SPSS

Survey Tips
A handy guide to help you
save time and money as
you plan, develop, and
execute your surveys

www.spss.com
Table of Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1
Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
Define your mission . . . . . . . . . . . . . . . . . . .3
Outline your research . . . . . . . . . . . . . . . . .3
Establish a budget . . . . . . . . . . . . . . . . . . . .4
Develop a schedule . . . . . . . . . . . . . . . . . . .4
Define the population . . . . . . . . . . . . . . . . .4
Determine your sample size . . . . . . . . . . . .5
Select a sampling technique . . . . . . . . . . . .7
Choose a survey method . . . . . . . . . . . . . . .9
Questionnaire design . . . . . . . . . . . . . . . . . .12
Determine your survey’s length . . . . . . . . .12
Choose appropriate question formats . . .14
Design your response choices . . . . . . . . . .18
Use professional formatting . . . . . . . . . . .21
Conduct a pilot study . . . . . . . . . . . . . . . .24
Finding problems in your survey . . . . . . . .25
Data collection . . . . . . . . . . . . . . . . . . . . . . .26
Use incentives as appropriate . . . . . . . . . .26
Give clear instructions . . . . . . . . . . . . . . . .28
Interviewer-administered surveys . . . . . . .28
Telephone surveys . . . . . . . . . . . . . . . . . . .29
Mailed surveys . . . . . . . . . . . . . . . . . . . . .32
Online surveys . . . . . . . . . . . . . . . . . . . . . .34
Survey field times . . . . . . . . . . . . . . . . . . .34
“Ideal” response rate . . . . . . . . . . . . . . . .34
Analysis and reporting . . . . . . . . . . . . . . . . .35
Use a unique identification variable . . . . .35
Clean and verify your data . . . . . . . . . . . . .36
Analyze open-ended questions . . . . . . . . .37
Keep track of your analyses . . . . . . . . . . . .37
Design readable reports . . . . . . . . . . . . . .42
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . .43
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . .44
About SPSS Inc. . . . . . . . . . . . . . . . . . . . . . . .51
SPSS products for survey
and market research . . . . . . . . . . . . . . . . .52

SPSS products named in this booklet are trademarks or registered

trademarks of SPSS Inc. For details, see pp. 52–56. Other names
are trademarks of their respective owners.
Preface
Are you currently involved in survey research?
Or are you developing a survey for the first
time? Regardless of your level of experience,
SPSS Survey Tips can help you plan, develop,
and execute surveys.

This booklet is divided into four major sections,

each related to a stage in the survey research
process and containing a number of tips to guide
you through it. Please remember that these
stages are not to be considered in isolation. A
decision made at one stage may influence your
work at other stages. Also, in some situations
you may work on several stages simultaneously,
rather than sequentially.

After the tips, you’ll find a glossary of terms

frequently used in survey research. These
terms are boldfaced the first time they appear
in the text.

As you read, you’ll see symbols that will help

you better understand the information in this
booklet.

This symbol indicates an example

illustrating a particular tip.

The § symbol indicates that the figure shown is

based on a survey of a very large population,
such as a national survey.

1
 This symbol highlights SPSS products
that help you perform a particular action
or specific survey application more
effectively and productively. A list of
SPSS survey research products can be
found on pages 52—56 of this booklet.

Keep SPSS Survey Tips by your side during the

survey process and use it to help you save
money, execute your survey in a timely manner,
and get the highest response rate and the most
useful results.

If you have questions about conducting or

analyzing your surveys, call your local SPSS
office. We offer a variety of training and
consulting programs to assist you. Information
is available online at www.spss.com.

Or, if you have suggestions for improving

our survey software, please e-mail us at
[email protected].

2
Stage one: Planning
Define your mission
Do you know why you and your organization
want to conduct this survey? If so, use these
reasons to develop your survey’s mission. Once
developed, refer to your mission throughout
the survey process to make sure that every
action taken supports it.

If you are uncertain about why you and your

organization want to conduct the survey, and
what you plan to do with the results, you may
be collecting unnecessary data.

When questioning your organization, if

you get an answer such as, “We do it
every year,” dig deeper for the real reason.

Outline your research

Develop a plan to implement your mission. Be
prepared to know which statistics to run and
what initial relationships and patterns you
expect to find.

If you want to predict purchase patterns

based on known demographic
characteristics, you need to specify the
demographics you think may be relevant.

3
Establish a budget
Mapping an action plan helps you justify your
study and budget. It also helps determine the
scope and size of your survey. Typically, the
major costs of a survey are data collection
and data entry. Some methods, such as e-mail
or online surveys, typically have a lower cost-
per-returned-questionnaire and may help you
stay within your budget.

Before beginning, be sure to ask how

much your survey will cost and how
long it will take to complete it and
deliver results.

Develop a schedule
If you’re new to survey research, assume that
the process will take longer than you expect.
When creating an entirely new survey, allow
time for review and performing a pilot study.
With any survey, be sure to allow enough time
for data collection, analysis, and reporting.

Try to anticipate what tables and graphics

should be constructed, so you can begin
these early in the process.

Define the population

Before you can develop questions and format
your survey, you need to consider the population
you will be contacting. The age, attention span,
and gender of potential respondents will
influence the design of your survey and the
data collection methods you use.

4
 A young child may be able to pay attention
to your questions for as little as five
minutes, or may not understand all of
the content. Older respondents may
not understand new phrases or jargon.
One age group may prefer phone or
mail surveys, another online surveys.

Consumers reached by phone at home

may be distracted by other activities and
unwilling to spend more than 10 minutes
being surveyed.

If any of your questions are gender-

specific, be careful to avoid
unintentionally offensive language,
and be sure that questions are directed
at the right audience.

Determine your sample size

The appropriate sample size for your survey is
influenced by your purpose in conducting the
survey, and by your budget. If your sample size
is too small, you could miss important research
findings. But if it’s too large, you could waste
valuable time and resources.

There are software tools that enable you

to determine critical variables (anticipated
effect size, confidence, statistical power,
and sample size) before you spend valuable
time and resources.

5
 There is usually no reason to survey
more than 1,000 to 1,500§ respondents.
While the precision of results tends to
improve as the sample size increases,
the increase in precision is negligible
when sample size is greater than
1,500§ respondents.

SamplePower

One approach
A conservative formula you can use to determine
the appropriate sample size N is based on the
amount of error you are willing to tolerate,
stated as a proportion or percent.

