Flashcards in Ch. 6 - Interviewing Deck (61):
The Survey Interview
- research conducted by the Vovici Company in 2010 (through a survey, of course) revealed that American adults are asked to take part in surveys 7 billion times each year and that the 80% who claim to complete the surveys they start provide 2.6 billion responses
- some of these are conducted F2F at home, malls, businesses, sporting events, hospitals, but a growing number are being conducted over the phone and via the Internet
- as the number of surveys has continued to escalate, so has the unwillingness to participate
- potential interviewees are concerned about confidentiality, privacy, telemarketers, survey accuracy, biases of survey organizations, and benefits to themselves and society
- overall there has been a markedly negative shift in attitudes toward public opinion researchers and polls across several dimensions between the mid 1990s and the first decade of the 2000s
- survey interviews: reliability (assurance that the same types of information are collected in repeated interviews) and replicability (the duplication of interviews from respondent to respondent) are essential characteristics and interviewers operate from meticulously planned and highly structured interviews during which they may ask only preplanned probing questions
- surveys reach out and touch everyone
- survey interviews are neither flexible nor adaptable
Purpose and Research
- survey interviews have multiple purposes
- before you start thinking about questions or ppl you may interview, determine precisely the purpose of your survey. the types of information you NEED to discover and HOW you will USE this information determine whether you will conduct a qualitative or quantitative survey.
- you will use a qualitative approach if you want to explore ideas and feelings, dig deeply into an issue, discover motivations, understand different perspectives, and understand behaviors
- your findings are presented in narrative from in which words are critical
- you will use a quantitative approach if you want to determine frequencies of behavior, degree of feelings, consensus of opinions, causes and effects, preferences,, averages and make predictions or strategic decisions.
- your findings are presented in quantitative form in which numbers are critical.
- your needs and q's may be tempered by the optimum length of your interviews.
- they may be as brief as three to five mins or as long as 15-20 mins.
- longer timed interviews cover more areas of interest - essential for qualitative surveys, and are more reliable, but they may be unnecessary or detrimental in quantitative surveys designed to determine a few attitudes or intentions.
Purpose and Research
- longitudinal studies reveal trends and changes over time
- don't assume adequate knowledge of a topic
- don't waste time learning what you already know
- one additional factor-time-will shape your purpose.
- you may have to complete a survey almost overnight to determine reactions to an event such as a political debate or a court decision, and this limits your purpose to a few questions probably conducted over the telephone
- other topics or issues may require weeks or months and allow you to develop longer and more complex interviews that delve thoroughly rather than superficially.
- time also determines the type of survey you need to conduct
- a cross-sectional survey takes a slice of what is felt, thought or known during a narrow time span and is used when you need to determine how interviewees are reacting at present.
- a longitudinal survey determines trends in feelings, thought or knowledge over time.
Purpose and Research
- as soon as you have a clearly define purpose, investigate all aspects of the topic.
- explore past, present and future as well as proposed and attempted solutions
- check resources such as organizational files and archives, correspondence and interviews with knowledgable people, government documents, professional journals, books, previous surveys on this topic, the internet, news magazine and news papers
- research reveals info already available in other sources that need not be gathered in your survey
- become and expert on the topic, particularly unique terminology and technical concepts
- you must determine the precise and understandable meaning of keys terms so they are clear to interviewees and acceptable to those who may read or use your results to form opinions or take actions.
- research will reveal past attitudes and opinions and speculations about current attitudes and opinions.
- thorough knowledge of the topic provides insights into areas you need to explore, the complexities of an issue, and potential intentional and unintentional inaccuracies in answers during interviews.
Structuring the Interview:
Interview Guide and Schedule
- when you have a precise purpose and conducted necessary research, develop a highly structured interview
- an interview guide is essential for survey interviews bc it dictates the topics and subtopics you will cover and primary and probing questions to be asked.
- begin your interview guide by listing major areas. for ex., if you are planning to survey Illinois educators on ways to reform the public schools in their state, major topics might include state-supported, all day kindergarten, parental involvement in the education process, etc
- if you are conducting a qualitative survey, you may develop a highly scheduled interview that includes open-ended questions, planned probes, and the possibility of unplanned probes that depend on the interviewee's responses.
- there is a degree of flexibility in questioning bc you are more concerned about depth of information than with statistical compilation of data
- the traditional interview guide (who, what, when, where, how and why) may be adaptable to qualitative surveys, but surveys often require a more detailed guide and schedule that ensures complete coverage of a topic or issue and the manner of organizing, reporting and interpreting answers
- if you are conducting a quantitative survey, you must elicit answers that are easy to record, tabulate and analyze. the flexibility and adaptability of the qualitative survey may lead to difficulties in coding and tabulation of results, so you will rely one a highly scheduled, standardized format that ensures replicability of interviews and accurate compilations of findings
- a detailed guide is easily transformed into a scheduled format
- standardization is essential for surveys
- although each interview is unique, like a small work of art with its own ebb and flow, a mini-drama that invovles real lives in real time, each respondent must go through as identical an interview as possible.
- compose an opening that typically includes a greeting, name of the interviewer, the organization conducting the survey, subject matter of the interview, purpose, amount of time the survey will take, and assurance of confidentiality.
