of interviewing.
5. Think through the extent to which the draft questions may evoke pressure
to give socially acceptable answers. Might recipients worry that their eligi-
bility could be jeopardized by what they say? If so, along with assurances of
condentiality, look for ways to remove any stigma that might attach to cer-
tain answers. One tactic for showing nonjudgmental acceptance is to insert
a prefatory comment such as “some people tell us [a particular opinion about
an issue],” before asking, “How do you see this issue?” This suggests that
answers like the one cited would not surprise or disturb the interviewer.
Administrators themselves may feel an urge, conscious or otherwise, to cir-
cle the wagons and put the best possible face on a less than ideal program
situation. Rather than asking them to identify what is “bad,” ask about “areas
that need improvement” to help minimize defensiveness.

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
498 Handbook of Practical Program Evaluation
6. The agenda for a semi-structured interview is never carved in stone. If a
conversation unexpectedly turns from the rst to the fourth topic, by all
means, reorder the topics on the y and return later to pick up the ones
that were skipped. Nevertheless, when drafting the tentative question order,
try to anticipate the most likely and smoothest sequence. These time-tested
guidelines can help the agenda fall into place:
rAfter customary pleasantries, as the actual interview begins, start with
a few extra easy, even throwaway questions to start a comfortable chat
before the more serious inquires begin. To break the ice, respondents
might be asked how long they have lived in the city or worked in the
program, for example.
rAfter establishing some rapport, turn next to more directly relevant but
still nonthreatening questions.
rWhen introducing critique questions, consider putting positive inquiries
rst: “What are the good things about X?” or “What do you like about
it?” Starting with positives allows those people who might be reluctant
to voice criticisms to share their complaints later because they already
offered some praise. Another advantage of this approach is that some
people, once they adopt a harsh critical tone, nd it difcult to say any-
thing good, as if they fear they would be contradicting themselves or
minimizing the seriousness of their grievance.
rAfter looking at the positive sides of a topic, turn next to its drawbacks,
disadvantages, disappointments, or areas that need improvement, always
taking a neutral, nonjudgmental tone. (There will be more information
about tone and delivery later in this chapter.)
rThe most potentially embarrassing, controversial, or awkward questions
should come toward the end. By this point the interviewer is no longer a
stranger but a pleasant, nonargumentative professional who seems gen-
uinely interested in the respondent’s opinions. Now respondents are
wishing that their kids, spouse, or co-workers showed as much respect
and attentiveness to their thoughtful opinions. So at this stage, more
sensitive questions can be introduced, along with reminders of conden-
tiality as needed.
rWhy are demographic questions best saved until the end? For many peo-
ple, few topics are more delicate or possibly painful than their current
marital status, their age, their amount of education, their income, and
so forth. These questions raise fundamental identity issues in a way that
mere opinion questions do not. Revealing one’s personal prole, even
in condence, can still feel like a privacy intrusion. Be sure to scruti-
nize the typical laundry list of demographic questions to omit all that

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 499
are not essential for this specic program evaluation. Refusals can be
minimized by asking about the broadest usable categories, instead of
exact amounts (for example, asking respondents to select from among
large income ranges rather than prying for precise annual income). A
few demographic variables, such as gender, often race, and perhaps gen-
eral age group, can be coded by the interviewer without needing to ask.
rTo end on a substantive note, consider concluding by returning to a
short, easy, program-related question, perhaps about the future.
Once developed, the interview guide, no matter how extensive its prepara-
tion, should still be considered a work in progress. It remains subject to change
for this reason: in the eld, as feedback quickly begins to accumulate, adjust-
ments will need to be made (Galletta, 2013). Perhaps the sequence of ques-
tions will have to be rethought, the way certain issues are posed will have to be
recast, and some unanticipated issues will emerge that seem sufciently impor-
tant that they should be added to all subsequent interviews. Agile researchers
will exploit these new insights to rapidly rene the interview guide and will
actually have planned for this possibility at three stages—after the rst inter-
view, after the rst round of interviews, and periodically thereafter.
After the rst interview, reassess everything. What works well, and what
needs to be modied? Some questions and topics may need to be added or
subtracted, expanded or condensed, recast or reordered. If more than one
interviewer is working on the project, have the most experienced and knowl-
edgeable person conduct the rst interview; then he or she can brief the oth-
ers and take the lead in rening the questions and agenda. After everyone on
the team has conducted an interview, schedule another opportunity to share
tips and experiences as well as to modify the questions as needed. Thereafter,
periodic, even daily if feasible, sessions allow the team members to continue
to review their work to identify any areas for which adjustments should be
made and to ensure that their individual approaches are not diverging too
much.
Decisions on modifying the interview guide in the eld are necessary judg-
ment calls. If, for example, one respondent volunteers a surprising and trou-
blesome problem that seems potentially quite threatening to the effective exe-
cution of the program under study, should subsequent respondents be asked
pointedly about that issue if they fail to volunteer it? Adding it to the guide
might be the safest strategy. If feedback from the next rounds of interviews
consistently dismisses that particular concern, it might then be taken back off
the guide and the initial response treated as an outlier. Ongoing reassessments
of the interview guide, particularly during the early waves of interviewing, let