N =_____
1
error2

For example, if an error of five percent (±5

percent) is acceptable, the formula calculates
the required sample size as:

N =_____
1 = _____
1 = 400
.052 .0025

A second approach
If you have reason to expect a high degree of
correlation between variables, then you can use
a smaller sample size for your survey. Base the
sample size on the minimum adequate sample
size of important subgroups in the population.

6
Many analysts suggest that there should be at
least 100§ cases in each subgroup. If possible,
know what proportion of the population (in the
real world) is in each subgroup and calculate
the total sample size required. Take into account
the expected non-response rate of this population
and increase the sample size by that factor.

You can use a crosstabulation or some

nonparametric tests; in addition, there are
procedures, such as exact tests, that work
very well with small subgroups.

SPSS Base, SPSS Exact Tests

Select a sampling technique

A sample is a part of an entire population
that possesses attitudes, opinions, habits, or
characteristics that you wish to study. A census
includes an entire population. Consider the size
of the population you’re sampling, and then
decide whether you want to survey a sample
or a census. If you elect to survey a sample,
you might use one of the following techniques.

Purposive sampling
Use non-probabilistic or purposive sampling,
if you want to learn about people with special
characteristics. Purposive sampling is often
used for focus groups.

7
 You want to survey females between the
ages of 18 and 25 who watch television
at least 10 hours a week. You’ll want to
conduct a purposive sample, to be
sure that you include all of the relevant
characteristics and exclude individuals
who don’t fit these requirements.

Using every “nth” name

Systematic sampling, the easiest random
sampling method, generates a multiple of a
number: for example, every 5th, 10th, or 34th
name from a population list. Beware of hidden
patterns in your list that could compromise the
integrity of the sample, however. The list may
include names sorted by frequency or recency
of contact, geographic origin, or similar
company size.

Using more than one sampling method

You may use more than one sampling method
for your survey.

You may conduct a simple random

sample but purposely over-sample one
stratum—households with a certain
level of income, for example. With
these additional data, you can perform
two studies: one on the whole population
and the other on a specific subset. This
type of sample is called a stratified sample.

SamplePower

8
 You can take equal numbers of various
subgroups—former, current, and new
customers, for example—to maximize
statistical power in tests.

Choose a survey method

There are two basic methods for conducting
a survey: self-administered and interviewer-
administered. Self-administered surveys can
be written—a paper questionnaire sent by
mail, for example—or conducted electronically,
either over the Web or through dedicated
survey stations. Self-administered surveys
can also be oral—as when participants phone
in responses to an automated system. Interviewer-
administered surveys can be conducted in
person or over the phone, with interviewers
recording results on paper or electronically.

If questions are personal or require

a lot of thought, self-administered
surveys are a good choice.

Allocate resources for data entry

Whatever method you choose, be sure to budget
time and resources for data entry. With online
surveys, your data will already be in an electronic
format. Data will also be recorded if you use
computer-assisted personal interviewing (CAPI)
or computer-assisted telephone interviewing
(CATI) solutions. Designing paper surveys so that
they can be scanned electronically makes data
entry for this type of survey faster and less costly.

SPSS Data Entry,

mrInterview,
mrPaper, mrScan

9
Consider the length of your survey
If you must ask a lot of questions, or expect the
respondent to spend an hour or more with your
survey, an interviewer-administered in-person
survey is recommended. If you can’t afford to
perform in-person surveys, the second-best
option is a self-administered survey.

A lengthy mail survey is more likely

to be completed than a 20-minute
telephone survey. A lengthy online
survey is more likely to be completed
if respondents can stop and start at will.

Consider your budget

Interviews conducted by phone can be more
expensive than other surveys when you consider
long-distance phone charges, wages for
interviewers, and supervisory costs (in-person
interviews also incur many of these charges).

Mail surveys may be economical if you have

a large sample or if your sample spans a large
geographical area. Remember to consider the
method of returning the mail survey. For example,
including postage on return envelopes increases
expenses but also can increase returns.

10
Online surveys are another cost-effective
method. You save on postage, printing, and
wages for interviewers—and you minimize the
costs of data entry and data cleansing, since
information is already captured in an electronic
format. Depending upon the number of surveys
your organization conducts, you may find that
a survey developed and/or hosted by an
application service provider (ASP) is an
economical alternative.

SPSS Data Entry,

mrInterview,
mrPaper, mrScan,
Dimensions ASP,
Dimensions Service Bureau

Consider speed
Web and telephone surveys are the fastest
methods of conducting surveys. The typical
time required for collecting data through a Web
or telephone survey ranges from a few days to
a few weeks, while mailing a survey can add at
least a month to the process.

Consider your survey population

Before you decide upon any form of self-
administered survey, you need to consider
whether the method is appropriate for the
population you are surveying.

11
 If you’re thinking of conducting a
Web-based survey, make sure you’re
able to obtain a valid sample population
with that method. For example, some
populations—such as members of certain
professional organizations or students
at many universities—have access to the
Internet. However, you can’t assume that
everyone does, or that they are willing
to complete a survey online.

In order to get accurate results with

Web-based surveys, be aware that with
“volunteer” surveys—a popular feature
on many Web sites—results may be
skewed. For example, if you are trying
to learn how many hours per day people
surf the Web, heavy Web users will be
more likely to respond than the average
person. You can minimize this risk by
using screening questions to control
the number of participants in various
subgroups, based on demographic or
other qualities or on their responses
to questions.

Stage two: Questionnaire design

Determine your survey’s length
The ideal length depends upon the topic and the
type of population you need to study. In general,
limit the length of your questionnaire to encourage
prospective respondents to participate.

12
 Here are some general guidelines:
■ A survey targeting children should

take no longer than 5 to 7 minutes

to complete
■ For most populations, self-administered

surveys should be no longer than

four pages
■ Web-based surveys should ideally

enable participants to answer each

question in a single step
■ Phone interviews should run no

longer than 10 minutes

■ Face-to-face interviews can continue

for an hour or longer

Balance length and information

With a shorter questionnaire, you should get a
higher response rate and reduce the chance of
error and missing data. However, you will also
get less information from your respondents and
may have a less comprehensive study. The key
to a successful survey is to ensure that your
questions are concise and easy to understand
and give you valid, reliable information.

Keep questions short

Make each question easy to understand and,
if possible, less than 25 words in length.
Avoid using “double negatives,” as this type
of phrasing often confuses respondents.

13
 Which of the following questions do you
find clearer?
“I agree that snack foods such as
candy bars should be available in
school vending machines.”

“I don’t agree that snack foods such

as candy bars should not be
available in school vending machines.”