- encourage interviewers to give this opening verbatim without reading it our sounding stilted
- to make the opening sound more natural, you may allow skilled interviewers to create their own openings as long as each opening includes all of the elements you have stipulated
- there are no icebreaker questions or small talk in surveys
- identify the interviewer and org, state the general purpose and length of interview. the survey schedule can provide instructions for the interviewer to follow and precode the question for ease of tabulating results when survey is completed
- the opening may not identify the group that is paying for it (political president, a pharmaceutical company, special interest groups) or the specific purpose (to determine which strategies to employ during a political, advertising, or lobbying campaign) bc such information might influence how the interviewee responds
- when a newspaper such as The New York Times or the Washington Post, a cable or TV network such as CBS or CNN or a well known polling group conducts a survey, the orgs name is used to enhance the prestige of the poll and the interviewer, to reduce suspicion that a candidate or corporation is behind the survey, and to motivate ppl to cooperate
- interviewers may have to show identification badges or letters that introduce them and establish legitimacy as survey takers
- bc of the increase of refusals to take part in surveys, particularly those online, and the fact that the quality of survey results depends on response rates, creators of surveys have increasingly focused on incentives ranging from simple assurances to prepaid monetary offers as high as $40 per interview
- incentives tend to come during the opening minutes of the interview when the interviewer must motivate a contact to take part in a survey
- although some research indicates the obvious, that the higher the financial incentive the greater the likelihood of participation, even token incentives may improve response rates
- on study showed that a simple prepaid ballpoint pen incentive made during the opening increased response rates and result in greater completeness in answers during the early portion of a survey interview
- non-monetary incentives included the civic obligation to help others and to be active citizens, establishing credibility of the org conducting or sponsoring survey, and stress the brevity of the survey for those with little available time.
- some survey researchers are concerned that emphasis on incentives may persuade some persons to take part to their detriment.
- the closing is brief and expresses appreciation for the time and effort expended to aid the survey. for example: that's all the questions I have. Thank you very much for your help
- if the survey organizer wants a respondent's telephone number to verify that a valid interview took place, the closing might be: that's all the questions I have. May I have your telephone number so that my supervisor can check to see if this interview was conducted in the prescribed manner? (gets the number) thank you very much
- if you can provide respondents with results of a survey, a common practice in research interviews, the closing might be: that's all the questions I have. Thanks for the help, if you'll give me your address I'll be sure that you receive a copy of the results of this study. (gets the address) Thank you very much
- some interviewees are reluctant to give their telephone numbers or email addresses to strangers. be prepared to back off from either request if the person appears anxious, suspicious or very reluctant.
- do not sacrifice the rapport and good will you created during the interview
- interviewees tend to prefer anonymity
- when one of the authors was conducting polls for a political party, he discovered many ppl did not want to give their phone numbers for fear they'd be deluged with calls from candidates during the campaign.
- with permission from the party, he stopped asking for phone numbers and the closing went much smoother
- ppl may be curious about as survey or interested in the topic and want to discuss it. this can be a good relationship builder and motivator for taking part in future surveys, but do so only if time permits, the interviewee will have no opportunity to talk to future interviewees, and the survey organization has no objections.
- create each question carefully bc you cannot rephrase, explain or expand on questions during interviews without risking your ability to replicate interviews, an essential element in surveys
- in quantitative surveys, all q phrasing and strategic decisions are made in the planning stage; not on the spot.
- in qualitative interviews, all primary questions and most probing questions are planned ahead of time
- interviewers cannot make on-the-spot adjustments
- every word in every questions may influence results
- all interviewees must hear the same q's asking in the same phrasing and manner
- a slight change in wording, vocal emphasis on a word, or facial expressions can generate different answers.
- "it is okay to smoke while praying": 90% said no
- "is it okay to pray while smoking": 90% said yes
- although the q's sound the same, ppl interpreted them differently
- the first sounded sacrilegious, lighting up while praying. the second sounded like a good idea, maybe even necessary.
- emphasis may change the focus and meaning of simple questions and this is critical in surveys in which you are striving for replicability
- a single word might alter significantly how ppl respond to a question, thereby altering the results of a survey.
- ex. where ppl viewed the word forbid as a stronger and more dangerous action than "not allow" - perhaps un-American, even though the effects of the governmental policy would have been the same.
- study discovered that validity was higher when using direct questions that presented response options and that ambiguously phrased questions took longer to ask and answer.
- make each q clear, relevant, appropriate to the respondent's level of knowledge, neither too complex or too simple, neutral, and socially and psychologically acceptable
- this is not a simple task when ppl may be of both genders, and differ widely in culture, age, income level, education, intelligence, occupation, geographical area, and experiences
- the increasing diversity of the American population may result in your target population representing widely diverse continents, cultures and nations
- be careful of using formal names or acronyms for persons or organizations with which your interviewees may not be familiar
- ppl of different cultures may be fluent in English but be confounded with abbreviations, colloquialisms, aphorisms, jargon, euphemisms and slang.
- avoid ambiguous words and phrases such as a lot, often, much, large school, or recently discovered ones that have many and vague meanings
- researchers warn against phrasing q's negatively bc they can be misleading and confusing
- a negative answer to a negative worded statement may not be equivalent to the positive answer to a positively worded statement
- even the explanation sounds confusing. they give this example: disagreeing with the statement 'my work is not meaningful' does not necessarily mean that the same individual would have agreed with the statement 'my work is meaningful'
- forcing a respondent to disagree with a negative statement can be confusing. think of the difficulties you have had with negatively phrases multiple choice questions on exams.