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
500 Handbook of Practical Program Evaluation
researchers catch areas for improvement early, avoiding the need to repeat
numerous interviews.
During the course of their interviews, some respondents (program admin-
istrators, for example) may refer to certain documents that the interviewers
may not yet have. Interviewers should keep a list of any such documents that
are of interest (and not subject to privacy restrictions) so these items can be col-
lected immediately after the interview if possible or, if not, then sent by e-mail
or mail. The post-interview thank-you note can serve as a convenient vehicle
for a reminder.
Starting the Interview
Interviewers should establish a positive rst impression. They should dress
professionally—men with ties even on Fridays—although slightly more casual
attire is preferable when the respondents are economically disadvantaged.
They should always arrive early and review the agenda, having allowed time
for potential travel delays and getting lost, rather than risk arriving even one
minute late. Thanking the respondent for the meeting and offering a business
card can add to professional credibility.
To record or not to record, that is a key judgment call. A small digital
recorder, if permission is granted, allows the interviewer to be more actively
engaged in the conversation as well as to ponder the best next question
instead of having to concentrate on writing down answers. However, if the top-
ics covered are at all sensitive, respondents may be inhibited by a recording
device, even if complete condentiality if promised and consent is given; some
people forget the recorder is running, but others stay wary. If the machine
option does seem acceptable, bring a small, unobtrusive digital recorder (or
a microcassette tape recorder) that has been veried as working properly
and is ready to use, so no fumbling is required. After gaining permission,
switch it on and say something like this, “OK, [respondent name], thanks for
letting me record this,” to document the conrmation of consent. Be fully
prepared, in case the respondent declines, to nonchalantly switch to taking
notes.
If the interview is not recorded electronically, alternatives for taking notes
include booklets, legal pads, electronic tablets, laptop computers, or smaller
notebook computers. The interview guide can be produced in a number of
layouts: bound as a booklet (with enough blank space between items for writ-
ing out the answers), put into a condensed format (perhaps even one page, so
you can see everything at a glance, with answers recorded separately), or even
printed on ve-by-eight-inch index cards that can be shufed as needs dictate