Choose appropriate question formats

Some questions can be easily answered with a
single answer, but others may require multiple
choices—a scale or, perhaps, a grid. And some
may best be answered through an open-ended
text response. Choosing an appropriate format
will make it easier for respondents to answer
questions clearly.

SPSS Data Entry,

mrInterview, mrPaper, mrScan

Avoid “double-barreled” questions

Double-barreled questions ask for opinions about
two subjects at the same time. For example,
“Are you satisfied with the amount and kind
of information you receive from your benefits
administrator?” When analyzing responses to
such a question, you won’t be able to tell which
part of the question the respondent is answering.

14
Avoid “leading” and “loaded” questions
Leading questions indicate—subtly or not so
subtly—the desired or acceptable answer. For
this reason, they discredit the objectivity of
your results.

Loaded questions are also to be avoided. These

questions use emotionally charged words like
“crisis,” “failure,” or “superb,” and tend to
elicit more strongly emotional responses than
questions phrased in emotionally neutral terms.

Be specific
If questions include vague qualifiers like “few,”
“many,” or “usually,” or undefined qualifiers like
“good” and “bad,” the meaning of respondents’
answers may be difficult to quantify.

Consider how responses might vary to

the following questions:
“Do you think the president is doing a
good job of handling foreign policy?”

“Do you think the president is doing

a good job of handling the current
foreign policy crisis?”

“Please rate how the president handles

foreign policy.”

The third version has the most balanced

phrasing and avoids both loaded terms and
vague qualifiers.

15
Don’t lead with responses
In interviewer-administered surveys, avoid
beginning questions with a phrase like “Do
you very often, frequently, seldom, or never...”
Since people pay attention to what they hear
first, respondents listening to a question with
this structure will focus on the choices, and
not the question.

Organize questions in logical groups

This provides a sense of order for respondents
and makes it easier for them to recall experiences
and express their views.

In online surveys, group similar questions on

a single screen or use grids to minimize
complexity and encourage survey completion.

Avoid antagonizing respondents

Respondents are more likely to respond if they
feel their answer is socially acceptable. Give
respondents a way out of answering questions
to which they might feel they may not have the
“right” answer.

The question, “Did you vote in the last

election?” is more likely to alienate a
non-voter than the following phrasing:
“There are many reasons why people
might not vote in a given election.
Sometimes they are ill, or very busy,
or have to take care of an emergency.
Thinking back to the last election, did
you happen to vote?”

16
Offer respondents a choice of languages
In certain situations, your survey can be conducted
in a single language. It’s increasingly common,
however, to create and field surveys in multiple
languages. You’ll want to provide questions,
response lists, and instructions in the languages
preferred by respondents while simplifying
data analysis as much as possible.

mrTranslate

Consider using open-ended questions

Most survey questions are closed-ended,
meaning that you provide response choices
for participants. Open-ended questions allow
people to express themselves in their own
words. Open-ended questions allow you to
explore a greater breadth of respondent
attitudes and preferences. There are two
types of open-ended questions. With one
type, there is a predetermined set of answers
that you expect to receive. With the second,
there is a wider range of potential answers.

A question might be, “Which breakfast

cereals have you eaten in the past
month?” In this example, you already
know the breakfast cereals on the
market; yet you have not supplied a
list for the respondent.

17
 The question, “Are there additional
features you’d like to see in our product?”
will generate comments that are not so
quickly classified as a simple list might
be. The benefit to this approach is that
it may uncover preferences or views
your organization might not have
considered suggesting. In this example,
it might lead to your organization
developing features that you may not
have imagined customers wanting.

SPSS Text Analysis for Surveys

Place open-ended questions at the end

If your survey is divided into sections, place
open-ended questions at the end of a section.
This gives you greater flexibility and more room
to record verbatim responses. Be careful not to
overuse open-ended questions, as they can
take more time for respondents to answer than
questions in other formats. This may cause
some respondents to quit before completing
your survey.

Design your response choices

If you provide a scale or choice of answers for
the respondent, it’s important that the provided
answers accurately reflect the respondent’s
intended response.

Likert scales—should you offer a middle choice?

The Likert scale is a ranked list of responses,
often five or seven, ranging from one pole to
an opposite pole.

18
Many researchers include a middle response
option in a scale. The middle answer offers a
comfortable response for subjects who have
legitimately divided or neutral opinions.

1 2 3 4 5
Strongly Agree Neither Disagree Strongly
Agree Agree nor Disagree
Disagree

Alternatively, if you prefer to force respondents

to make a choice, you can use a four-point
scale that offers no middle choice.

1 2 3 4
Very Satisfied Dissatisfied Very
Satisfied Dissatisfied

Note that the items above are not interval-scaled.

Interval-scaled means that the distance
between “strongly agree” and “agree” is the
same as the distance between “agree” and
“neither agree nor disagree.” Treating ordinal
variables as if they were interval-scaled can
lead to biased statistical results by skewing
the weighting of each response. However,
there are statistical packages that can correctly
do statistical analysis of such items.

mrInterview,
SPSS Categories

19
Offering “Don’t know” as a response option
Should you offer “Don’t know” as a response
option? By including it, you provide a response
choice for respondents who genuinely don’t
know. However, including this alternative enables
some respondents to avoid stating an opinion.

Research has shown that using “Don’t know”

doesn’t affect the relative proportion of other
responses. You might, however, find patterns
in the groups of people that responded “Don’t
know” to specific questions.

Statistical software can help you find and

understand patterns in the respondents who
answer “Don’t know,” so that you can determine
if they share similar characteristics.

SPSS Base, SPSS Missing Value Analysis,

SPSS Classification Trees

Offering “Couldn’t say” or “Not applicable”

Depending on the question, you could offer
respondents the option not to answer, if you
prefer that the opinions they express be
grounded in experience. Then the respondent
can tell you, “I couldn’t say” or “Not applicable,”
rather than “I don’t know.” By giving respondents
the opportunity to give a more precise answer,
you will gain more information about them.

20
Statistical software can help you find and
understand patterns in the groups of people
that answer “I couldn’t say” or “Not applicable,”
so that you can determine if they share similar
characteristics.

SPSS Base, SPSS Missing Value Analysis,

SPSS Classification Trees

Use professional formatting

You don’t have to start from scratch when
developing your survey. You can borrow styles
from large, well-known surveys such as the
General Social Survey developed by the United
States’ NORC (National Opinion Research Center).
Look into survey research books that provide
examples for business, academic, or government
environments, or for your particular industry.
Using pre-existing questions not only saves
time, it is also likely to improve the effectiveness
of your survey.

If the Gallup organization groups

income categories in a certain way,
those breaks may be good for your
survey, too.

