You are on page 1of 5

Developing Effective Questionnaires

CHRISTOPHER E. BORK
and J. BRUCE FRANCIS

Many research questions are effectively answered by collecting data about the
opinions, beliefs, and perceptions of large numbers of persons in widely dis-
persed areas. The questionnaire is an economical means of gathering such
information. Properly designed questionnaires can collect valid and reliable data
for analyzing a research problem. This article reviews fundamental principles of
questionnaire design and presents a matrix to guide the construction of a
questionnaire. Methods to avoid common problems and pitfalls in developing a
questionnaire are also presented. Field testing is recommended to ensure validity
and reliability.
Key Words: Physical therapy, Research design.

The need for knowledge that goes beyond the basic and hypotheses. Before developing a questionnaire, however, a
clinical sciences increases as the physical therapy profession researcher must decide whether its use is appropriate for a
continues to develop and mature. An understanding of the particular study.
feelings, attitudes, and concerns of the members of the profes-
sion, the patients served, and the public in general is necessary
when clinicians plan, implement, and assess changes in the ADVANTAGES AND LIMITATIONS OF A
practice of physical therapy. The realm of survey research QUESTIONNAIRE
includes concepts and strategies that enable the researcher to
accumulate data concerning the attitudes, opinions, and We are assuming the researcher has identified the problem
knowledge of a population or sample.12 Survey research is to be studied, translated it into a testable form, and defined
also useful to describe a particular knowledge, such as what the variables operationally. The next tasks are to determine
physical therapists know about the legal implications of in- what data must be collected and analyzed, identify sources
dependent practice.1 where the data can be obtained, and describe how data will
One of the most potent survey research tools is the ques- be gathered. As a tool to collect data, the questionnaire has
tionnaire, which can be administered directly, by telephone, advantages and limitations that must be taken into account.
or through the mail.3 We will discuss only questionnaires sent The main advantages of using a questionnaire are the
through the mail in this article. Because a questionnaire allows following:
a researcher to gather data conveniently from large samples, 1. Data are gathered economically because the cost for each
it is a useful tool to collect and analyze data from groups that
may be affected by changes in professional practice or edu- completed response is low.2-6
cation. Unfortunately, many questionnaires contain major 2. Distributionalflexibilityis provided because the question-
flaws that render them incomprehensible to respondents or naire can be sent almost anywhere.4-6
do not collect the intended data. If a questionnaire is incom- 3. Data can be gathered from a large number of subjects.4-6
prehensible to the recipient, it is unlikely to be returned; a 4. The data gathered can be standardized, especially if an
questionnaire that does not gather valid information does not item requires a response from a group of choices provided
allow the researcher to address the research problem being by the researcher.4
studied.3 5. The process is time efficient because it generally involves
A researcher's attention to several fundamental principles less of the researcher's time to gather data in comparison
for developing a questionnaire greatly enhances the likelihood with an equivalent number of interviews.2, 5, 6
that valid and reliable data will be gathered. This article is The main limitations of a mail questionnaire are the follow-
intended to assist the reader to develop questionnaires that
collect valid and reliable data. Generally, the term validity ing:
refers to the ability of an instrument, such as a mailed 1. The subject's motivation while completing a questionnaire
questionnaire, to measure actually what the researcher intends is unknown.4
to measure. Reliability refers to the consistency of a question- 2. Data cannot be gathered from subjects who are illiterate.5
naire in gathering data. Optimally, each time an individual 3. The researcher must assume all subjects have sufficient
responds to an item he or she should choose the same re- knowledge to complete the questionnaire.3,4
sponse. Thus, carefully developed questionnaires tend to 4. The researcher must assume that all subjects are willing to
gather data that accurately represent the knowledge, attitudes, answer all questions.3, 4
or opinions of the respondents and address the researcher's 5. All subjects who receive questionnaires may not complete
and return them.2, 3, 5, 6
Dr. Bork is Associate Professor and Director of Graduate Studies, Depart- 6. Questionnaires do not allow follow-up questions such as
ment of Physical Therapy, College of Allied Health Professions, Temple Uni-
versity, Philadelphia, PA 19140 (USA). an interviewer may pose to clarify a given response.5
Dr. Francis is Academic Vice-President, Walden University, Minneapolis, 7. Although each subject receives the same items in a ques-
MN 55401.
This article was submitted September 6, 1983; was with the authors for
tionnaire, test conditions cannot be standardized.5
revision 51 weeks; and was accepted December 12, 1984. 8. Subjects may not interpret items contained in a question-

Volume 65 / Number 6, June 1985 907


enable the reader to fill in cells of the matrix and to avoid
pitfalls frequently encountered when developing a question-
naire.