- many ppl will fail to hear the word "not" in a question during an interview so those in factor of a statement such as 'the US should not establish diplomatic relations with Iran' might disagree with the statement. and those who actually disagree may answer the same way. Edwards and Thomas warn, "you may never know which is which"
Sample Question Development
- q's evolve as you develop a schedule, particularly for a quantitative survey. here is how a q about texting while driving might evolve during preparation: How do you feel about the state-imposed law against texting and driving?
- take a closer look at this seemingly neutral q. "state imposed" may bias results bc it may sound tyrannical and unconstitutional to some respondents. the openness of the question and the ambiguity of the word "feel" may result in a wide range of answers, some positive and some negative. others may give long answers for and against the law that will create recording and coding nightmares.
- try a second version that closes up the question and eliminates the word "state-imposed"
- are you for, against, or have no feelings about the state's law against texting while driving?
- this version eliminates the potential bias of the first, resolves recording issues, but it may be too closed for qualitative and quantitative purposes.
- develop a third version: do you strongly agree, agree, disagree, or strongly disagree with the state law against texting while driving
- this option assess intensity of feelings, is easy to record and code, leaves undecided as an unstated option, provides instructions for interviewers and includes built in secondary "why" questions to discover reasons for strong approval or disapproval.
- build in secondary q's for reasons, knowledge, level and qualifiers
- work with each question until it satisfies phrasing criteria and is designed to obtain the info needed. careful phrasing avoids confusion and inaccurate results
- less frequent and usually planned in survey interviews. for instance, if a respondents gives an unclear answer, you might ask "what do you mean by that" "how do you mean that?" "what I hear you saying is..."
- if a respondent gives a very brief answer or appears reluctant to elaborate, use a silent probe a nudging probe, or an informational probe such as "tell me more about that"
- if you are unsure a respondent has told you everything of relevance to the question, use a clearinghouse probe such as "anything else?"
- remember to record probing questions and answers carefully, clearly, accurately for later tabulation and analysis.
- your goal is to perform nearly identical survey interviews over and over. one the spot probing may result in interviewer bias if you or other interviewers phrase q's verbally or nonverbally in ways that suggest the answers you prefer or lead different ppl to provide different answers
- if some interviewers ask probing questions and others don't, the amount and type of info attained will differ from one another and result in unreliable data or data that is impossible to tabulate and analyze with a degree of confidence.
- there are five question strategies that enable interviewers to asses knowledge level, honesty and consistency; reduce undecided answers; prevent order bias; and incorporate probing questions
1. Filter Strategy
2. Repeat Strategy
3. Leaning Question Strategy
4. Shuffle Strategy
5. Chain or Contingency Strategy
1. The Filter Strategy
- the filter strategy enables you to determine interviewee knowledge of a topic.
- are you familiar with the water company's proposed rate increase for next year?
- yes I am
- What is the water company proposing?
- if an interviewee says no, go on to the next question. if the interviewee says yes, you ask the person to reveal the extent and accuracy of knowledge.
- this follow-up question may discover that the person is confused or misinformed.
- many interviewees will say yes to bipolar questions even when they have no clue what the interviewer is talking about, to avoid appearing uninformed.
2. Repeat Strategy
- enables you to determine if an interviewee is consistent in responses on a topic, particularly a controversial one.
- you may ask the same question several minutes apart and compare answers for consistency.
- a variation of this strategy is to disguise the question by rephrasing it.
- do you supervise your children's use of computers at home?
- do your children have free access to computers at home?
- another example of repeat strategy is to go from a moderately closed to a highly closed question such as: how often during a week do you drink alcoholic beverages?
- I am going to read a list of how frequently you drink alcoholic beverages each weak. stop me when I read the frequency that best describes your drinking
- do not make the repetition too obvious or close to the initial question and be sure the rewording does not change the intent of the initial question
- repeated questions must be essentially the same to determine consistency in answers
3. Leaning Question Strategy
- respondents may be reluctant to take stands or make decisions, often bc they do not want to reveal their feelings or intentions.
- employ a leaning question, not to be confused with a leading question, to reduce the number of "undecided and "don't know" answers
- the following isa typical leaning question strategy:
9a. do you plan to vote for or against the school referendum in the fall election (if undecided, ask Q. 9b)
9b. which way are you leaning today?
- the "undecided" option remains in question 9b bc an interviewee may be truly undecided at the moment. a variation of the leaning question is "well if you had to vote today, how would you vote?"
- clearly stated "undecided" and "don't know" options may invited large percentages of these answers, particularly when a q asks for criticism of people, organizations or products.
- however, some sources recommend that you always include "don't know" or "not applicable" answer options in all questions, unless all interviewees will have a definite answer, to reduce interviewee frustration and provide the most honest and accurate answers.
- leaning questions urge respondents to take a stand or make a decision
4. Shuffle Strategy
- the order of answer options in questions may affect interviewee responses
- research indicates that last choices in questions tend to get negative or superficial evaluations bc interviewees get tired or bored but that interviewees also tend to select an option bc it is the first mentioned or the last heard.
- the shuffle strategy varies the order of answer options from one interview to the next or prevent order bias
- the method or rotation is carefully explained when training interviewers
- order bias is both fact and myth
- potential order bias has resulted in strange events in political, persuasive, and advertising surveys
- a political candidate in Indiana changed his name legally so it would begin with A. this placed him at the top of the ballot on election day, the belief being that voters select the top names in lists of candidates.
- he lost, but his and similar actions have led states to shuffle names on ballots.
5. Chain or Contingency Strategy
- highly standardized and highly scheduled formats allow allow for preplanned questions that enable you to probe into answers.
- the chain of contingency strategy is illustrated in the following series from a market survey.