SPSS Data Entry,

mrInterview

Distinguish questions from answers

Make it as easy as possible for your respondents
to fill out self-administered surveys.

21
 For example, you can format questions
in upper- and lower-case text and
responses in all-capital letters and
boldface type.

Did you purchase a car in the last year?

(Please circle one choice)
YES NO

Use checkboxes rather than blanks

Make it obvious where the respondent is
supposed to make marks so it’s easy for
respondents to complete the survey. Instead
of leaving a blank space, use:
■ Boxes ❑

■ Parentheses ( )

■ Brackets [ ]

■ Circles ❍

Leave enough space for comments

If you ask respondents to make comments, be
sure you provide enough room for them to
write them.

At the end of your survey, you may want

to ask respondents for general comments
and thank them for their assistance.

Don’t use lines

If you use open-ended questions, don’t
supply lines on which respondents can write
their response. The lines limit the amount of
feedback you’ll receive and don’t allow for
different sizes of handwriting. Instead, leave
plenty of white space.

22
Make your survey look attractive
The presentation of a survey can either
encourage or deter a person from responding.

Make a paper survey inviting by:

■ Including plenty of white space, so that the

document does not appear intimidating

■ Including color, if possible—but no more

than two colors

■ Using no more than two typefaces

With an online survey, you can include elements

that cannot be included in a paper survey. These
include animations, video, and sound files that
make an online survey vibrant and engaging.
Such elements enable you to pose questions
in different ways—simulating a shopping expe-
rience, for example—and can elicit different
types of responses than other types of surveys.

SPSS Data Entry,

mrPaper, mrScan,
mrInterview

Make your survey look professional

Spending extra time and money to produce
a professional document shows that you take
your research seriously. This increases the
likelihood that your respondents will take the
same attitude. If your paper survey is several
pages long, print it as a booklet. If you staple
several pages together, you risk losing pages
that get separated.

23
If you are creating an online survey, use images,
sound, and animation, where appropriate, to
make your survey engaging. If it is more than
just a few items long, you can improve the
response rate if respondents can take the survey
in stages, without having to re-enter answers.

Convey a consistent image by following corporate

style guidelines, whether you’re creating surveys
for your own organization or for your clients. To
save time, develop templates that can be
applied to multiple surveys, or use a service
that can do this for you.

mrPaper, mrScan,
mrInterview,
Dimensions ASP,
Dimensions Service Bureau

Conduct a pilot study

Pretesting, or conducting a pilot study, is like
a dress rehearsal for your survey. Although it
takes time, conducting a pilot study saves time
in the long run because it helps you identify
potential problems with your survey’s design
while there is still time to fix them.

Conduct at least two pilot studies for all new

surveys. With the first, focus on correcting
problems with the questionnaire, openly asking
for help and comments. Then conduct the
second as if it were the “real” survey, looking
for additional problem areas.

24
 If you plan a mail survey, it’s a good
idea to mail the surveys to a test sample
to gauge response time and other factors.

How large a sample should you use for

a pilot study?
Usually, no more than 75§ respondents are
needed for a pilot study. The pilot study
population should be similar in characteristics,
though not in size, to the population of the
planned survey.

Finding problems in your survey

The following tactics will help you find flaws
in your survey.
■ Read questions aloud or to someone else

and observe how they sound. If the subject

or sentence is complicated, split the question
in two or rephrase it.
■ Look for overuse of conjunctions such as

“and,” “or,” and “but” that create compound

sentences. Eliminate compound sentences
when possible.
■ Minimize the use of common prepositions

such as “with,” “except,” and “by” as

phrase connectors. These also can lead
to complicated sentences.

Ask respondents to write in comments and

additional responses. This may suggest
response options you never previously
considered. You can then add these to your
response lists, minimizing the number of
write-in or “other” responses.

25
Common survey problems
Look for the following common problems
in surveys:
■ Interviewers have problems reading

questions or recording answers

■ Many questions are left unanswered

■ The layout of the survey is cluttered and

confusing
■ Instructions are confusing

■ Completing the survey takes longer than

anticipated
■ There are unintentionally repeated or

redundant questions
■ There is too little space for responses to

open-ended questions

Other problems to watch for

When analyzing the responses from a pilot
study, you may find the following problems:
■ Little or no variance among responses

■ Too many “don’t know” responses

■ Too many “other” responses

■ Unclear skip patterns or branching

■ Misinterpretation of open-ended questions

If you find these problems, consider the wording

of a question, and its relevance to the topic.

Stage three: Data collection

Use incentives as appropriate
The most appropriate incentive depends upon
your population. In general, monetary rewards
are often an effective incentive. By studying
previous survey efforts and the effect of
incentives, you can get a clearer picture of what
kinds of incentives motivate the respondents
you’re hoping to reach.

26
 SPSS Classification Trees,
mrInterview

Examples of tangible incentives

Incentives that may be appropriate for your
survey population include:
■ Cash

■ Gift certificates

■ Chances to win prizes

■ Promotional or premium items, such as

key chains or calendars

■ A nine or ten-year-old might

participate in an hour-long study
in exchange for a video game or
movie pass.
■ A businessperson might participate
in exchange for a chance to win a
useful business tool, such as a
laptop computer or PDA.
■ Academics and some business
audiences are motivated by the
promise to share survey results.

Examples of intangible incentives

Some respondents will react more favorably
to intangible incentives.

“Help our organization better understand

this learning disability for the benefit of
other sufferers,” is an example of an
intangible incentive. So is the following:
“By telling us how you feel about our
products, you will help us better under-
stand consumers’ needs, which will
lead to the development of better,
more useful products.”

27
Give clear instructions
Begin your survey by briefly explaining its purpose.
Let respondents know how the data will be
used, and that their privacy will be protected.
In online surveys, you can also let respondents
see how far they’ve progressed, which encourages
them to complete the survey.

To minimize confusion and errors, tell the

respondent at the beginning of a written self-
administered survey what instrument (pen or
pencil) to use. At the beginning of each section,
give instructions on how to respond to the
questions. Also, don’t forget to tell the respondent
how to return the form when it’s completed.