Deciding What Information to Collect


A common tendency in survey research is to gather more
data than necessary.3,4 This results in questionnaires that are
too long and discourage the respondents from completing
them.4-6 Gathering more data than necessary also makes the
data analysis more complicated because decisions must be
made about which data are essential to the study and which
are superfluous. Justifying each item by the matrix ensures
that only items that gather information for testing the hy-
potheses of the study are included in the questionnaire.

To help refute or validate a hypothesis,3,4, 7 include demo-


graphic information that can be used to test the relationship
of sample to population and possibly the relationship of
variables to one another. Sometimes more than one item for
one hypothesis is necessary to elicit the information the
researcher seeks.3, 5 For example, multiple items are needed if
detailed information is sought or if the information is of a
sensitive nature and must be elicited in a manner to keep the
Fig. 1. Questionnaire item construction matrix. respondents inclined to provide the information.

naire in the same manner. Perceptions of what an item Knowledge of the Respondents
asks may differ.4, 6
After carefully reviewing the advantages and limitations of The characteristics of the potential respondents are impor-
using a questionnaire for a particular study, a researcher must tant to the design of the questionnaire and the items. The
decide if the advantages outweigh the limitations. The poten- researcher must be confident that the respondents possess
tial effects of using a questionnaire on the validity and relia- adequate knowledge to complete the questionnaire.3, 4 If the
bility of the study must also be considered.4 Using a question- questionnaire must use technical language or jargon, the
naire simply because it is economical and easily administered respondents must be able to comprehend fully this language.3
may result in the collection of meaningless data. For example, A knowledge of the characteristics of the respondents is im-
suppose a researcher wishes to compare, one year after dis- portant to avoid asking unanswerable questions. An example
charge, the functional ability of individuals who received a of an unanswerable item is, Do you prefer TENS or cold laser
particular exercise regimen with individuals who did not. therapy for pain reduction? If asked of physical therapists
Using a questionnaire to ascertain individuals' functional with access to both modalities, the question would be accept-
abilities probably would not yield valid and reliable data with able, but if asked of physical therapists in general, the question
the possible exception of what the subjects perceived their might not be answerable to those clinicians without access to
functional ability to be. In this example, the researcher should both modalities. If the researcher incorrectly assumes that all
decide that a questionnaire is inappropriate for the study. respondents will be able to answer an item and no provision
is made for the respondents to indicate that they are unable
USING A MATRIX TO TRANSLATE THE to provide the information, the data gathered might be inac-
RESEARCH PROBLEM INTO ITEMS curate or misleading. Some simple guidelines to avert this
problem include avoiding complicated syntax (ie, lengthy,
Translating a research problem into items suitable for a complex statements), omitting technical jargon unfamiliar to
questionnaire requires an understanding of the kinds of in- any potential respondents, choosing only topics familiar to
formation necessary to validate or refute each hypothesis.2,4 respondents, and adding a response to indicate insufficient
Equally important is a knowledge of the characteristics of the knowledge for respondents who are not able to complete the
persons who will be answering the questionnaire, the respond- item.3, 5
ents.2-4 A questionnaire can be designed in a logical and
efficient manner if the following concerns are addressed: 1) A second important reason for understanding the respond-
What information must be elicited from the respondent to ents is to determine whether the respondents will volunteer
test the hypothesis? 2) Are the respondents likely to be willing the desired information. A researcher must be cognizant of
and able to provide this information? and 3) How will the social and cultural mores that may affect the willingness of
data be analyzed? the respondents to provide information.2-4 For these reasons,
We have developed a matrix that has been helpful in questions about the potential respondents are important to
constructing questionnaires and checking items (Fig. 1). The answer in the matrix. When the researcher has completed this
following sections of the article correspond to the major section of the matrix, she or he is ready to construct question-
sections of the matrix. Each section provides information to naire items.

