Conducting a Survey: Key words
& terms
Week 1: Lecture and Doing Surveys Online chapters 1
and 3
Key word / term Definition
Administrative data Data provided by persons or organizations for regulatory or other
government activities.
Transaction data Data generated as an automatic byproduct of transaction and activities.
Social media data Data created by people with the express purpose of sharing with others.
Sensor data Data generated by apps on, e.g., smartphones.
GIGO Garbage In Garbage Out: if you did not design your survey well, you will
get bad results.
Address-based Sampling method that uses addresses as the sampling frame.
sampling
CAPI Computer assisted personal interviewing.
CATI Computer assisted telephone interviewing.
CASI Computer assisted self-interviewing (no interviewer present).
Coverage error An error of non-observation in which there are discrepancies between
the frame population and the actual target population.
Digital divide The discrepancy that people with Internet access differ from those
without Internet access.
Frame population A list or another mechanism of the target population used to draw the
sample.
Intercept surveys Survey in which every ‘nth’ person that visits a particular website is
invited to participate in the survey.
Internet penetration The percentage of the target population with Internet access.
Item non-response Individuals who fail to provide the answer to an individual question or
error item.
List based samples A sample based on a list of possible respondents that are invited to
complete the survey.
Measurement error The deviation of the answers retrieved from the respondents from their
true value.
Modes Different ways of conducting a survey, such as online, face-to-face, and
paper-and-pencil. Modes can differ in five dimensions: interviewer
involvement, interaction with the respondent, privacy, channels of
communication, and use of technology.
Non-probability Non-random sampling is used, some people have zero chance of being
surveys involved in the survey. Three types: polls as entertainment, unrestricted
self-selected surveys and volunteer opt-in panels.
Non-response error Three types of error due to the missing of answers: unit non-response
error, item non-response error, and partial non-response error. Only
leads to error when the non-respondents differ from the respondents in
features that are central to the questionnaire.
Partial non-response Individuals who drop out during the survey.
error
Polls as Mostly unscientific Internet polls, like a ‘question of the day’ poll. An
1
, entertainment example of a non-probability survey.
Pre-recruited panels A panel that tries to obtain a probability sample of the full population:
of the full population non-Internet users are provided with alternative equipment.
Pre-recruited panels A panel consisting of participants recruited via random digit dialling or
of Internet users address-based sampling.
Probability-based Surveys in which participants are recruited randomly and in which all
surveys members of the target population have a chance to be selected. Five
subtypes: intercept surveys, list-based samples, web option in mixed-
mode surveys, pre-recruited panels of Internet users, and pre-recruited
panels of the full population.
Qualitative research Investigating a research problem by collecting (often textual) data in
order to understand underlying reasons and motivations. No
representative sample needed, useful in small-scale research. Three
different methodologies: passive analysis of Internet data, active analysis
of Internet data (researcher gets involved in the conversations) and web
surveys.
Quantitative research Investigating a research problem by generating numerical data and using
that data for statistical testing. Useful in large-scale research, close ended
questions.
Random digit dialling Sampling method that uses random dialling of telephone numbers to
recruit respondents.
Sampling error Occurs when the respondents in the sample do not accurately represent
the frame population.
Social desirability bias Respondents giving social desirable answers rather than the truth. A
downside of surveys where the interviewer is present.
Total survey error A conceptual framework which describes statistical error properties of
sample surveys, and which attempts to display a balance of costs and
errors at the design stage of the survey. Survey researchers should
choose the design that brings the smallest total survey error while taking
cost of that design into account. Four fundamental concepts to design a
good survey are: accurate measurements, high response rates, high
coverage, and proper sampling.
Unit non-response Individuals not participating in the survey.
error
Unrestricted self- Surveys that do not use random sampling. An example of a non-
selected surveys probability survey.
Volunteer opt-in A large database of potential participants that are willing to fill out
panels surveys on a voluntary basis. An example of a non-probability survey.
Web option in mixed A survey in which completion on the web is seen as one of the mode
mode surveys options.
Client-side paradata A type of paradata which contains observations about the behaviour of
the participants on the web page during completion of the survey. If
researchers collect this data without informing the participant, they act
against the principles of informed consent.
Confidentiality The right participants have to know who can access their data. Refers to
the way data are treated.
Data fabrication A way of breaching publication ethics: fabricating and reporting on non-
existing data.
Data falsification A way of breaching publication ethics: manipulating research methods or
results.