Provide respondents with examples

On printed surveys, show respondents:
■ How to fill in their address

■ How to shade in a circle

■ How many circles or boxes to mark

Special considerations for interviewer-

administered surveys
With phone or in-person surveys, respondents
can’t reread items on a page, so it’s particularly
important that both questions and response
options are as clear as possible.
■ Keep questions short—no more than

25 words
■ Limit the number of response options—

no more than five

■ Limit the number of items to rank—no

more than three

28
Use good interviewers
A good interviewer is essential to a successful
telephone or in-person survey. A good interviewer:
■ Understands respondents and is empathetic

■ Listens well

■ Maintains an attitude of genuine interest

■ Is articulate

■ Can be objective

■ Accepts rejection—doesn’t take “no” personally

■ Doesn’t let mood affect performance

Special considerations for telephone surveys

For telephone interviews, script the entire call,
from the greeting through the closing. Make
the script clear and direct. You want to make
sure that the interviewer can easily follow the
script, because simple blunders in speech can
affect the confidence respondents have in your
survey and, therefore, affect the overall results.

mrInterview CATI

Break long phone surveys into sections

You need to keep respondents interested and
alert during a long stretch of questions in an
interviewer-administered survey. Breaking your
survey into sections will help accomplish this.
Even if there is not a logical subject change,
find a way to make a break.

29
Prompt the interviewer
There are tactics the interviewer can use to
help the respondent answer the questions. In
the interviewer’s script, tell him or her to:
■ Mention response alternatives within the

question (but not at the beginning)

■ Repeat a portion of the question for items

in a series so that respondents don’t forget

what is being asked

Give your respondents a chance to remember

Respondents may have a hard time remembering
details about past behavior. Give them time to
reflect if you are conducting a phone or in-person
survey. Help them fix a time period in their
minds by referring to events or specific dates
rather than “in the past five years.” Ask them
to check personal records.

Help the interviewer

During an interview-administered survey,
interviewers have a lot to do: They must talk,
listen, and record responses. To make their
job easier and receive more accurate responses:
■ Clearly differentiate questions and responses

from instructions (anything NOT to be read

to the respondent)
■ Arrange the questionnaire’s check-boxes or

other response options so that it’s easy for

interviewers to record data
■ For in-person paper-based surveys, provide a

list of common answers to questions so that

interviewers don’t have to write out the same
response multiple times

30
■ Conduct a practice session so you can listen
to the interviewer and provide feedback
■ Coach the interviewer on how to answer
common questions that may arise

Use numbers to record answers when

respondents rate something on a scale,
so that the interviewers don’t have to
write lengthy answers.

Anticipate unavailability of respondents

Give the interviewer a long list of names to contact,
in case certain individuals aren’t available or
are unwilling to participate. To be safe, you
should have between seven and 15 times
more names than the number of completed
surveys you want.

You can often increase the response

rate by giving respondents a choice of
ways to respond—in person, by mail, or
online, for example. To make data
analysis simpler, use survey research
tools that allow you to store data collected
in multiple modes in a single database.

mrInterview,
mrPaper, mrScan

Reduce the barriers to interview response

Lack of respondent availability and lack of
cooperation are the main barriers to response
for telephone and in-person interviews. To
compensate for lack of availability, interviewers
must keep contacting prospective respondents.
To help the interviewer, keep the introduction

31
concise. Also, at the beginning, mention any
incentives being offered to survey respondents.
If the person does not want to participate,
don’t press them; say “Thank you,” and move
on to another individual.

Special considerations for self-administered

mailed surveys
To increase the response rate, write a straight-
forward cover letter, no longer than one page,
explaining:
■ Who is sponsoring the survey

■ What is the purpose of the survey

■ The seriousness/importance of the survey

■ How the respondent’s opinions will be used

■ How their privacy will be protected

■ How the respondent benefits from participating

If possible, have an important or influential

person sign the letter to give the survey
greater legitimacy.

Certified mail can increase response rate

Although it increases costs, sending a survey
by certified mail can greatly increase the response
to a lengthy questionnaire.

Express mail and e-mail improve response

and timeliness
Both express mail and e-mail can be effective
when sending a survey to a business. These
methods improve both the response rate and
the timeliness of the survey’s return.

32
Notify respondents in advance
Send respondents an e-mail or postcard to alert
them that the survey will arrive in approximately
one week. Pre-notification will increase the
likelihood of response because the respondents
are more likely to recognize the survey when it
arrives. The note may also spark curiosity in the
respondents, and they will look for the survey.

Use commitment cards or e-mails to

estimate response
Commitment cards and commitment e-mails
ask prospective respondents to indicate that
they agree to participate in the survey. This
will allow you to estimate the response rate
to your survey.

Reminders increase the response rate

Send a reminder by postcard or e-mail one or
two weeks after the initial survey mailing.
These reminders should thank people who
have completed the survey while reminding
those who have not yet responded to do so.
Research shows this greatly increases a survey’s
response rate.

Re-mail surveys to increase response

Send out duplicate surveys to non-respon-
dents (or to the entire sample) 10 to 14 days
after the thank you/reminder postcard.

Telephone follow-up also improves response

Telephone follow-up also can increase the
response rate, when those making the calls are
properly trained. The personal contact tends to
emphasize the importance of the survey and
encourages people to participate. This does
increase survey costs, however.
33
Special considerations for online surveys
When designing surveys for fielding online,
minimize the number of screens respondents
must click through and the amount of scrolling
they must do to answer a question. But do not
try to fit all your questions on a single screen,
as this can be intimidating. Graphics and
images increase the attractiveness of online
surveys. Also, allowing respondents to partially
complete a survey, stop, and finish it at a later
time is a convenience that helps increase
response rates.

Survey field times

A survey conducted online may be completed
in a day—even in a few hours. Those conducted
in person or by phone may take a few days.
With mailed surveys, it may take weeks to
receive all responses. You can begin processing
and analyzing data even before all responses
have been received. Doing so can help you
identify questions that may be unclear to
respondents and also gauge whether you are
reaching the right subgroups within your sample.

mrInterview,
Dimensions ASP,
Dimensions Service Bureau

What is an “ideal” response rate?

There is no ideal response rate for surveys.
Many factors affect the response rate to a
survey, including:
■ Subject matter

■ Method of administration

■ Presence of incentives/perceived rewards

34
■ Level of difficulty of the survey
■ Aesthetic appeal of the survey
■ Perceived cost of completion

What does non-response imply?

Non-response, by itself, should not be a cause
for concern. But if non-respondents differ from
respondents in relevant ways, this may introduce
error into your results. Software tools can help
you analyze missing data patterns and account
for non-response variables. If you find that
respondents are systematically different from
non-respondents:
■ Weight your results so that the sample

matches known population values

■ Draw conclusions more carefully due to

missing responses

SPSS Missing Value Analysis

Stage four: Analysis and reporting

Use a unique identification variable
Analysis delivers the value from your survey
data. There are several steps in the process,
and several ways to evaluate your data. Begin
by placing a unique identification number on
each survey or record returned. This unique
number, sometimes called a “case ID,” will
help you track down problems in data cleaning
as well as flag cases of particular interest
during analysis.