908 PHYSICAL THERAPY


PRACTICE

CONSTRUCTING ITEMS quently or infrequently. Consider how a respondent will


complete the item if she or he strongly agrees that patients
A researcher must decide how many items are needed to should be reevaluated but not necessarily frequently. By en-
gather data to test each hypothesis. The appropriate number suring that an item makes only one declarative statement or
depends on the complexity of what is being studied and the asks only one question, this pitfall may be avoided.
willingness of the respondents to provide the information. In A final common error in item construction is an implicit
some cases, more than one item is necessary to gather relevant or explicit bias in wording.2, 3 Because the researcher is seeking
data. For example, consider how a researcher would approach
information from the respondents, any statement or phrase
gathering data to test the hypothesis that physical therapists'
that tends toward favoring a particular response must be
attitudes about a specific therapeutic technique are related to
avoided. An accurate portrayal of each respondent is sought.
their knowledge of the technique. Several items relating to
Consider the following examples of an item to determine the
various aspects of the technique may be appropriate to ascer-
happiness of Californians:
tain the respondents' knowledge of the technique. In contrast,
1. Do you think that people residing in sunny California are
only one item may be necessary to ascertain therapists' atti-
tudes concerning the technique. In some cases, more than happier than people residing elsewhere in the country?
one item may be necessary to approach a potentially sensitive 2. Do you think that people residing in California are happier
issue gradually. The decision of whether more than one item than people residing elsewhere in the country?
is necessary is thus dependent on each hypothesis and the The first sample item biases the respondents by using the
nature of the data being sought.2-4 word "sunny," which may tend to evoke positive feelings
toward the sunshine and climate of California.
Because the purpose of a questionnaire is to elicit infor-
Poorly constructed or biased items decrease the likelihood
mation from respondents in a consistent, and optimally, a
that the information gathered will be reliable and valid. By
standardized manner, the utmost care must be exercised when
conscientiously examining each item for construction errors,
constructing each item. The information elicited by an item
the researcher can avoid needless frustration later on when
may be a fact, an attitude, or an opinion. If an item is poorly
analyzing the data.
constructed, it may gather erroneous information that will be
of questionable use to the researcher. Flawed items result in
communication errors by respondents. These can be avoided Response Formats
by attention to the information desired and the format of the
item. The response format provides the respondent with a means
to communicate facts, attitudes, or opinions. An inappro-
Common Errors in Item Construction priate response format generates incomplete or erroneous
information. An earlier example requested information on
Ambiguity in an item forces the respondent to interpret the whether clinicians preferred TENS or cold laser treatment for
item and make decisions based on those interpretations. If patients with pain. We pointed out that many respondents
several interpretations are possible, the researcher may draw might not be familiar with one or both of the modalities and
erroneous conclusions from the data because the respondents' would lack the knowledge to complete the item. If the re-
interpretations differ substantially from the researcher's.3 An sponse format included no choice to indicate a lack of famil-
example of an ambiguous question requiring an interpretation iarity with the modalities, respondents would have no way to
is, Do you use transcutaneous electrical nerve stimulation communicate this important information. Therefore, the re-
(TENS) frequently? Respondents answering this item must searcher must select response formats that best convey the
interpret what is meant by "frequently." Consequently, what information each questionnaire item seeks from the respond-
the researcher considers "frequently" and what the respond- ents.
ents consider "frequently" may vary significantly. Thus, the One means of classifying response formats is open- or
information provided by this item is likely to be misleading closed-ended. The criterion for classifying response formats is
and almost useless to the researcher. whether the respondent is limited in response choices. If the
To avoid errors of ambiguity, the researcher must be as respondent is required to choose from a limited set of re-
precise as possible when developing items.7"9 The less subject sponses, the format is closed-ended. If there is no limit, the
to interpretation an item is, the more likely it accurately response format is open-ended.
represents the respondent's view. Optimally, an item should The open-ended format is often called the essay response
be reliable, that is, convey the same meaning to all respond- because the respondent is asked to respond by writing the
ents. In the previous example, we can correct the ambiguity answer in her or his own words. By allowing the respondent
by changing the nonspecific words: Do you use transcutane- such freedom, the researcher is provided a response that is
ous electrical nerve stimulation (TENS) more than three times rich in content. The open-ended format has the advantage of
per week? Note that the term "frequently" has been specified allowing for all possible responses to a given item.10 Many
to mean "more than three times per week" so that respondents individuals designing questionnaires use the open-ended for-
are less likely to interpret the question differently. mat to determine the range of responses to an item.2, 3 If a
Another common error in item construction is an item that researcher is interested in the respondents' knowledge about
asks the respondent to address two or more questions or a question, the respondents' frames of reference, or their
statements.9 Consider this poorly worded example: Physical motivations, the open-ended format should be chosen.
therapists should reevaluate their patients frequently. The Open-ended formats are not without several drawbacks.
declarative statement asks for the respondent's opinion on When the respondent is free to choose the frame of reference,
two issues: 1) Whether physical therapists should reevaluate the response may not provide the information anticipated by
patients, and 2) whether patients should be reevaluated fre- the researcher.3 Completing a large number of items with an