- notice the build in instructions and precoding for ease of recording answers and tabulating data
1a. During the past month, have you received any free samples of breakfast cereal?
yes - Ask Q.1b
no - Ask Q.2a
1b. Which breakfast cereal did you receive?
1c. (ask only if great grains is not mentioned in question 1b, otherwise, skip to 1d)
Did you receive a free sample of Great Grains?
yes - ask question 1d
no - ask question 2a
- all probing questions i surveys are included in the schedule
- replicability means to reproduce interviews
- the chain or contingency strategy enables you to probe into answers while maintaining control of the process and ensuring that each interview is as identical as possible
- evaluative interval scales (likert scales)
- frequency interval scales
- numerical interview scales
- a variety of scale questions allow you to delve more deeply into topics and feelings than bipolar questions and to record and tabulate data more easily.
- interval scales provide distances between measures.
- for ex., evaluative interval scales (often called Likert scales) ask respondents to make judgements about persons, places, things, or ideas. the scale may range from 5-9 answer options (five is most common) with opposite poles such as "strongly like....strongly dislike" "strongly agree....strongly disagree" or "very important.....not important at all"
- you may provide respondents with cards (color-coded to tell them apart) for complex questions or ones with many choices or options.
- a card eliminates the faulty-recall problem that respondents experience.
- they can study the answers or objets they are evaluating, rating or tanking without trying to remember all of the options given orally.
- example of card use: please use the phrases on this card to tell me about how the recent TV ads for the new water park has affected your interest in visiting the park
5 increases my interest a lot
4 increases my interest a lot
3 will not affect my interest
2 decreases my interest a little
1 decreases my interest a lot
- cards or color-coded cards provide aid for interviewee recall of answer options
- frequency interval scales ask respondents to select a number that most accurately reflects how often they do something or use something.
- how frequently do you eat pork?
__ more than once a week
__ once each week
__ every other week
__ once or twice a month
__ less than once a month
- frequency scales deal with number of times.
- numerical interval scales ask respondents to select a range or level that accurately reflects their age, income, educational level, or rank in an organization
- I am going to read you several age groupings. please stop when when I read the one that applies to you:
__ 65 and over
- nominal scales deal with naming and selecting
- nominal scales provide mutually exclusive variables and ask respondents to name the most appropriate variable.
- these are self-reports and do not ask respondents to rate or rank choices or to pick a choice along an evaluative, numerical, or frequency continuum.
- choices may be read in any order. for example:
Do you consider yourself to be a:
- sometimes "other" options ask you to write the name of the "other" you choose.
- in nominal questions the options are mutually exclusive and include most likely options from which to choose. "other" is the final option bc the respondent must be able to choose of the named options or provide an option.
- ordinals scales ask for ratings or rankings
- ordinal scales ask respondents to rate or rank the options in their implied or stated relationship to one another.
- they do not name the most applicable option as in interval and nominal scales.
- The following is a RATING ordinal scale
as you have traveled around the country during the past 5 yrs, i'm sure that you have stayed in many hotels. please rate each of the following applicable hotels and motels as excellent, above average, average, below average, poor or not applicable "N/A"
- note that this rating scale generates 6 responses, including not applicable for an unused hotel or motel.
- The following is a RANKING ordinal scale
one this card are the names of 5 news programs. rank order them in terms of accuracy and dependability with one being the highest and 5 being the lowest
ABC world news
CBS evening news
- or ordinal questions can ask respondents to select from among options and rank them in order: pick the three you think are most important and rank them in order of importance to you.
Bogardus Social Distance Scale
- Bogardus scales measure effect of relational distance
- the bogardus social distance scale determines how people feel about social relationships and distances from them
- you want to know if a person's attitude or feeling changes as the issue comes closer to home. this scale usually moves progressively from remote to close relationships and distances to detect changes as proximity narrows.
- for example, you might use the follow Bogardus Social Distance Scale to determine how interviewees feel about expanded oil drilling
1. do you favor expanded oil drilling in the US? yes/no
2. do you favor expanded oil drilling in the midwest?
3. do you favor expanded oil drilling in your state?
4. do you favor expanded oil drilling in your county?
5. do you favor expanded oil drilling in this township?
- in many questions, respondents feel safely removed from the attitude or feeling they are expressing about a product, issue action or person. The Bogardus Social Distance Scale brings an issue ever closer to home so it is no longer something impersonal or one that affects others "over there"
- question scales are designed to obtain a range of results and level 2 and 3 disclosures, but respondents may try to "out psyche" survey takers.
- students do this when taking standardized tests. for ex, respondents may try to pick "normal" answers in nominal and ordinal scales and safe, moderate, or middle options in interval scales.
- rather than admit they don't know the correct answer, even when there is no correct answer, respondents may pick the option that stands out, such as the second in a list that include 10 percent, 15 percent, 20 percent, 30 percent and 40 percent.
- respondents who first agree that a certain activity would make most people uneasy are less likely then to admit ever engaging in that activity and may attempt to change the subject.
- phrase scale q's carefully to avoid game playing, guessing and confusion. listen and observe reactions during pretesting interviews to detect patterns of responses, level of interviewee comprehension, and hesitancy in responding
- long scales, complicated rating or ranking, and lengthy explanations may confused respondents, perhaps without either party realizing it at the time.
- minimize guessing in surveys
- anticipate confusing in scale questions
- question sequences complement question strategies
- the tunnel sequence is useful when no strategic lineup of questions is needed.
