class 11questionnaire design

Upload: nitesh-panch

Post on 03-Apr-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/28/2019 Class 11questionnaire Design

    1/38

    Designing Questionnaire

    Dr. Sasmita Mishra

    KSOM, KIIT University

  • 7/28/2019 Class 11questionnaire Design

    2/38

    INTRODUCTION

    Despite much progress, designing questionnaire is still an art.

    Neither a basic theory nor even a fully systematised approach tothe problem has been developed.

    The extensive experience of many researchers and organisedexperiments have led to a considerable understanding of theproblem and to a long list of"dos and don'ts" and rules of thumb.

    These help researcher avoid many pitfalls but they can not besubstituted for creative imagination.

    A procedure can be developed by systematizing the steps.

    Stages are presented in a particular sequence.

    Rarely to develop a questionnaire in this step-by-step fashion. A

    more typical development will involve some iteration and looping. Secondly, these steps should be taken as a guide. With

    questionnaire

  • 7/28/2019 Class 11questionnaire Design

    3/38

    Specify what information will be sought

    Determine type of questionnaire & method of administration

    Determine the contents of the individual questions

    Determine the form of response to each question

    Determine the question phrasing

    Determining the number of questions and sequence of each question

    Pre-test the questionnaire and revise, if necessary

    STEPS IN QUESTIONNARE DESIGNING

  • 7/28/2019 Class 11questionnaire Design

    4/38

    Step 1: Specify what information is needed

    Basically a questionnaire must serve two functions:

    1)Translate research objectives into specific questions, and

    2)Motivate respondent to corporate and furnish inf. correctly.

    Research objectives generally lead to the formulation of

    hypotheses and hypotheses guide the questionnaire. Hypotheses not only guide what information is needed but they

    also help in deciding the type of questions and form of response.

    During the preparation of questionnaire additional hypotheses mayemerge.

    Therefore, if a new hypothesis is important and vital to the

    understanding the problem, it should be included. On the otherhand, if any hypothesis is interesting but not vital to the problemunder consideration should be dropped.

    Lengthening may cause problems in administration and analysisand often increases non-response.

  • 7/28/2019 Class 11questionnaire Design

    5/38

    Step 2: Type of Questionnaire and Method of

    Administration

    After basic information, the next step is how it is to be gathered.

    This is with respect to the structure/disguise and whether mail,telephone, or personal interview will be used to administer it.

    If a disguised unstructured questionnaire has to be employed thenthis is not possible with telephone and also for mail.

    Mail administration is not recommended for unstructuredundisguised questionnaire with open ended questions.

    Type of data can have important effect on these questions.

    Example: the relationship between behaviour and demographic, onemay use either mail/personal/telephone interview.

    If the researcher wants measuring attitudes through attitude scaletelephone interview will be out of question.

    Thus research must specify what primary data are needed, how thesemight be collected, what degree of structure and disguise will beused, and then how the questionnaire will be administered.

  • 7/28/2019 Class 11questionnaire Design

    6/38

    Step 3: Determine the Contents of Individual Questions

    The following points are in the nature of standards againstwhich to check possible questions.

    1) Is the question necessary? Extra questions add to theexpense of the survey and increase the non response

    Question should fulfils any objective then it should beincluded

    2) Do the respondents have the necessary information?

    The respondents will be able to give a reliable answer.

    The answer will not require a great deal of time and effort

    Expl: "Which statement true of Metallic Metal Act?

  • 7/28/2019 Class 11questionnaire Design

    7/38

    Step 3: Determine the Contents of Individual

    Questions

    The researcher keeps the following with respect to each question.

    Answers respondent cannot give or cannot give accurately?

    Is the issue within the respondents experience?

    Respondent to do a great deal of work to answer the question?

    Question for opinion on matters so unfamiliar to the respondent?

    Whether the respondent is the best source of information?

    Can the respondents be expected to remember the information?

  • 7/28/2019 Class 11questionnaire Design

    8/38

    Step 3: Contents of Individual Questions (Cont)

    3) Will Respondents Provide Information?