Volume 65 / Number 6, June 1985 909


Please place a check in t h e box which that best represent her or his response to the item. Often
b e s t r e p r e s e n t s your r e s p o n s e t o t h e respondents are asked to rank their choices to provide more
s t a t e m e n t below. information for the researcher.
Another variation of the closed-ended response format is
There should be certification for the scale, whereby respondents must select a point on a
physical therapy specialists. continuum of possible responses (Fig. 2). Scales are generally
used to gain information on the relative intensity of attitudes
or opinions.
The question of whether to use a closed- or open-ended
response format depends largely on the information the re-
searcher is seeking. Most experts recommend the thoughtful
use of both response formats so that all essential information
Fig. 2. Item with a scale response format. can be obtained.2, 3 An important consideration in selecting
response formats is how the responses will be analyzed. The
investigator must be certain that the data collected are appro-
open-ended response format can also be tiresome to respond- priate for the statistical tests available to test each hypothesis.
ents. Many may not complete and return the questionnaire.6
Some respondents may have difficulty expressing themselves
Data Analysis
in written language, and words chosen by a respondent to
convey a certain meaning may represent an entirely different The matrix requires the researcher to describe the type of
meaning to the researcher.6 Written communications, unless data to be collected and the proposed method of analysis. The
carefully worded, tend to be imprecise and are interpreted in potential usefulness of the data and the appropriateness of the
ways other than those intended. Finally, the diversity, individ- method of analysis can be evaluated. Hence, the matrix serves
uality, and richness of open-ended responses make them not only to identify what information should be sought but
difficult and time-consuming to code for analysis.6, 10 also to determine the appropriateness of the information.
Open-ended response formats simply provide space to allow
the respondent to answer the item. The respondent may be SEQUENCING ITEMS
asked to provide any one of the following:
1. One bit of information—What do you feel is the principal The final series of cells in the matrix begins with the
value of independent practice by physical therapists? sequence of items in the questionnaire. To fill these cells, the
2. More than one bit of information—What qualities should investigator must determine and justify the sequence of the
a person possess if interested in becoming a physical ther- items. The arrangement of items should follow a sequence
apist? that appears logical from the respondent's viewpoint.3, 5 A
3. The ranking of information—The types of patients, edu- sequence that appears logical to the investigator for providing
cational benefits, and salary are three important factors information to test a hypothesis may confuse and frustrate
influencing initial job selection by entry-level graduates. the respondents, who are then less likely to complete and
Please rank these factors from the most important to the return the questionnaire. Arranging items in a sequence that
least important. appears logical to the respondents increases the probability
When the researcher is aware of all the respondents' frames that questionnaires will be completed and returned.5
of reference and all potential responses, the closed-ended To reduce frustration and increase interest, begin the ques-
format may be used. Advantages of a closed-ended response tionnaire with items that seem to be the easiest to answer and
format include ease of answering and coding items because the most interesting.1 Although some authors suggest that the
of the limited and simultaneously focused choices.2, 3 On the initial items should be objective and impersonal to establish
other hand, the closed-ended response format may sacrifice a mind-set, whereby the respondents are encouraged to re-
richness; by selecting this format, the researcher assumes that main objective, holding the respondents' interest remains a
all possible responses are represented by the choices given.2 primary concern.8
Finally, because of different interpretations by respondents The items should be sequenced in ways that focus the
and the researcher, the response chosen by a respondent may respondents' attention to help them provide information.1
not truly represent the information the respondent meant to Items that are dependent on the ability of the respondent to
convey. recall an event from the recent past are best sequenced by
Closed-ended response formats may be structured in several asking first about current events and gradually about events
different ways. The designs called dichotomies or two-way up to two months in the past.10 Generally, researchers are
answers ask the respondent to choose between two alternatives concerned about the accuracy of individuals' memories of
that, optimally, are mutually exclusive.9 Some authors also events after two months have passed. Items calling for specific
suggest including an alternative for respondents to designate information should be preceded by related items that ask for
that they lack the information to answer the item.9 An ex- general information and gradually increase in specificity.3
ample of this response format is illustrated by the item, Do Items that ask for personal information and items that are
you think all physical therapists should have a master's degree the most difficult to answer should appear near the end of the
to work in a specialty? The responses include Yes, No, or questionnaire.10 Respondents who are frustrated or are made
Don't Know. defensive by difficult questions may elect to skip them; yet,
Another form of the closed-ended response format is the they have already provided substantial information elsewhere
multiple-choice design. In a multiple-choice design, the re- in the questionnaire. Respondents who have answered all
spondent must choose either one or a number of alternatives previous items may not be as likely to be intimidated or

910 PHYSICAL THERAPY


PRACTICE

frustrated by difficult or personal items. In effect, what has sequentially. Response formats must be constructed to reflect
taken place is the establishment of a type of rapport between accurately the information sought, and instructions to the
the researcher and the respondents through the medium of respondent must be clear. Before printing, a questionnaire
the questionnaire.1, 5 When this happens, respondents may should be field tested and revised if necessary.
not feel threatened and will provide the information re- Survey research can illuminate many problems and issues
quested. of concern to the physical therapy profession. This article
serves as a starting point for those who wish to contribute to
Instructions the body of knowledge and elect to use a questionnaire.