- Gallup's quintamensional design sequence, or a variation of it, is appropriate when exploring intensity of attitudes and opinions.
- funnel, inverted funnel, hourglass and diamond sequences include open-ended questions, so answers may be difficult to record, code and tabulate. they are more appropriate for qualitative surveys bc the wealth of information interviewers obtain for open questions is worth the problems involved.
- a study of the effects of question order suggests that general questions should come first followed by more specific questions. this is a funnel sequence.
Defining the Population
- interviewees are the source of your data. the best schedule of questions is of little help if you talk to the wrong people at the wrong time or in the wrong place.
- the first step in selecting interviewees is to define the population you wish to study.
- the population may be small and similar such as members of sales staff or large and diverse such as all employees of an auto plant.
- you may select a subset of a large population such as all employees over age of 50 at a department store.
- the identified population should include all persons who are able and qualified to respond to your questions and about whom you want to draw conclusions.
- if a target population is small (members of a fitness club), you may want to interview all of them.
- most surveys, however, deal with populates that far exceed time, financial and personal limitations - the 35,000 graduate students at a university of residents over the age of 18 in a city of 250,000
- dozens of interviewers could not reach, let alone interview all of these people, so you interview a sample of them and extend findings to all of them.
- a common pitfall is not to spend the time necessary to ensure that the population from which the sample is drawn is complete and well defined
- the fundamental principle is that a sample must accurately represent the population or target group under study
- old-time watermelon sellers practiced this principle when they carefully cut out a triangular plug from a watermelon. this plug represented the entire watermelon
- each potential respondent from a define population must have an equal chance of being interviewed
- you determine the probability that each person might be selected by deciding upon an acceptable margin of error
- the precision of a survey is the "degree of similarity between sample results and the results from a 100 percent count obtained in an identical manner"
- most surveys attain a 95 percent level of confidence, the mathematic probability that 95 out of 100 interviewees would give results within 5 percent points (margin of error) either way of the figures you would have obtained if you had interviewed the entire target population.
- survey results reported in the media routinely state that a survey had a margin of error of 4 percent. this means that if 42 percent of respondents approve of the way congress is doing it's job, the real figure might be as low as 38 percent or as high as 46 percent
- a tolerable margin of error depends on the use of survey results
- if you want to predict the outcome of an election or the effects of a new medical treatment, you must strive for a small margin of error, 3 percent or less.
- if you're conducting a survey to determine how employees feel about a new recreation facility, a higher margin of error is acceptable, 4 or 5 percent.
- a sample is a miniature version of the whole
- margin of error determines the worth of a survey
- a sample is the actual number of persons interviewed.
- determine sample size by the size of the population and the acceptable margin of error.
- some survey orgs produce accurate national surveys with a margin of error in the 3 percent range from a sample of 1,500.
- standard formulas reveal that as a population increases in size, the percentage of the population necessary for a sample declines rapidly. (smaller the population, more ppl needed in sample)
- in other words, you must interview a larger percentage of 5,000 people than of 50,000 people to attain equally accurate results.
- formulas also reveal that you must increase greatly the size of sample to reduce the margin of error from 5 percent to 4 percent to 3 percent.
- the small reduction of the margin of error may not be worth the added costs of conducting significantly more interviews.
- sample sizes necessary for 5 percent margin of error and 95 percent level of confidence: infinity-384, 500,00-384, 100,000-383, 50,000-381, 10,000-370, 5,000-357, 3,000-341, 2,000-322, 1,000-278
- creative research systems offers a sample size calculator as a public service. you may employ this calculator once you know the confidence interval (margin of error of 3, 4 or 5), confidence level (percentage of sureness you have in results such as 95%), and target population size
- size of the sample is important, but you how select the same is of utmost importance to the validity of the survey
- As W. Charles Redding warned years ago, a bad survey is worse than no survey.
- there are two general types of sampling, probability and non-probability.
- in probability sampling, you know that each member of your population has a certain change of being interviewed.
- in non-probability sampling, you do not know the chance that each member of your population will be interviewed.
- there are five common methods of probability sampling (random sampling, table of random numbers, skip interval or random digit, stratified random sample, and sample point), the most accurate method of sampling.
Probability Sampling - Random Sampling
- random sampling is the simplest method of selecting a representative to sample.
- for ex. if you have a complete roster of all persons in a population, you place all names in a container, mix them thoroughly and draw out one name at a time until you have a sample.
- random sampling is like drawing names out of a hat
Probability Sampling - Table of Random Numbers
- a more complicated random sampling method is to assign a number to each potential respondent and create or purchase a table of random numbers.
- with eyes closed, place a finger on a number and read a combination up, down, across, left or right or diagonally.
- select this number as part of the sample or decide to read the last digit of the number touch (46) and the first digit of the numeral to the right (29) and thus contact respondent number 62.
- repeat this process until you have the sample needed.
Probability Sampling - Skip Interval or Random Digit
- in skip interval you select every nth name from a list
- you may choose every 10th number in a telephone book, every fifth name is a roster of clients, or every other person who walks into a supermarket.
- the random digital dialing system now in wide use for conducting surveys randomly generates telephone numbers in target area-code and prefix areas, gives every telephone number int he area an equal chance of being called and ensures anonymity bc no interviewee names are used.
- this common sampling technique may have some built-in flaws.
- for ex. a growing percentage of the population has unlisted phone numbers or uses cell phones.
- on the other hand, a growing number of households have more than one telephone number and this increases the probability that a particular household may be contacted more than once.