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 501
in the conversation. In any layout, it can be helpful to use a large font, with pri-
ority questions in boldface or color coded. One additional strategy should be
considered if resources permit and a recorder is not used: a two-person team.
Using a division of labor, one person can take the lead in asking questions and
ensuring that all key topics are addressed, while the other person, who does
not have to be mum, concentrates on taking notes.
When taking notes, interviewers should do the best they can with their own
shorthand systems. They should use quotation marks when writing verbatim
phrases. Many notes will be paraphrases and should be treated as such when
writing the report, but it is important to write down word-for-word and put
quotation marks around any particularly valuable or memorable comments.
Put any interviewer observations (for example, of respondent laughter, ner-
vousness, or anger) in square brackets. If needed, it is ne, even attering, for
interviewers to ask respondents to pause for a moment in order to nish writ-
ing down an important comment. Immediately after the interview, interviewers
should be able to rush to a computer to clarify and expand their scribbles and,
while the chat is fresh, add any other key remarks that they recall but did not
write down at the time.
At the start of the interview, the matter of condentiality must be addressed
clearly. If the respondent is going to be quoted by name in the report, that must
be explained and consent obtained. However, unless there is some compelling
reason to do otherwise, condential interviews are generally better because
they are more likely to elicit candid answers. If that route is taken, it is worth
explaining and emphasizing condentiality to the respondent in the initial
invitation and at the start of the interview. For example:
We’re trying to learn how the XYZ program operates and get your
suggestions on how it can be improved. We’re not auditors and we’re not
employees of any government agency. We’re independent. Nothing you say
today will be quoted with your name. We’ll be submitting a report, but no
condential names will be mentioned. Do you have any questions about
our pledge of condentiality?
If later, when writing the report, some observation stands out as so brilliant
that you really ought to give credit to that person, contact him or her to ask
permission. Likewise, if a valuable comment obviously must have been voiced
by a particular person, he or she might be asked, “May I quote you on that by
name?” during the interview, or the evaluator might go back later to ask per-
mission, so as not to violate the condentiality promise. The evaluation should
also establish procedures for handling and storing the information collected,

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
502 Handbook of Practical Program Evaluation
especially if it is condential. Procedures may range from not entering respon-
dents’ names into any project databases to randomly generating identication
numbers maintained in encrypted computer les.
Polishing Interview Techniques
Preparation is vital. Interviewers should know the questions thoroughly, under-
stand the purpose of each question and the overall priority level of each ques-
tion in the overall research scheme. When asking questions, tone is extremely
important. SSI interviewers should take a casual, conversational approach
that is pleasant, neutral, and professional, neither overly cold nor overly
familiar. In this relaxed, comfortable setting, probing is accomplished with-
out the interviewer sounding astonished by anything said, interested but not
shocked.
Not all SSI practitioners agree about the exact persona that the interviewer
should adapt. Some advise interviewers to “appear slightly dim and agreeable”
(McCracken, 1988, p. 38), so that respondents won’t feel threatened. Even
though being sensitive to people’s feelings is certainly a good idea, too much
playing dumb might make respondents decide they are wasting their time and
should not bother getting into complex discussions with such a clueless inter-
viewer. The opposite extreme, a superior, all-knowing stance, seems likely to
be off-putting and counterproductive as well. Probably the best balance is to
appear generally knowledgeable, in a humble, open-minded way, but not to
pose as more expert than the respondent (Leech, 2002). In that vein, inter-
viewers should not debate with or contradict a respondent; interviewers must
be sure they understand his or her views. They should maintain that calm, non-
reactive demeanor, even when faced with a respondent whose personality or
comments are offensive.
From time to time, it can be constructive to restate concisely in one or
two sentences, using mainly the respondent’s own words, what was just said.
This technique of active listening reinforces that the interviewer is indeed
intently interested and can ensure that the interviewer does, in fact, under-
stand a point. After any unrecognized lingo or acronyms, interviewers should
not apologize, but just repeat the mystery word in a questioning tone.
Prompting respondents to elaborate can be done in many ways besides
just asking, “Why is that?” “Could you expand on that?” or “Anything else?”
Sometimes a simple “yes?” with a pause, repeating a key word, or even nodding
in silence is sufcient to signal that the interviewer would like to hear more. A
tilted head can also be a green light for more detail. Avoid asking, “What do