35
Clean and verify your data
If you’ve collected data using electronic
methods, your data may already be clean
and ready for analysis. When collecting data
entered manually, however, allow time to
perform data verification and cleaning. As
the analyst, you should have an idea of how
your file should look. Software is available
that can streamline the data cleaning process.

SPSS Base, SPSS Data Entry,

SPSS Tables, SPSS Missing Value Analysis

When cleaning data, you’ll want to make

sure that the answers to these questions
are consistent with your experience:
■ How many people should be

managers, how many staff

■ The average number of years of

education of the sample

■ The proportion of missing cases you

should have for a question that did

not pertain to everyone

Run a series of crosstabulations before doing

further analysis to look for:
■ Inconsistent relationships (such as someone

saying she is a female, but whose relationship

to the head of household is “son”)
■ Unexpected averages

■ A large number of missing values

36
Analyze open-ended questions
Developing a coding scheme for open-ended
questions can be time-consuming and may
require data entry personnel who have a good
understanding of the subject matter. However,
there are techniques available that make reliable
analysis of open-ended survey responses fast
and efficient. With these tools, it’s easier to turn
unstructured text responses into information you
can analyze along with other quantitative data.

SPSS Text Analysis for Surveys

Keep track of your analyses

When performing complex analyses, keep a
record of the procedures you perform or the
ways you create new variables. This record will
help you reconstruct your analyses, if necessary,
as you write your final report.

SPSS Base

What are your data like?

Different statistical procedures are appropriate
for different variables, depending on what you
want to learn and the level of measurement of
the variable.

SPSS Base, SPSS Categories

37
Using categorical variables
Categorical or nominal variables provide a list
of choices with no meaningful order to the list.

Examples include gender, hair color,

and the type of organization that a
person works for.

An arithmetic mean of a categorical variable is

meaningless. Instead, use the mode and run
frequencies and crosstabulations using cate-
gorical variables. To display this type of data,
use pie charts or bar charts.

SPSS Tables, SPSS Classification Trees,

mrTables

Using ordinal variables

Ordinal variables have an implied order
between the response choices.

When asking opinions about an issue, a

code of 1 (strongly agree) means more
agreement than a code of 2 (somewhat
agree), but how much more is unknown.

Examine the median and mode for these variables

and run crosstabulations. Bar charts display the
choices well.

SPSS Base, SPSS Categories,

SPSS Advanced Models
mrTables

38
Using interval or continuous variables
Interval or continuous variables have an implied
order and an implied distance between the
response options.

With a variable such as age in years,

a one-unit difference is the same
throughout the distribution.

Interval or continuous variables lend themselves

to a much broader range of statistics than do
nominal or ordinal variables.

Use continuous variables where appropriate;

they will give you more information. If necessary,
you can always collapse a continuous variable
into a categorical variable.

Regression
Regression is one of the more popular statistical
procedures using interval-level variables.
Scatterplots and histograms are appropriate
graphical displays for these kinds of variables.

SPSS Base, SPSS Advanced Models,

SPSS Regression Models,
SPSS Complex Samples, Amos

Crosstabulation
Performing a crosstabulation is appropriate
when you have two or more categorical variables.
(Continuous variables don’t lend themselves to
crosstabulation, since you would get as many
rows or columns as there are different responses.)

39
When looking at crosstabulations, if the
probability of a chi-square is .05 or less, it is
usually small enough for the analyst to feel that
the distribution did not result from chance.

The chi-square statistic does not measure the

strength of the relationship; instead, it measures
whether a relationship is likely due to chance.
The smaller the probability of the chi-square
statistic, the more comfortable you can be that
the patterns you see are “real.” If your chi-square
statistic is significant, you must then follow up
with a column proportion test to indicate the
reasons for differences that you see in the table.

SPSS Tables, SPSS Complex Samples,

mrTables

Factor analysis
In survey research, factor analysis can be
helpful. It can be used to show the underlying
structure of a large number of variables, to
simplify the discussion of the data, or even
to suggest new, combined variables for use
in other analytic procedures.

SPSS Base

40
Differences in the mean
Use a t test to learn about differences in
means between two groups.

Are men who receive a certain surgical

procedure more likely to be younger than
women receiving the same procedure?
A t test can help you analyze data to
answer questions like this.

You can determine the average age for each

group, but you need a procedure like a t test
to confirm if the observed difference is due to
chance, or if it can be considered “real.” If the
significance is less than .05, you will usually
conclude that the differences in the observed
averages are not due to chance, and that they
reflect real population differences.

When presenting the results of a t test, use a

bar chart in which the height of each bar is the
average score for each group. Error bar charts
show both the group means and the precision
with which the mean was estimated (often the
95 percent confidence band). If you have more
than two groups that you would like to compare,
use the ANOVA procedure instead of a t test.

SPSS Base, SPSS Complex Samples

41
Design readable reports
Keep your audience in mind when reporting
survey results. Organize your report logically,
write clearly, and avoid jargon that may be
confusing of unfamiliar to your readers. You
may want to create report templates, not only
to save time but also so that the look of your
reports supports your organization’s brand, or
that of your clients.

mrInterview

Use graphs and tables to communicate results

It’s well known that using graphs and charts
helps people understand data more easily. So
show your results in appropriate visual formats.
What’s appropriate depends upon the kinds of
questions you’ve asked in your survey and the
nature of the data you’ve collected.

Number graphs and tables so that your audience

can find them easily. Also, be sure to label
axes and other elements. You may find that by
adjusting scales or by dividing overall results
into subgroups that reveal interesting differences,
you can give your graphs greater impact and
communicate results more clearly.

SPSS Base, SPSS Tables, SPSS Maps,

mrTables

42
Deliver timely reports
You conducted your survey because people
wanted information. Deliver results to them
faster and they’ll be able to put that information
to use sooner. To save time, you might want
to use templates to standardize reports or
presentations. You can use software that
enables you to provide analysts or clients with
online access to real-time results, or even
develop customized applications to deliver
information wherever it’s needed.

SPSS WebApp Framework,

SmartViewer Web Server,
mrInterview, mrStudio

Conclusion
This booklet has briefly touched on a number
of the things you’ll want to bear in mind as
you plan and conduct survey research projects.
Some related topics, such as predictive
modeling, are beyond the scope of this booklet.
If you want to explore this, or study any of
the topics covered here in greater detail, we
recommend you seek out a college-level
textbook on marketing and survey research
tools and practices. Or attend an SPSS training
course in the application of our products to
survey research. Information on these courses
can be found at www.spss.com/training.