    Sometimes respondents have necessary information but theydo not provide it.

    They may not possess necessary communication skills or theymay simply not wish to convey this information. Hide the question in a group of questions.

    Phrasing the question in terms of others.

    Providing response in terms of a number of categories.

  • 7/28/2019 Class 11questionnaire Design

    9/38

    Step 3: Determine the Contents of Individual

    Questions

    4. Are Several Questions Needed Instead of One?

    "Why do you use Colgate?

    One respondent may reply, "To reduce cavities" while of may reply,Because my dentist recommended it".

    Break this one question into two separate questions.

    (i) How did you first happen to use Colgate?

    (ii) What is your primary reason for using it?

  • 7/28/2019 Class 11questionnaire Design

    10/38

    Step 4: Determine Forms of Response/Response Format

    Open Ended Questions: Respondents are free to reply to openended questions in their own words. The following examplehighlights this type.

    1) What is your age?

    2) How much do you earn per month? Open-ended questions are often used as "openers" and for

    probe for additional information.

  • 7/28/2019 Class 11questionnaire Design

    11/38

    Forms of Response/Response Format (Cont)

    Multiple-choice Questions:

    1. How old are you

    a) Less than 20 b) 20-29 c) 30-39 d) 40-49 e) 60-above

    2. Why did you purchase western colour TV?

    i) Low Price ii) quality iii) service iv) Sharp picture v) Better Warranty

    Easy to interpret, tabulated or analysed.

    Problems

    1) Do not permit individuals to elaborate their true positions

    2) Susceptible to potential order bias.

  • 7/28/2019 Class 11questionnaire Design

    12/38

    Forms of Response/Response Format

    (Cont)

    Dichotomous Questions: Only two alternatives,

    for example:

    Do you intent to purchase a two-wheeler this year?

    Yes No

    Offers the ultimate in case of coding tabulation and less interview bias.

    There may be more than two answers.

    For example, in purchase intention question the answers may be "Definitely intend to buy

    probably will buy" Definitely intend not to buy

    "undecided"

    Dichotomous questions can often be framed in multichotomousquestion.

  • 7/28/2019 Class 11questionnaire Design

    13/38

    Forms of Response/Response Format (Cont)

    Scales:

    For instance, when inquiring about the various sewingmachines features the following question might be asked.

    How often do you use zigzag stitch on your machine?Never ( ) occasionally ( ) sometimes ( ) Often ( )

    The advantage of this pattern is that the descriptors could bepresented at the top of the page and a number of possiblefeatures could be listed along the left margin.

  • 7/28/2019 Class 11questionnaire Design

    14/38

    Forms of Response/Response Format (Cont)

    Ranking questions:

    The respondent may be required to rank feature of a product.

    Please rank the following factors, for the most important (1) to the leastimportant to you, buying a refrigerator.

    i) Company reputation ( )

    ii) Price ( )iii) Compressor ( )

    iv) Guarantee ( )

    v) Colour ( )

    vi) After sale service ( )

    vii) Maintenance cost ( )

  • 7/28/2019 Class 11questionnaire Design

    15/38

    Step 5: Determine the question phrasing/wording

    Translation of the desired question content into words andphrases that can be understood easily and clearly by therespondents.

    In general, questions, should be simple and straight forward.

    It is to ensure that the respondents and the researcherassign exactly the same meaning to the question.

  • 7/28/2019 Class 11questionnaire Design

    16/38

    Step 5: Determine the question phrasing/wording

    1. Meaning of Words

    Technical terms should be avoided unless they are inevitable. They shouldbe clarified or explained.

    Sometimes common words also create equally serious problems.

    How many members are there in your family, "Does family mean nuclear

    family or a joint family. Paynes advice: ask the following five questions .

    i) Does it mean what we intend?

    ii) Does it have any other meaning?

    iii) If so, does the content make the intended meaning clear?

    iv) Does the word have more than one pronunciation?

    v) Is a simpler word or phrase suggested?

