The survey Broad-based information on a population
“Getting the lay of the land”
Surveys A social science survey is a research method where a number of people answer a fixed set of questions concerning their knowledge, attitudes and behaviors The people are usually chosen carefully to represent a larger population Question topics and wording are crucial to the success of the method
Uses of surveys Surveys are mainly used to: Gain an understanding of the characteristics of populations Public opinion Media use Population demographics, etc. Develop and test theories concerning relationships among variables in large populations
Self-report measures Depending on respondents to provide information about themselves has strengths and weaknesses as a data-gathering method Respondents may be the only source for much of the information you want to know Knowledge, experiences, etc. A number of biases occur in people’s provision of information about themselves
Census v. sample survey When you conduct a census you measure every member of a population When you conduct a sample survey you measure a subset of the population Sample surveys are used to estimate what a census would have found
Census U.S. census Course evaluations US News survey of US colleges
The sample survey Most telecommunications surveys are sample surveys because populations researchers study are usually very large a census would be very expensive and inefficient
Polls Polls usually are short surveys looking to determine public opinion on a topic of current interest Usually are not used to develop more advanced theory Often look to identify opinions of subgroups (demographic clusters)
Sample surveys in telecommunications: Nielsen ratings and surveys Arbitron ratings and surveys Gallup polls Simmons Experian gfkMRI
Pew Research Center surveys
Survey validity
How do you collect the data? Personal interviews Phone interviews Mail interviews Computer-mediated interviews
Why choose one over the other? Cost Response rate Respondent need for guidance Anonymity/confidentiality Speed Control over data collection
Personal interviews
An interviewer asks the respondent a number of questions face-to-face
Source:
Advantages of the personal interview Interviewer can monitor respondent’s answers Interviewer can react to nonverbal cues Survey can include visual stimuli Movie posters, videos, etc. Interviewer can probe for deeper answers Response rates are high
High level of control over the interview situation Respondent identity Interviewer can prevent input from other people in household, noise and other distractions May be the only way to reach certain populations Homeless LGBT Undocumented aliens
Disadvantages of personal interviews Expensive Slow Supervision of interviewing staff is difficult Significant potential for interviewer bias
Telephone interviews An interviewer asks questions of the respondent over the phone Very common method Computer-automated dialing Computer-aided interviewing Large facilities with multiple interviewing stations
Advantages of telephone interviewing Moderate cost Can be carried out quickly Supervising interviewers is relatively easy Interviewers can help respondents with their questions and concerns High response rate Callbacks are relatively easy Personal touch (human voice)
Disadvantages of telephone interviews Less control over the interview situation Cannot use visuals No face contact Respondents get bored quickly Probes, depth limited
Disadvantages of phone interviews Response rate is lower than with personal interviews Unlisted numbers Wrong numbers (turnover is rapid) Class attempt to use UK phonebook was a disaster Cell phones Refusals Not-at-homes (answering machines)
Questionnaires v. interviews Questionnaires are presented to the respondents, who fill them out themselves Distribution can happen in a wide variety of ways Product warranties Restaurants Doctors’ offices Magazines Blogs
Mail-distributed questionnaires A questionnaire is sent through the mail, self-administered Used for radio/tv diaries, disks with ads on them, product warranty cards, political polls by representatives
Advantages of mail questionnaires Low cost Wide sample possible No field staff to manage Confidentiality No interviewer bias Respondent is not rushed, can answer questions at her leisure Can include limited graphics
Disadvantages of mail questionnaires Low response rates May be biased in favor of those interested in topic Respondents must interpret questions without help available Complicated questions cannot be asked (nor can extensive probes be used) No ability to be certain the respondent is who he says he is
Disadvantages of mail questionnaires Respondents must be literate in the language on the survey U.S. has a high adult illiteracy rate English may not be the respondent’s first language Slow response May take weeks or even months
Nielsen’s diary
Other distribution systems for paper questionnaires Handed out at worksites, doctor’s offices, etc. Group administration Targeted audiences (not random) Efficient High response rate
Internet-based questionnaire distribution The survey is distributed in computer file form, either to list of people via or administered to those who visit a website Popularity rapidly rising May be too popular—people are simply ignoring
Advantages of Internet-distributed questionnaires Very inexpensive Data can automatically be included into the database without inputting Skip patterns can be programmed in Data can be collected quickly Audiovisual materials can accompany the questionnaire
Disadvantages of computer- mediated questionnaires Sample bias Many people have limited Internet access Upscale homes, workplaces Internet access is not the same as use Little help is available for respondent Little control (skipped questions, ‘help’ from others during response, etc.)
Disadvantages of computer- mediated questionnaires Respondent self-selection Low response rate Multiple response Groups may want to influence survey outcome, though this is relatively rare
PersonalTelephon Computer Cost Response rate Control Flexibility Speed Help Confidentiality Rapport Sample bias