43
Glossary
Analysis of variance (ANOVA) — A method of
analysis used when dealing with a continuous
or integral dependent variable and one or more
categorical or nominal variables

Arithmetic mean —The sum of all observations

divided by the number of observations. Also
known as the mean.

Branching — See skip pattern

Categorical variable — A variable for which

numbers are simply identifiers and do not have
mathematical properties, such as order. For
example, the sales territory in which a company’s
customer lives (Central, North, South) is a cate-
gorical variable. Also called a nominal variable.

Census — An accounting of an entire population,

as opposed to a survey of a sample of that
population

Chi-square — A statistic often used in

crosstabulations to test the hypothesis that
the row and column variables are independent;
that is, whether the observed distribution is
likely due to chance

Closed-ended question — A question for which

response categories are provided

44
Coding scheme — A method for assigning a
code (usually in the form of a number) to
responses to a question. For example, if you
are researching customers’ opinions of a
certain product feature, you might devise a
coding scheme to identify a positive opinion
with a 1, a negative one with a 2, and a neutral
one with a 3. Coding schemes are also used to
turn open-ended text responses into data that
can be analyzed.

Commitment card — A card that asks respondents

to commit to participating in a survey

Confidence band or interval confidence — A

specified range around a survey result for
which there is a high statistical probability that
it includes the value that would be calculated
from the whole population (if that were possible).
Such confidence intervals are commonly
calculated for confidence levels of 0.95 or 0.99.

Continuous variable — A variable whose

response options have an implied order and
distance and for which one unit represents
the same quantity throughout the scale. For
example, age in years or weight in pounds or
kilograms. Also called an interval variable.

Crosstabulation — A table that shows the

relationship between two or more variables
by presenting all combinations of categories
of variables

45
Error bar chart — A chart that plots the
confidence intervals, standard errors, or
standard deviations of individual variables

Exact tests — Tests that calculate the probabilities

exactly, rather than by using estimates, to
determine if there is a relationship between
variables. Exact tests are necessary when you
have small datasets, small subgroups, or
unbalanced distributions.

Factor analysis — An analytic technique that

groups quantitative variables according to their
degree of correlation

Focus group — A moderated group discussion

about a particular topic. The discussion typically
lasts about two hours and is led by a moderator
who follows a topic guide but does not use a
fixed questionnaire.

Frequencies — A table showing what number

or percentage of respondents gave each
answer to a question

Histogram — A bar chart in which continuous

variables are shown in groups

Imputation — A methodical process for making

an assumption about the value of missing
data. For example, if certain demographic
information is missing from a respondent’s
questionnaire, a model can be built comparing
information that is available to data provided
by other respondents. The model would then
assign a likely value to the missing data.

46
Interval variable — See continuous variable

Level of measurement —The way in which a

question may be answered. There are four levels
of measurement: nominal, ordinal, interval,
and ratio. (See separate entries for descriptions.)

Mean — See arithmetic mean

Median — A measure of central tendency for

continuous or ordinal data, defined for
ungrouped data as the middle value when
data are arranged in order of magnitude

Missing data — Incomplete or invalid data.

Data can be missing for a number of reasons:
for example, questions left unanswered,
marked incorrectly, or marked “Don’t know.”
Missing data are usually excluded when
calculating percentages. However, sometimes
missing values can be assigned using imputation.

Mode — The value of a variable that occurs

more frequently than any other value

Nominal variable — See categorical variable

Nonparametric tests — Statistical tests that

require either no assumptions or very few
assumptions about a population’s distribution

Non-response rate — The proportion of sample

population that did not respond to a survey

Open-ended question — A question for which

no response list is provided. Respondents are
expected to supply a response in their own words.

47
Ordinal variable — A variable whose response
options have an implied order but no implied
distance. For example, a scale that ranges
from “strongly agree” to “strongly disagree.”

Pilot study — The administration of a

questionnaire under field conditions to a
small sample in order to time it and/or
uncover problems. Also called a pretest.

Population — The totality of things or people

that you wish to study

Pre-notification card — A card alerting prospective

respondents that a survey will arrive

Pretest — See pilot study

Purposive sampling — A sampling procedure

in which each element of the population is
purposely selected for some characteristic or
characteristics of interest

Questionnaire — A set of questions designed

to generate data necessary to accomplish the
objectives of the research project

Random digit dialing — The technique of dialing

random numbers in working telephone exchanges
so that people with unlisted phone numbers
are not excluded from a sample population

Random sampling — A sampling procedure

that selects population elements based on
chance. This ensures that the sample accurately
represents the population.

48
Ratio variables — Variables that have order
among points, equal distances between
adjacent points, and an absolute zero

Regression — An estimation of the linear

relationship between a dependent variable
and one or more independent variables

Response rate — The proportion of a sample

population that responded to a survey

Sample — A subset of a population from

which information is collected in order to
obtain information and draw conclusions
about the total population

Scatterplot — A graph of data points based on

two continuous variables. One variable defines
the horizontal axis and the other variable
defines the vertical axis.

Simple random sampling — (SRS) A sampling

procedure by which population members are
selected directly from the sampling frame. This
results in there being an equal probability of
selection for all population members that
appear in the frame.

Skewed — A distribution whose frequency

curve is not symmetrical about its mean,
having one “tail” longer than the other

Skip pattern — A method of questionnaire design

that enables respondents to skip questions,
based on their response to a previous question.
Also called branching.

49
Strata — (Plural of stratum) In sampling,
groups defined by certain characteristics
(See stratified sampling.)

Stratified sampling — A sampling procedure

in which respondents are separated into
subgroups or according to characteristics
of interest, and samples drawn from each
subgroup. Income level, race, and business
title are examples of characteristics that might
be used to create a stratified sample.

Systematic sampling — A random sampling

method that is equivalent to a simple random
sample

Survey — The process of collecting information

about a topic or issue by means of sampling
and interviewing selected individuals

t test — A hypothesis test that uses the t

statistic to determine whether or not two
means are equal in the population

Weighting — Assigning a numerical coefficient

to an item to express its relative importance in
a frequency distribution

White space — On a printed page, an area that

contains no text or graphics

50
About SPSS Inc.
For more than 35 years, SPSS Inc. (NASDAQ:
SPSS) has helped commercial, academic, and
government organizations maximize productivity
and accuracy through every phase in the survey
research process, from ad hoc research projects
to high-volume, complex programs. SPSS Inc.
offers a broad range of end-to-end survey and
market research tools to support survey
design, authoring, and sampling, as well as
data collection, analysis, and publishing.