  • 7/28/2019 Class 11questionnaire Design

    17/38

    Question phrasing/wording (Cont)

    2. Biased Words and Leading Questions

    Emotionally coloured, a feeling of approval or disapproval.

    Leading questions suggest what the answer should be or indicatethe researcher's own point of view.

    Both result in a consistent measurement error. Biased phrases are difficult to deal with because phrases that are

    neutral to one group may be emotionally charged to another.Phrases such as luxury items and leisure times are neutral tomany people, yet negative overtones to others.

  • 7/28/2019 Class 11questionnaire Design

    18/38

    Question phrasing/wording (Cont)

    3. Implied Alternatives

    If the implied alternatives are changed into explicit one then it wouldchange the selection of the response category more or less as with thecase of implied remains implicit. For example.

    "Do you think workers should have the right to elect representative on theBODs of the company they work for?" Produced 61% yes answers.

    When the phrase "or should all the directors be elected by the owners ofthe company"? The response dropped to 53 percent.

  • 7/28/2019 Class 11questionnaire Design

    19/38

    Question phrasing/wording (Cont)

    Implied Assumptions

    Questions are frequently asked in such a way that the answerdepends on assumptions about factors outside the question itself.

    "Are you in favour of curtailment of amount of advertiser allowed

    on TV? Will elicit differing responses, depending on the respondents

    assumptions converting the effects this might have on thequantity and quality of TV programming.

    A more effective way of wording the question would be

    "Are you in favour of curtailing the amount of advertising allowed

    on TV if this would have an impact on TV programming". Therefore, failure to state assumptions often produces different

    responses.

  • 7/28/2019 Class 11questionnaire Design

    20/38

    Question phrasing/working (Cont)

    5. Frame of Reference

    The wording of the question will determine which frame reference orview point the respondent will assume.

    Consider the following versions of a question to be answer by therecent claimants of an insurance company.

    i) Does GIC provide satisfactory or unsatisfactory settlement ofclaims?

    ii) Were you satisfied or unsatisfied with GIC settlement of yourrecent claim?

    Each of these versions provides the respondent with a some what

    different frame of reference. The first version calls for an objective answer that may include therespondent perceptions of other peoples standards for claimsettlement and how adequately GIC meets these expectations.

    The second question involves the personal experience of therespondent.

  • 7/28/2019 Class 11questionnaire Design

    21/38

    Step 6: Number and Sequence of Questions

    Researcher is ready to putt all questions together.

    A very long questionnaire can appear quite imposing and may deterthe respondent from even reading the instructions.

    Questions sequence is also important. A typical questionnaire

    contains two types of information: 1) Basic information: Refers to the subject of the study for example,

    intentions or attitude of respondents towards some phenomenon.

    2) Classification information: Data collected to classify respondent.These are basically the demographical socio-economic characteristicsof respondents.

    Therefore, a proper questionnaire sequence is for question securingbasic information to be presented first and those seekingclassification information to be presented last.

    The basic information is most critical, without it there is no study.Thus, the researcher should not risk alienating the respondent beforegetting to the body of the study.

  • 7/28/2019 Class 11questionnaire Design

    22/38

    Number and Sequence of Questions (Cont)

    The basic information itself can also present some sequence problem.The early questions should not be sensitive ones.

    Opening questions should be designed to secure the respondentsinterest and cooperation.

    These questions should be simple and if possible should arousecuriosity enough so that they will be willing to cooperate.

    The sensitive questions should be relegated to the body of thequestionnaire and intermixed and hidden among some not sosensitive ones.

    The order in which the questions are presented can affects the

    responses. "what improvements are needed in the companies' service policy"?"How do you like the quality of service".

    Funnel Approach should be used

  • 7/28/2019 Class 11questionnaire Design

    23/38

    Step 7: Re-examination, Revision and Pre-testingthe Questionnaire

    Re-examination and revision are the order of the day.

    Revision of each question should be done by the researcher criticallykeeping in mind whether the question is an ambiguous, confusingpotentially offensive.

    The real test how it performs under actual conditions of datacollection.

    For this assessment, the questionnaire pre-test is vital.