- a voter, customer or member roster may be outdated.
- time of day, day of the week and location may determine the type of ppl available for interviews.
Probability Sampling - Stratified Random Sample
- a stratified sample most closely represents the whole.
- random sampling procedures may not provide adequate representation of subgroups within a population
- if a population has clearly definable groups (males and females, ages, education, income levels, and diverse cultural groups), employ a stratified random sampling method.
- this method allows you to include a minimum number of respondents for each group, typically the percentage of the group in the target population.
- for instance, if a targeted population consists of 30 percent first years, 25 sophomores, 20 juniors, 20 seniors, and 5 graduate students, your sample would represent these percentages.
Probability Sampling - Sample Point
- a sample point is usually a geographical area
- a sample point represents a geographical area (a square block or mile, for ex.) that contains specific types of persons (grain farmers or retired ppl).
- instructions may tell interviewers to skip corner houses (corner houses are often more expensive) and then try every other house on the outside of the flour-block area until they have obtained two interviews with males and two with females.
- the US department of agriculture uses aerial photos of farm areas and crops to determine which famers to interview to determine the amounts of various crops planted and possible yields of these crops each year.
- the sample point or block sample gives the survey designer control over selection of interviewees without resorting to lists of names, random digits and telephone numbers
Non-Probability Sampling - Self-Selection
- there are two common forms of non-probability sampling methods, the LEAST accurate form of sampling. survey interviewers employ them bc they are convenient and inexpensive (Self-Selection and Convenience)
- self-selection is the least representative of sampling methods
- most inaccurate sampling method
- you see this voluntary method use nearly every day in radio and tv talk shows, newscasts and on the web
- who is most likely to call C-SPAn, Rush Limbaugh, or a TV station? you guessed it-those who are very angry or most opposed to/in favor of an action
- moderates rarely call or write
- it's easy to predict how self-reporting surveys on gun control, health care reform and labor unions will turn out.
Non-Probability Sampling - Convenience
- popular bc respondents are numerous and easy to reach.
- ex. are interviewers stopping students as they exit a classroom building, shoppers as they walk through a mall, or ppl as they walk down the street
- the only criterion for selection is convenience for the interviewer
- randomness and representation of diverse elements of a target population build into probability forms of sampling are not considered.
Selecting and Training Interviewers
- creating a survey instrument and developing a careful sample of interviewees are critical, but so is selecting interviewers and training them to conduct interviews properly.
- you can rarely do it all by yourself
- if you plan to interview a small # of ppl and the interviews will be brief, one interviewer may work
- most often you will need several interviewers, particularly when interviews will be long, the sample is large, time is limited, and interviewees are scattered over a large geographical area
- large and difficult interviewing assignments result in serious interviewer fatigue and decline in motivation; both will reduce the quality of interviews and data received.
Selecting and Training Interviewers
- a highly scheduled, standardized interview doesn't require the interviewer to be an expert on the topic or skilled in phrasing questions and probing into answers
- it does require a person who can learn and follow guidelines, read the q's verbatim and effectively, and record answers quickly and accurately.
- if you are using a high scheduled interview format that requires skillful probing into answers, interviewers must have the ability to think on their feet, adapt to different ppl, handle unanticipated objections and concerns, and react effectively and calmly to strange answers.
- in this type of interview, professionally trained interviewers are more efficient and produce more accurate results
- experienced interviewers obtain higher rates of acquiescent reports than to inexperienced interviewers, even after accounting for potential differences in interviewer and respondent characteristics. these differences across interviewers are not mediated by differential pace of the interview, as measured in interview length, implying that there may be differences in interview behaviors for experiences and inexperienced interviews.
- interviewers must follow the rules
Selecting and Training Interviewers
- interviewers who are older, have a nonthreatening demeanor, or have an optimistic outlook get better response rates and cooperation, regardless of their experiences.
- age generates credibility and self confidence, optimism motivated interviewees to cooperate.
- one study showed that personality and attitude of the interviewer are the most important elements in shaping interviewee attitude toward surveys
Selecting and Training Interviewers
Personal Characteristics - Personal Skepticism
- nearly 1/3 of respondents believe that answering survey q's will neither benefit them nor influence decisions, that there are too many surveys, that surveys are too long, and that interviewers ask too many personal q's.
- some 36% one respondents in one study said they had been asking to take part in "false surveys" sales or political campaign interviews disguised as informational surveys.
- the public's questions indicate a healthy dose of skepticism about polling. their q's are usually accompanied by a strong and sincere desire to find out what's going on
- clearly survey interviewers must be aware of relational dimensions such as warmth, involvement, dominance, and trust and make every effort to establish a positive relationship with each respondent by appearing to be friendly, relaxed and trustworthy
- interviewees are increasingly wary of surveys
Selecting and Training Interviewers
Personal Characteristics - Similarity of Interviewer and Interviewee
- similarity, but not a mirror image, may be important.
- similarity is an important relational dimension in survey interviews
- you should dress similar to interviewees bc if interviewers look like me, i am more likely to cooperate and answer appropriately.
- an in-group relationship with the interviewee (black to black, senior to senior, hispanic to hispanic) may avoid cultural and communication barriers and enhance trust bc the interviewer is perceive to be safe, capable of understanding and sympathetic
- it may be essential that interviewers can speak to interviewees in their own language, including dialects or regional differences
- poor execution can undo thorough preparation
- conduct training sessions with carefully written instructions for all interviewers, regardless of experience
- training results in greater use of appropriate probing q's, feedback and giving instructions
- discuss common interviewee criticisms of surveys and stress the importance of following the question schedule exactly as printed.
