Colleges van het vak Toepassing van onderzoeksmethoden en statistiek

College 1 correlationeel
Qualitative research: study people in natural environment; holistic approach; interviews,
focus groups, text analysis.
Correlational research: quantitative; relationships between variables; causal study difficult.
Experimental research: researcher manipulation; experimental and control group:
randomization of respondents into groups; quantitative measures; suitable for causal
research.
Different types of data, for example:
customer satisfaction, political polls,
governmental statistics.
Data are generated in different ways:
incidentally (organic, ready make) and
purposively (designed, custom made).
We design a study and collect data to:
describe the social reality, study
(causal) relationships, generalize to the
target population.
Inferential goals: description, causation, prediction.
Survey modes: face-to-face (CAPI), mail, telephone (CATI), internet, mixed-mode.
CAPI  computer assistant personal interview
CATI  computer assistant telephone interview
Mixed-mode  two out of four other modes are combined.
Differences between the modes: degree of interviewer involvement; degree of interaction
with the respondent; degree of privacy; channels of communication: visual, auditive;
technology use.
Types of mixed-mode design:
- One mode for some respondents, another mode for others (e.g. online survey with
mail component for those without internet).
- One mode for recruitment, another for survey administration (e.g. mail invitation for
an online survey).
- One mode for data collection, another for reminders, follow-up (e.g. telephone
reminders for an online survey).
- One mode for main part of the interview, another for some subset of answers (for
example sensitive items).
- One mode for one wave of the panel survey, another for others (e.g. first wave face-
to-face, following waves online to save costs).
Cross-sectional survey  enquête die op één moment wordt afgenomen.
Panel survey  interview respondents over time (content usually the same, but may differ).
- Advantages: we can assess within-person change & causality; we can disentangle
age, period, and cohort effects.
Cohort effect: bijvoorbeeld een jaarlaag, die in hetzelfde jaar gestart zijn.

, - Potential errors: attrition (drop-out or consecutive wave-nonresponse); panel
conditioning (learning effects).
Learning effects: bijvoorbeeld weten welk antwoord je altijd invult op
dezelfde vragenlijst.
Dekkingsfout/coverage error  fout die optreedt
doordat een onvolledige lijst van de populatie gebruikt
wordt.
Steekproeffout/sampling error  op basis van toeval
treedt er een groot verschil op tussen de parameter van
de populatie met de statistiek van de steekproef.
Nonrespons fout  fout die optreedt doordat
geselecteerde proefpersonen deelname weigeren.
Reasons: technical difficulties; lack of
motivation/interest; lack of trust with sensitive
questions.
Unit-nonresponse  people chosen for the sample refuse to cooperate with the
survey request entirely.
Item-nonresponse  people chosen for the sample refuse to answer a certain
question of the survey.
Nonresponse bias  systematic differences between respondents and non-respondents on
key study variables.
Adjustment error  because of coverage and nonresponse, statistics based on respondents
may depart from those of target population:
- e.g. underrepresentation of some groups (urban response rates lower than rural
response; new households/new employers in the frame). Adjustment: “weighting” up
the underrepresented.
- e.g. nonresponse to some questions. Adjustment: imputation.
Measurement error  occurs if respondents provide inaccurate answers due to: survey
mode effects; poor question wording; aspects of their behavior.
Processing error  occurs between receiving response of the respondent and publishing
statistics. Examples: data entry mistakes, coding/recoding errors, errors with handling
confidential information, calculation errors.