SPSS survey research products are used for

a wide variety of research purposes:
■ Market research

■ Product enhancements

■ Product preferences and usage

■ Customer attitudes and behaviors

■ Demographic surveys

■ Referral identification

■ Population/member demographics

■ Employee satisfaction surveys

■ Employee attitude studies

■ Work schedule preferences

■ Benefit analysis

■ Government agency surveys

■ Policy analysis

■ Constituent preferences

■ Crime analysis

■ Academic surveys

■ Testing and evaluation

■ Student interest studies

■ Professor/course evaluations

51
SPSS is also a leading provider of predictive
analytics software and solutions. SPSS predictive
analytic technologies connect data to effective
action by drawing reliable conclusions about
current conditions and future events. More
than 250,000 commercial, academic, and
public sector customers rely on SPSS technology
to help increase revenue, reduce cost, improve
processes, and detect and prevent fraud.
Founded in 1968, SPSS is headquartered in
Chicago, Illinois. For additional information,
please visit www.spss.com.

SPSS products for survey and

market research
SPSS offers many products to assist survey
researchers.
■ With our SPSS family of products, you can

create questionnaires and analyze them on

the desktop, and then create and distribute
reports over your network or online
™
■ With our Dimensions family of survey

research products, you can create more

complex and customized questionnaires,
field surveys in multiple modes, and collect
data using advanced sample management
and data storage features
■ We also offer the Dimensions ASP and the

Dimensions Service Bureau to enable your

organization to develop and field surveys
over the Web, gaining the benefits of our
Dimensions software without having to own it

52
■ You can use SPSS Text Analysis for Surveys
to create code frames and categorize open-
ended text responses more quickly and
reliably, regardless of the program used to
create your questionnaire

By using SPSS products for survey and market

research, you build a system that meets your
needs today and is flexible enough to help you
meet tomorrow’s goals.

Planning
SamplePower® — Easily determine the sample
size you need, saving your organization both
time and money

Questionnaire Design
SPSS Data Entry™ — Design surveys and forms
on the desktop

mrInterview™ — Use a Web browser to design

questionnaires for fielding online or, with the
mrInterview CATI option, by phone

mrPaper™ and mrScan™ — Create paper

questionnaires and set them up for electronic
scanning

mrTranslate™— Manage translations of

questionnaires and reports

53
Data collection
SPSS Data Entry™ — Enter survey responses
through your desktop or networked data
entry stations

mrInterview — Collect survey data online

or by phone

mrPaper and mrScan — Collect survey data

on paper and scan to capture responses

Data preparation and management

SPSS Base — Offers a number of tools and
wizards to help you put data in a format
that’s optimized for analysis

SPSS Missing Value Analysis™ — Quickly

analyze the extent of your missing values
problem and take appropriate action for
more valid conclusions

SPSS Text Analysis for Surveys™ — Categorize

text responses quickly and reliably

Data analysis
SPSS Base — Includes all the tools you need
to analyze your data, including frequencies,
descriptives, crosstabs, t tests, regression,
and more.

SPSS Conjoint™ — Discover preferences and

develop attribute ratings

54
SPSS Exact Tests™ — Make correct conclusions
when working with small samples

Reporting, presentation, and distribution

SPSS Maps™ — Accurately plot your strategic
course with integrated, powerful mapping
analysis

SPSS Tables™ — Present complex tabular

reports with multiple responses and
missing values

SPSS WebApp Framework™ — Build custom

Web applications powered by SPSS

SmartViewer® Web Server™ — Distribute your

results electronically with Web-based online
tables and charts

mrStudio™ — Automate the creation of tables

and reports

mrTables™ — Create tables from your survey

data

Modeling
SPSS Advanced Models™ — Analyze complex
relationships using this high-end modeler’s
toolkit

SPSS Categories™ — Reveal underlying

relationships in categorical data through
perceptual maps with powerful optimal
scaling and dimension reduction techniques

55
SPSS Classification Trees™ — Create highly
visual classification and decision trees to find
specific subgroups and relationships in your
data that you might not uncover using more
traditional statistics

SPSS Complex Samples™ — Design, analyze,

and interpret non-random surveys and more
easily and accurately study populations

SPSS Regression Models™ — Make better

predictions with powerful regression procedures

Amos™ – Understand and predict behaviors,

attitudes, or actions with state-of-the-art
structural equation modeling (SEM)

Products in the SPSS family are available

for Microsoft® Windows®, UNIX®, and other
platforms. Amos and products in the Dimensions
family run in a variety of Windows operating
environments.

For detailed information on SPSS products,

visit our Web site at www.spss.com.

56
What the experts are saying about SPSS
products for survey research:

“…with SPSS our senior managers are alerted

to potential problems a lot more quickly
and accurately. That helps them take
action, and ultimately, to ensure more
people are in our seats, rather than
those of our competitors.”

– King Douglas
Senior Analyst and Database Manager
American Airlines

“As our primary data collection engine for

online interviewing, mrInterview enables us
to deploy sophisticated surveys online. The
depth of mrInterview’s features, including
sample management and quota control,
coupled with its customization options,
makes it an ideal addition to our research
infrastructure.”

– Michael Reuscher
Vice President, IT Project Management
Synovate

“By using the software tools provided by SPSS

we have cut our process time by 70 percent.
… Because the output generated by SPSS
is so impressive, it makes people in authority
sit up and take notice of our results.”

– Jacqueline Martin
Principal Support Officer
East Ayrshire Council

57
 “SPSS gave me answers to my ‘how’ and
‘why’ questions on customer preferences.
We learned exactly where customers
think we are struggling and can allocate
resources to address the problem.”

– Sharon James
Process Specialist
Haworth Furniture

“With SPSS, I save about two weeks of

time each quarter (along with $20,000
in annual consulting fees) from data
input to number crunching.”

– Julie Frandsen
Assistant Editor
Aspen Publishers

“SPSS Text Analysis for Surveys’ ability to

distinguish between positive and negative
open-end responses is the ‘Holy Grail’ of
survey text analysis.”

– Bob Muenchen
Manager, Statistical Consulting Center
The University of Tennessee Office of
Information Technology

SPSS has a worldwide network of distributors.

To locate the SPSS office nearest you, go to
www.spss.com.

SPSS is a registered trademark and the other SPSS

products named are trademarks of SPSS Inc. All other
names are trademarks of their respective owners.
© 2004 SPSS Inc. All rights reserved. STIP-1004

ISBN 1-56827-090-9 Printed in the U.S.A.