    Therefore, data collection should never begin without an adequatepre-test of questionnaire.

  • 7/28/2019 Class 11questionnaire Design

    24/38

    Step 7: Re-examination, Revision and Pre-

    testing the Questionnaire

    The pre-test can be used to assess both individual questions and theirsequence.

    A pre-test must use respondents who are as similar as possible to thetarget respondents.

    This typically involves drawing a sub sample from the larger sample. Thequestionnaire should be administered in the some manner as the finalsurvey.

    Churchil: Researcher who avoids a pre-test is either naive or a fool.

    The pre-test is the most inexpensive insurance the researcher can buy toassure the success of the questionnaire and the research project.

  • 7/28/2019 Class 11questionnaire Design

    25/38

    Scaling Techniques

    Comparative Scales Non-Comparative Scales

    PairedComparison

    RankOrder

    ConstantSum

    Q Sort

    ContinuousRating Scales

    ItemizedRating Scales

    LikertScale

    SemanticDifferential

    StapleScale

  • 7/28/2019 Class 11questionnaire Design

    26/38

    Comparative Scales

    Paired Comparison

    Respondents are presented

    two or more objects and

    asked to compare two

    objects at a time andchoose one on some

    criterion.

    Example:5 ad copies are

    given and asked to comparepair-wise on believability.

    Ad

    copy

    A B C D E

    A X A C A AB X X C B B

    C X X X C C

    D X X X X E

    E X X X X X

    N (n-1)/2, 5 (5-1) 2

    5* 4 = 20/2 =10

  • 7/28/2019 Class 11questionnaire Design

    27/38

    Comparative Scales

    Rank-Order Rating Scale: Respondent is required to arrange objects according to some

    criterion---for example, quality, taste, or attractiveness.

    Result is ordinal scale, describes objects from most favored to least

    favored, but says nothing about the distance between objects.

    Examples, Rank the following cars on the basis of looks: (place Ithat has the best looks and 4 that is the lowest).

    _4__ Maruti _2_ Zen

    1_ Matiz 3_ Santro

    More realistic in representing the actual situation, where consumermakes direct comparisons among all alternatives.

  • 7/28/2019 Class 11questionnaire Design

    28/38

    Comparative Scales (Cont)

    Constant-Sum Rating Scale:

    In this scale the respondents are required to allocate a given number of

    points among a number of objects according to some criterion.

    Respondents are told to allocate their points in proportion to their

    preferences for the objects.

    For example, Please divide 100 points among the following automobile

    brands so as to reflect how much overall quality you believe each one has:

    Matiz 30 pts

    Santro 60 pts

    Zen 10 pts

  • 7/28/2019 Class 11questionnaire Design

    29/38

    Non-Comparative Scales

    Continuous Rating Scales (graphic-ratings scales)

    Respondents indicate their rating by placing a check at theappropriate point on a line that runs from one extreme of theattribute to the other.

    The scale given below is an example. Please evaluate each attribute in terms of how important is

    to you by placing an X at the position on the horizontal linethat most reflects your feelings.

    Attribute Not Imp V Imp

    Courteous Service ______________________ Convenient Location ______________________

    Convenient Hours _______________________

    Low Interest _______________________

    Advantage is the ease to construct and use.

  • 7/28/2019 Class 11questionnaire Design

    30/38

    Non-Comparative Scales: Itemized Rating Scales

    Respondents are provided with a scale that has anumber of brief description associated with eachcategory.

    The categories are ordered in terms of scaleposition and the respondents are required toselect the specific category, the best describesthe object being rated.

    The most commonly used itemized rating scalesare Likert type scale, Semantic Differential scale,and Staple scale.

  • 7/28/2019 Class 11questionnaire Design

    31/38

    The Likert scale (Summated-rating Scale)

    Most widely used scaling and allows respondents to express theintensity of their feelings.

    Scale Construction: Researchers develop a number of statementsthat relate to the product or attribute in question.

    Subjects are asked to indicate their degree of agreement ordisagreement with each statement in the series.