- explain complex q's and recording methods
- be certain interviewers understand the sampling techniques employed
- emphasize the need to replicate interviews to enhance reliability and attain an acceptable margin of error and level of confidence
- describe the entire process and purpose discuss the nature and danger of interview bias. you may need to assist interviewers in reading maps and identifying households.
- rehearse the interview, including the opening, asking and recording of q's, which is critical if probing questions are included or interviewers need to create them on the spot, the closing of the interview.
- the following are typical instructions for interviewers:
Preparing for the Interview
- guard against interviewer bias
- study the question schedule and answer options thoroughly so you can ask rather than read questions and record answers quickly and accurately.
- dress appropriately, and be neat and well-groomed.
- do not wear buttons or insignia that identify you with a particular group or position on the issue to avoid biased responses
- choose an appropriate time of week and day
Conducting the Interview:
- be friendly, business like, and interested in the topic. speak clearly, at a good pace, and loudly enough to be heard easily. maintain eye contact and don't be afraid to smile.
- ask all questions clearly, without hesitation, and naturally
- adopt and informal speaking manner that avoids the appearance of reading or reciting openings, questions and closings
Opening the Interivew
- motivate the interviewee from the moment the interview commences.
- state your name, identify organization, and present you credentials if appropriate
- explain purpose, length, nature, and importance of study; then move to first q without appearing to pressure the interviewee to take part
- no question can be altered in any way
- ask all q's including answer options exactly as worded.
- you may repeat a q but not rephrase it or define words
- don't change the order of questions or answer options unless instructed to do so
- if you are doing a qualitative study, probe carefully into answers to obtain insightful and thorough answers free of ambiguities and vague references
Receiving and Recording Answers
- give respondents adequate time to reply and then record answers as prescribed in your training and one the schedule.
- write or print answers carefully. remain neutral at all times, reacting neither positively nor negatively to responses
- maintain a pokerface
Closing the Interview
- when you have obtained the answer to the last q, thank the person for cooperating and excuse yourself without being abrupt
- be polite and sensitive, making it clear that the interviewee has been most helpful. don't discuss the survey with the interviewee.
Conducting Survey Interviews:
Pretesting the Interview
- with preparation completed, it's time to pretest the interview with a portion of the targeted audience to detect potential problems with question and answer options
- lack of pretesting invites disaster
- the best plans on paper may not work during real interviews
- try out the opening, questions, recording answers and closing
- leaving nothing to chance
- may help you discover which questions to eliminate or adjust, or help you realize you need to probe into something
- researcher discovered that scale questions tended to confuse elderly respondents, so he added special explanations to complex questions
- make sure questions are understood and wording is clear, ppl don't react negatively, you can collect desired info, recording went smoothly, etc.
- leave nothing unquestioned
- once you have studied the pretest results and made alterations in procedures, questions and answer options, you are ready to conduct the full survey
- ideally, the survey interview would be conducted in F2F personal interview that obtains a good response rate bc of the personal touch and interviewees can see and hear the interviewer and see, feel, touch, experience and taste products.
- the F2F interview has a number of advantages over other forms of survey taking.
- easier for interviewer to establish credibility through physical appearance, dress, eye contact, presentation of credentials
- interviewers can be sure of interviewing target responses including marginalized populations in specific places at specific times.
- respondents are more willing to take part in lengthy interviews that enable interviewers to ask more questions on complex issues and focus on in-depth attitudes and information
- allow interviewers to observe attitudes facial expressions, eye contact, gestures and posture
- ppl are more likely to provide self-generated and more accurate answers bc of the naturalness of the situation
- disadvantages: expensive, time consuming, slow, requires good trained interviewer who may or may not interview the representative sample the survey requires
- may be impossible to do F2F interviews over a wide geographical area
- societal changes have made it hard to predict when ppl might be available
Interviewing by Telephone
- may be inexpensive in money but costly in results
- particularly with the advent of Random-Digit Dialing technology; has become dominant
- studies say telephone & f2f produce similar results but researchers urge caution when selecting interview methods
- one study showed many interviewers don't like phone interviews. one found that fewer interviewees particularly older ones prefer the telephone
- ppl feel uneasy about talking ab sensitive issues with strangers that can't see, and its hard to make convincing confidentiality guarantees when they are not face to face with ppl
- ppl can filter out unwanted calls, including surveys
- cellphones create problems bc can't use phone book to retrieve cell numbers, only landlines
- this has led researchers to warn against coverage bias when cell phone only respondents are excluded (usually young ppl)
- they have become dominant bc of significant changes - inexpensive, fast results, overnight if desired, fewer interviewer bias bc increased uniformity in manner and deliver and no effect from appearance, expressions and eye contact
- interviewers feel safer, ppl provide fewer socially unacceptable answers and prefer anonymity
Opening Telephone Interview
- most refusals in phone surveys occur prior to the first substantive question: 1/3 in opening seconds, 1/3 in orientation, 1/3 at point of listing household members.
- speaking skills (pitch vocal variety loudness rate and distinct enunciation, especially during the opening, are more important than content.
- respondents react to cues communicated by the interviewer's voice and may grant or refuse an interview on that basis
- phone interviewers must establish trust through verbal and vocal analogs to the personal appearance, credentials and survey material that enhance trust in f2f interviews
How to Use the Telephone
- don't give a person reason or opportunity to hang up.