Debriefing Has three functions: the ethical function (reveal the true purpose of the
2
& terms
Week 1: Lecture and Doing Surveys Online chapters 1
and 3
Key word / term Definition
Administrative data Data provided by persons or organizations for regulatory or other
government activities.
Transaction data Data generated as an automatic byproduct of transaction and activities.
Social media data Data created by people with the express purpose of sharing with others.
Sensor data Data generated by apps on, e.g., smartphones.
GIGO Garbage In Garbage Out: if you did not design your survey well, you will
get bad results.
Address-based Sampling method that uses addresses as the sampling frame.
sampling
CAPI Computer assisted personal interviewing.
CATI Computer assisted telephone interviewing.
CASI Computer assisted self-interviewing (no interviewer present).
Coverage error An error of non-observation in which there are discrepancies between
the frame population and the actual target population.
Digital divide The discrepancy that people with Internet access differ from those
without Internet access.
Frame population A list or another mechanism of the target population used to draw the
sample.
Intercept surveys Survey in which every ‘nth’ person that visits a particular website is
invited to participate in the survey.
Internet penetration The percentage of the target population with Internet access.
Item non-response Individuals who fail to provide the answer to an individual question or
error item.
List based samples A sample based on a list of possible respondents that are invited to
complete the survey.
Measurement error The deviation of the answers retrieved from the respondents from their
true value.
Modes Different ways of conducting a survey, such as online, face-to-face, and
paper-and-pencil. Modes can differ in five dimensions: interviewer
involvement, interaction with the respondent, privacy, channels of
communication, and use of technology.
Non-probability Non-random sampling is used, some people have zero chance of being
surveys involved in the survey. Three types: polls as entertainment, unrestricted
self-selected surveys and volunteer opt-in panels.
Non-response error Three types of error due to the missing of answers: unit non-response
error, item non-response error, and partial non-response error. Only
leads to error when the non-respondents differ from the respondents in
features that are central to the questionnaire.
Partial non-response Individuals who drop out during the survey.
error
Polls as Mostly unscientific Internet polls, like a ‘question of the day’ poll. An
1
, entertainment example of a non-probability survey.
Pre-recruited panels A panel that tries to obtain a probability sample of the full population:
of the full population non-Internet users are provided with alternative equipment.
Pre-recruited panels A panel consisting of participants recruited via random digit dialling or
of Internet users address-based sampling.
Probability-based Surveys in which participants are recruited randomly and in which all
surveys members of the target population have a chance to be selected. Five
subtypes: intercept surveys, list-based samples, web option in mixed-
mode surveys, pre-recruited panels of Internet users, and pre-recruited
panels of the full population.
Qualitative research Investigating a research problem by collecting (often textual) data in
order to understand underlying reasons and motivations. No
representative sample needed, useful in small-scale research. Three
different methodologies: passive analysis of Internet data, active analysis
of Internet data (researcher gets involved in the conversations) and web
surveys.
Quantitative research Investigating a research problem by generating numerical data and using
that data for statistical testing. Useful in large-scale research, close ended
questions.
Random digit dialling Sampling method that uses random dialling of telephone numbers to
recruit respondents.
Sampling error Occurs when the respondents in the sample do not accurately represent
the frame population.
Social desirability bias Respondents giving social desirable answers rather than the truth. A
downside of surveys where the interviewer is present.
Total survey error A conceptual framework which describes statistical error properties of
sample surveys, and which attempts to display a balance of costs and
errors at the design stage of the survey. Survey researchers should
choose the design that brings the smallest total survey error while taking
cost of that design into account. Four fundamental concepts to design a
good survey are: accurate measurements, high response rates, high
coverage, and proper sampling.
Unit non-response Individuals not participating in the survey.
error
Unrestricted self- Surveys that do not use random sampling. An example of a non-
selected surveys probability survey.
Volunteer opt-in A large database of potential participants that are willing to fill out
panels surveys on a voluntary basis. An example of a non-probability survey.
Web option in mixed A survey in which completion on the web is seen as one of the mode
mode surveys options.
Client-side paradata A type of paradata which contains observations about the behaviour of
the participants on the web page during completion of the survey. If
researchers collect this data without informing the participant, they act
against the principles of informed consent.
Confidentiality The right participants have to know who can access their data. Refers to
the way data are treated.
Data fabrication A way of breaching publication ethics: fabricating and reporting on non-
existing data.
Data falsification A way of breaching publication ethics: manipulating research methods or
results.
Debriefing Has three functions: the ethical function (reveal the true purpose of the
2