    The responses are scored that are consistent in terms ofdirectionality (a strongly agree response to a favourable and astrongly disagree response to an unfavourable statement would

    both receive similar scores, For example, Let us assume that we are trying to measure theattitudes of students toward a hypothetical University X

    Step I: A number of favourable and unfavourable statements withregard to the university. only attitude-revealing statements, andavoid statements that are facts

    Possible statements we might use are:

  • 7/28/2019 Class 11questionnaire Design

    32/38

    2 1 0 -1 -2

    Statement SA A NAND DA SDA

    1 X Univ is a terrific place to go to

    2 It has a very attractive campus

    3 If I had to do over again, I would

    have gone someplace else

    4 As a graduate of this univ, I will

    probably get good job offers

    5 It is very difficult to make friends

    here.

  • 7/28/2019 Class 11questionnaire Design

    33/38

    Step II: Test and refine the instruments by using asample who are similar to the eventual respondents. Thepurpose is to eliminate non-contributory statements with

    regard to the attitude score.

    1) Statements for which most of the respondents tend toprovide the same response.

    2) Statements generating responses that are unrelated to

    total attitudinal score.

    Once the instrument has been administered, theresponses are translated into their numerical counterparts.At this point each individuals response to unfavourable

    statements must be multiplied by -1 for the purpose ofdirectional consistency so that +ve responses will reflectfavourable attitudes and -ve responses will reflectunfavourable attitudes.

  • 7/28/2019 Class 11questionnaire Design

    34/38

    S.No

    Statement Respondents

    numerical

    responses

    Directionalcorrection

    Corectedrespons

    es

    1 X Univ is a terrific place togo to

    1 - 1

    2 It has a very attractive

    campus

    2 - 2

    3 If I had to do over again, Iwould have gonesomeplace else

    2 X (-1) -2

    4 As a graduate of this univ,I will probably get a lotof good job offers

    2 - 2

    5 It is very difficult to makefriends here.

    2 X (-1) -2

  • 7/28/2019 Class 11questionnaire Design

    35/38

    Semantic-Differential Scale

    The respondent is asked to express feelings relative to an object

    by selecting a position along a scale bounded by bipolar

    adjectives.

    Frequently used in corporate and brand image studies.

    It is typical to have at least two brands (objects) subjected to the

    same series of adjective pairs with the data from all respondents

    averaged for each question.

    For Examples, Loud - quiet, fast - slow in case of shirts, say,Charag Din and Peter England.

    Suppose Charag Din identified with slow and quiet while Peter

    England with fast and loud.

    The result is a profile analysis:

  • 7/28/2019 Class 11questionnaire Design

    36/38

    The profile analysis:

    Peter

    England

    Charag Din

    Modern :______:______:______:______:______:______:______:Old

    fashioned

    Fast :______:______:______:______:______:______:______:slow

    Bright ______:______:______:______:______:______:______:Dark

    Loud :______:______:______:______:______:______:______:Quiet

    Friendly _____:_____:______:______:______:______:______: Unfriendly

    High Prices______:______:______:______:______:______:______: LowPrices

  • 7/28/2019 Class 11questionnaire Design

    37/38

    Stapel Scale

    It is a modification of the SDS. It differs in that (1) adjectives are testedseparately instead of simultaneously as bipolar points, (2) points on the

    scale are identified by number, and (3) there are ten scale positions rather

    than seven.

    Respondents are told to rate how accurately each of a number ofstatements describes the object of interest,

    You would select a plus or minus number for words that you thinkdescribe (Pizza Hut) accurately.

    you can select any number from +5, to5,

    +5 +4 +3 +2 +1 (Courteous Service) 1 2 3 4 5

    +5 +4 +3 +2 +1 (Good Food) 1 2 3 4 5

    +5 +4 +3 +2 +1 (Low Prices) 1 2 3 4 5

    It is less popular than semantic-differential scale.

    A profile analysis for objects evaluated with Stapel scale is generally

    similar to that of SDS.

  • 7/28/2019 Class 11questionnaire Design

    38/38

    THANK YOU