- develop an informal but professional style that is courteous, not demanding, and friendly, not defensive. get the interviewee invovles as quickly as possible in answering questions bc active involvement motivates ppl to take part and cooperate
- listen carefully and actively. give your undivided attention to what the interviewee is saying by not drinking, eating, sorting papers or playing with objects on your desk.
- don't communication nonverbally with others in the room and saying nothing you do not want the interviewee to hear even if you believe you have the mouthpiece covered.
- explain any pauses or long silences of more than a few seconds or signal you are listening with cues such as uhuh, yes, okay
- use your voice effectively. talk directly into the mouthpiece, speak loud enough, slowly, clearly and distinctly bc the interviewee must rely solely on your voice.
- state each answer option with vocal emphasis on important words and pause between each option to aid in comprehension and recall
- use a computer assisted telephone interview system that allows you to dial random numbers quickly and to compile results within minutes of completing interviews.
- opening the telephone interview is critical
- do nothing but ask and listening during phone interviews.
Interviewing through the Internet
- increasing number of survey interviews take place via the internet
- email, webpages, and computer direct
- substantially less expensive and faster than face to face to and phone interviews
- survey posted on popular website may generate thousands of responses within hours
- bc they are highly flexible, they can target large populations over great distances
- a major problem of survey interviews - interviewees attempting to give socially acceptable answers - is lessened bc of the anonymity and perceived safety of the internet interview
- interviewer bias, which is a significant concern in telephone and face to face interviews, is not a problem
- ppl give more honest answers to sensitive topics.
- ppl tend to provide more detailed answers to open ended questions, perhaps bc it is easy and quick to type long answers and they can reply when it suits them
- critical nonverbal communication that aids face to face and phone interviews is lost when you use the internet
- response rates may suffer bc it is more hard to establish credibility of the survey and its source or to distinguish survey interview from a slick telemarketer sales interview
- while the internet gives ppl time to think through answers, it may lose the spontaneity of interactions it f2f and phone interviews; they are basically electronic bulletin boards
- however, real time chat software can ensure spontaneity
- it's nearly impossible to probe into answers or employ question strategies such as shuffle, leaning and repeat.
- evidence indicates that completion rates are lower for long surveys, respondents grow bored or tired and log off
- hard to target specific audiences you wish to sample in wide-ranging internet surveys and you may not know who in a family, school or state replied to your survey
- your sample and results may be highly questionable
- those who feel strongly about an issue usually with negative attitudes may overwhelm the results in self-selected internet surveys
- there is no doubt that the unrepresentativeness of current internet access remains the greatest problem for data collection online
Coding and Tabulation
- final phase of the survey invovles coding, tabulation and analysis of info received
- begin final phase of the survey by coding all answers that we not precoded, usually for open-ended questions.
- if question 20 elicits a wide variety of answers and the question is coded #20, each answer can be coded 20 plus 1, 2, 3, 4, etc. like 20-1, 20-2, 20-3
- answers to open ended q's may require analysis and structuring before developing a coding system.
- you may need to make categories before coding and code based on the categories
- record answers to open ended questions with great care
- analysis is making sense of your data/information
- once all the answers are coded and the results are tabulated, the analysis phase begins
- can be overwhelming. lots of information
- be selective. ask "what findings are likely to be most useful?" and what will i do with this info once i get it. if you have no idea, do not ask for it.
- capitalize on the potential of data. subject data to comparative breakdowns to discover differences between demographic subgroups
- dig for the gold. what is the really important stuff hidden within raw data and simple tabulations?
- look for what's missing. what you do not find may be more important that what do you do find what info did you not obtain?
- during analysis of data, ask these questions: what conclusions can you draw and with what certainty? what segment of a target population can you generalize? what are the constraints impose by the sample, schedule of q's, interviewing process, and interviewers? why did ppl respond in specific ways to specific q's? what unexpected events or changes have occurred since the completion of the survey that might make this survey dated or suspect?
- caution is essential fora ll survey takers. journalists must be cautious when writing headlines and making predictions, orgs must be cautious when basing policy decision on survey results. voters must be careful when casting votes according to candidate preference polls.
- when the analysis of data is complete, determine if the purpose and objectives of your survey are achieved. if s, what are the best means of reporting the results
- know the limitations of your survey
- be careful in using survey results
The Respondent in Survey Interviews:
- people are often involved in highly structured interviews.
- rarely compulsory, you can just say not but when you do you may forfeit a chance to influence an important decision that affect you your family etc.
- the opening: take an active part in the opening and discover through observing, listening, and asking q's the identify of the interviewer, credentials, org, purpose, why and how you were chosen, length, how info will be used, and confidentiality.
- become thoroughly oriented prior to answering q's.
- if interviewer doesn't provide important info, ask for it.
- the opening minutes enable you to determine if the interview is a survey or a slick persuasive interview under the guise of a survey.
- listen to each q carefully, to answer options in scaled interviews
- if a q or option is hard to recall, ask to repeat the q slowly
- if something is unclear, and why and ask for clarification.
- avoid replays of earlier answers, especially if you think you "goofed"
- don't try to guess what a q is going to be from first words
- respond clearly and precisely
- answer should represent beliefs attitudes and actions
- don't fall into interviewer bias
- you can refuse to answer poorly constructed leading q's or to give specific data you find invasion of privacy
- expect and demand tactful sensitive and polite treatment from interviewers
- ask for adequate time to answer q's
- remind interviewer about agreement of time limit if there is one
- survey interviews can be fun interesting and informative if both parties treat one another fairly