Internal
Links

Top
 Outline on  Survey Construction
External
Links
  -  Research Instruments
Link
  -  Surveys
Link
  -  Samples of Surveys
Link
  The advantages of self-administered questionnaires are that they 
- are the least costly type of survey
- allow the respondents all the time they need
- allow the respondents to feel more comfortable... privacy		  
  Disadvantages of self-administered questionnaires are that they
- have a low response rate
- are not good for complex questions
- allow Respondents a variability of effort in that some respondents fill it out hastily, while some are very thorough		  
  There are many advantages of Face-to-Face Interviews
Hi cost
Better response rate
Better quality responses
More consistent effort in responses
Interviewer can probe for more detailed response/clarification		  
  The advantages of telephone surveys include
the avoidance most of the problems of mail-in surveys
the availability of random-digit dialing
& the ease of interval number selection 		  
  Disadvantage of telephone surveys are that they are over-used by marketing companies & they must be limited to 15 to 20 minutes  
  In choosing questions, keep in mind
a. your own research goals
b. the impression a question may make
c. the response motivations        ( desire to appear knowledgeable, to express PC opinions )		  
  To achieve all of these goals, a survey should use Multiple Indicators, ( discussed below )
    i.e., several questions on the same topic, 
    to validate respondents responses
But this goal conflicts w/ the goal of keeping the survey short		  
  The Master Survey should label & group together the Multiple Indicators, 
    while the Survey should mix them up		  
  Sensitive questions may seem too personal unless the respondent sees them as justified
Instead of apologizing, make sure respondent is aware of the bearing a question has on the research
Make questions as specific as possible		  
  Questions about the Sentiments are the most common type
They are most effective when they are related to concrete realities by being combined w/ questions concerning behavior		  
  Remember, most people do not have an accurate view of their sentiments
They have an ideal view
As opposed to their operative sentiments
Example: 
Respondent believes:  "I am not sexist.  I am fair."
Operative sentiment:  "I need the best worker I can get--a man."		  
  Thus link sentiments w/ behavior anytime you can
This gives the respondent a greater chance for a more accurate view of their sentiments		  
  Knowledge questions are sometimes asked for their own sake & sometimes as filters to determine if respondent has sufficient info to provide meaningful opinions

Respondents will automatically agree or disagree w/ something they know nothing about

  
  Each question is one of the Multiple Indicators of the concepts that underlie the research design
Because the concepts being explored are likely to be complex, the survey should come at them from different angles or in slightly different ways
Thus you must use a series of related questions to produce multiple measures of a single concept		  
  Structured Questions have a fixed set of alternatives
Easy to answer
Responses are easy to code & record as data
Researcher must be able to anticipate most of the ways in which respondents will answer
 
  Unstructured Questions permit respondents to answer as they choose
Most useful when you expect a wide range of responses or when responses are likely to be detailed
But respondents may ignore or answer in brief
 
  In Wording the Questions, be as concise & direct as possible
Avoid technical jargon & over-elaboration
The meaning of every question must be clear to all respondents
Do not assume that the vocabulary of social scientists will have meaning for any others		  
  Avoid loaded questions:  those so biased that a respondent would find it difficult to take a particular position
That is, it subtly suggests that one position is preferable to another		  
  Utilize one-dimensional questions:     you should only survey one point at a time
Complex topics must be broken down into a series of one-dimensional questions		  
  Respondents should not be able to skim a questionnaire, answering many questions of a similar format w/ a similar answer (such a pattern of answers is called yea-saying  or   nay-saying )  
  Mix the types of formats:
- 7 point Likert scale
- 5 point Likert scale
- yes-no 
- agree-disagree, etc.
- mix patterns of answers ( reverse order or wording of agree-disagree )		  
  In Constructing the Survey consider both form & content
Until now, we have talked most about content,  i.e. what is in the research question		  
  Form Considerations should include 
a.  Making the survey easily readable in grammar & sentence construction
b.  Big print
c.  Don't clutter up the page
d.  Use a format that is easy to follow.		  
  Sequence the questions w/ the initial aim of capturing respondent's attention
This is especially important on surveys, since they are no use if not filled out
In an interview, the sequence must also be to establish rapport
Opening questions should be general, non-threatening, & easy to respond to		  
  The Body of questionnaire should progress through the topics following some logical pattern that the respondent is likely to recognize & will facilitate an orderly interchange  
  Sometimes the nature of the questions suggests best order
You may want to try alternative orderings to determine the best
Begin w/ the most general, advance to the most specific		  
  Sensitive questions should be reserved for the end
Subject should feel comfortable by this time
Word sensitive questions to be as inoffensive & intrusive as possible		  
  Layout & response organization should spread questions & answers out evenly on the page w/ sufficient blank space for the Subject to note where one item ends & another begins  
  Layout minimizes TWO Common Occurrences on Surveys
a. failure to respond
b. tendency to respond inappropriately or in a quick standard manner		  
  When you design the questionnaire, assume the worst:  respondents will rush through, ignore questions, etc.  
  The Mold should be consistent
It is helpful to include a sample question		  
  Filter or Contingency Questions are used to identify a subgroup of respondents
Explicit instructions should direct the subject to the next item they should answer
Use of different color pages can help isolate a series of questions  for a specific category or respondents
Arrows may also be used to direct respondents to a series of questions		  
  Use similar formats for interview schedules as on self-administered questionnaires
Sensitive questions are a greater problem for the interviewer
Possible answers on a closed list could be printed on a card & handed to the respondent to choose from		  
  The Researcher needs an Introduction for almost all types of field research:  surveys, interviews, etc.
In general, there should be TWO Introductions of any Research Instrument
a.  The Printed Introduction is printed on the survey, or other research instrument
b.  The Presented Introduction is read to, explained, etc. to the subject		  
  The tone of the introduction should be professional  
  a.  The Printed Introduction on the Research Instrument explains FIVE Facets of the Research  
  i.  Explain what the study is about in a way that captures the attention of the respondent
Keep it simple.  do not use jargon.  You can not necessarily tell the whole story because it may bias their answers		  
  ii.  Impress the Subjects w/ the importance & their participation
Emphasize that respondents should be conscientious / take care to fill out the most correct answers, but don't "second guess" themselves; go w/ 1st impressions
Respondents must be impressed upon to express their own ideas		  
  iii.  Assure Subjects that all data will be handled in a way that protects their identity
Emphasize that not even you will be able to tell who participated		  
  iv.  Link the study to any sponsor that is known & trusted
Identifying the project as a class project is less effective than linking it to a college or university		  
  v.  Explain the Disclaimer form & give 1 to the Subject to sign & allow them to keep another  
  Instructions for self-administered surveys must be short  
  b.  The Presented Introduction is read to, explained, etc. to the subject  
  The Printed Introduction is optional but strongly recommended  
  The Presented Introduction covers the same material as the Printed Introduction but attempts to cover points in greater detail, w/ greater emphasis on selected portions, & allows for some questions  
  One problem w/ surveys is Response Error in which results when a respondent misunderstands a question or intentionally give an untrue answer  
  Response Error may happen if the questionnaire contains ambiguous or double-barreled question  
  A Double-Barreled question includes more than one issue, so that the answer cannot be clearly interpreted  
  Example of a Double-Barreled question:  "Have your hours of work or your working conditions recently changed?"  
  Response Error my be the result of questions that the respondent interprets as prying into areas that are sensitive or confidential  
  Response Error is difficult to detect, & it threatens the validity & reliability of the survey  
  The best manner to determine if there is Response Error, Double-Barreled questions, & other problems w/ survey construction is to conduct a Pilot Study
Link
The End