The matrix concludes with cells requiring the researcher to REFERENCES


identify and justify instructions. Instructions given the re-
1. Babbie ER: The Practice of Social Research, ed 3. Belmont, CA, Wads-
spondents for completing the mail questionnaire should be worth Inc, 1983, pp 209-228
clear and concise. If the researcher wishes to emphasize the 2. Nisbet JD, Entwistle NJ: Educational Research Methods, New York, NY,
instructions, several methods are available, such as printing Elsevier Science Publishing Co Inc, 1970, pp 44-53
3. Sheatsley PB: Questionnaire construction and item writing. In Rossi PH,
the instructions in a different color from the items or printing Wright JD, Anderson PB (eds): Handbook of Survey Research. New York,
key words or phrases in capital letters.5 In either case, the NY, Academic Press Inc, 1983, pp 195-230
objective is to present the instructions as clearly as possible. 4. Mouly GJ: Educational Research: The Art and Science of Investigation.
Boston, MA, Allyn & Bacon Inc, 1978, pp 189-201
When the researcher changes topics or response formats, a 5. Brook L: Postal Survey Procedures. In Hoinville G, Jowell R: Survey
transition statement should accompany the change.1, 3 A tran- Research Practice. London, England, Heinemann Educational Books,
1977, pp 125-143
sition statement helps respondents change their frames of 6. Dillman DA: Mail and other self-administered questionnaires. In Rossi PH,
reference. The investigator who changes topics without in- Wright JD, Anderson PD (eds): Handbook of Survey Research. New York,
cluding transition statements risks confusing, frustrating, or NY, Academic Press Inc, 1983, pp 359-376
7. Francis JB, Bork CE: Design and Administration of Mail Questionnaires,
antagonizing the respondents, who may then become unwill- ed 2. Naples, FL, Action Research Associates, 1983, pp 3-37
ing to provide useful information. 8. Sax G: Principles of Educational Evaluation and Measurement. Belmont,
CA, Wadsworth Inc, 1974, pp 435-442
9. Payne SL: The Art of Asking Questions. Princeton, NJ, Princeton University
Press, 1951, pp 228-237
FIELD TEST 10. Courtenay G: Questionnaire construction. In Hoinville G, Jowell R: Survey
Research Practice. London, England, Heinemann Educational Books,
Before printing and mailing a questionnaire, the investiga- 1977, pp 27-54
tor should test the usefulness of the questionnaire.3, 5, 10 The
questionnaire should be completed by a group of individuals
who are most nearly like the target population for the ques-
tionnaire. The purpose of this trial is to discover and correct
any unforeseen problems with the questionnaire. In addition,
the researcher will gather preliminary data that will provide a
sense of the quality of the forthcoming study data.
The field test provides the researcher an opportunity to
validate the questionnaire by examining whether the data
gathered were appropriate to test the hypotheses and if the
data accurately represented the respondents' knowledge, opin-
ions, and attitudes. A researcher can also assess the reliability
of items by interviewing the respondents in the field test and
comparing their verbal responses with their responses to the
questionnaire item. If the questionnaire is reliable, the differ-
ences will be minimal.
After the questionnaire has been field tested and any nec-
essary changes incorporated, the researcher can have it printed
and mailed to the target population. One fundamental maxim
must be remembered when designing and administering ques-
tionnaires: No matter how excellent a questionnaire appears,
some oversights are likely. Therefore, although a researcher
should, and indeed must, work conscientiously to construct a
questionnaire, the instrument produced will not always be
perfect.

SUMMARY
The questionnaire, when properly designed, is a useful
approach for survey research. In constructing a questionnaire,
the researcher should include items that gather data to address
specific hypotheses. Using a matrix in developing a question-
naire can help the researcher ensure that important aspects of
construction are addressed and that all items included are
justified. Items should be worded concisely and arranged

Volume 65 / Number 6, June 1985 911

You might also like