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 503
you mean?” because that accuses the respondent of failing to communicate.
Asking for an elaboration or for an example is better. However, if something
is unclear, do not hesitate to obtain a clarication. In the give-and-take of the
conversation, take care not to interrupt and be overly controlling. Even when
respondents drift into irrelevant territory, wait until they nish before making
a soft segue back to a priority topic.
Sometimes, even if the recorder is running, it may be helpful to jot down
an occasional note or star something on the agenda to return to later. Inter-
viewers should be sufciently familiar with the general questions that they do
not have to bury their heads in the text to read each question verbatim. A
glance down at the text should be sufcient. Because the actual conversation
may diverge from the original order on the agenda, the interviewer must be
exible and ready with subtle improvisations to weave back to other issues.
Entirely unexpected but promising avenues of interest may open up, and, if
so, interviewers should feel empowered to pursue them. Such developments
should be promptly shared with colleagues to determine whether those new
topics should be explicitly added to the agenda or if the agenda order should
be revised. After conducting numerous interviews, interviewers may sometimes
nd that answers begin to seem predictable. They have to remain alert, how-
ever, for different nuances in the answers that may be worth exploring or at
least noting in the report.
Near the end of each session, nothing is wrong with asking for a moment to
review the agenda guide to ensure that no key questions were missed. If time is
running out, the interviewer will have to make a quick decision about whether
to omit some of the remaining questions (and which ones), to ask to extend the
visit a bit longer, or to request a short follow-up meeting at a later date. At the
conclusion of the interview, the interviewer should thank the respondent cor-
dially and condently (not apologetically) for helpful comments. Before the
day is out, the interviewer should send a short thank-you e-mail or, to be more
traditional, a handwritten note; this extra expression of appreciation makes
a difference in how respondents remember the experience and the people
involved.
Other important tasks should also be completed daily. Interview notes
should be cleaned and claried so they will make sense to other members of
the research team (and to the original interviewer a few weeks later). If notes
were handwritten, they should be entered into a computer right away, and
even if a small computer was used to take raw notes, these notes still have to
be reviewed and edited while fresh. Maintain a master list of any abbreviations
used in the interview summaries. Even if the session was recorded, some addi-
tional documentation (date, time, site codes, and so forth) should be led. To

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
504 Handbook of Practical Program Evaluation
save time, some interviewers prefer to dictate their elaboration of their hand-
written notes; the dictation then must be transcribed later.
Analyzing and Reporting SSIs
Once the SSIs have been completed, the next step is to explore the results.
As something of a hybrid between a standardized survey and a focus group,
semi-structured interviews are suitable for a report that is a hybrid, too. If,
as recommended, a few closed-ended questions were employed, there will be
some hard numbers to cite, maybe suitable for a few simple tables and graphs.
Percentages can be brought to life by the follow-up responses to open-ended
questions. Look for ways to consolidate themes found in multiple answers and
to supplement them with well-chosen, illustrative quotations. This aspect of
the SSI report resembles a focus group report. (For a published example of a
report of focus group ndings, see Adams, 2005, Chapter 8.) For example, if
roughly three out of four program administrators complained they were bur-
dened by mountains of paperwork, quote some comments to make this issue
vivid and explain it in more detail. Ordinarily, omit the highly unrepresenta-
tive outliers, unless for some reason a particular comment, even if rare, should
be conveyed to decision makers.
The time involved and method chosen for analyzing the open-ended ques-
tions will depend heavily on the number of people interviewed and the num-
ber of topics addressed. Summarizing a dozen SSIs of top managers will not
be onerous, but systematically assessing SSIs of several hundred program ben-
eciaries will be a challenge. For more advice about the analysis phase, see
Chapter Twenty-Two, which explains in detail the techniques that can be used
in coding open-ended answers, along with the software programs that assist
in the process. A judicious appraisal of the ndings should yield a depth of
understanding about the issues at hand beyond that possible from the alterna-
tive survey techniques alone. All in all, effectively conducted semi-structured
interviews, even though labor intensive, should be worth the effort in terms of
the insights and information gained.
References
Adams, W. C. Election Night News and Voter Turnout: Solving the Projection Puzzle. Boulder, CO:
Lynn Rienner, 2005.

JWBT1565-c19 JWBT1565-Newcomer Printer: Courier Westford June 5, 2015 20:56 Trim: 7in ×9.25in
Conducting Semi-Structured Interviews 505
Dexter, L. A. Elite and Specialized Interviewing. Evanston, IL: Northwestern University Press,
1970.
Galletta,A.M.Mastering the Semi-Structured Interview and Beyond. New York: NYU Press, 2013.
Leech, B. L. “Asking Questions: Techniques for Semistructured Interviews.” PS: Political
Science & Politics, 2002, 35, 665–668.
McCracken, G. The Long Interview. Thousand Oaks, CA: Sage, 1988.
Merton, R. K. The Focused Interview. New York: The Free Press, 1956.