://share.osf.io/trove/index-card/c7a7c5f1-b084-4b5c-8a4f-db22a23aa63c

trove:Indexcarddcat:CatalogRecord
more details...
trove:focusIdentifier
https://osf.io/8xajd
https://doi.org/10.17605/OSF.IO/8XAJD
dcterms:issued
rdf:type
type
dcterms:modified
trove:resourceMetadata

://osf.io/8xajd

osf:Registration
VacWeight
more details...
foaf:isPrimaryTopicOf
is primary topic of
dcterms:conformsTo
dcterms:created
2021-09-26
dcterms:rightsHolder
Hannah Bucher
dcterms:modified
2021-09-30
dcterms:subject

://api.osf.io/v2/subjects/584240da54be81056cecac48

skos:Concept
Social and Behavioral Sciences
more details...
skos:prefLabel
Social and Behavioral Sciences
rdf:type
type
skos:inScheme
dcterms:dateCopyrighted
2021
dcterms:title
VacWeight
dcterms:publisher
dcat:accessService
dcterms:identifier
https://doi.org/10.17605/OSF.IO/8XAJD
https://osf.io/8xajd
osf:archivedAt
dcterms:description
The popularity of online surveys has increased the prominence of using sampling weights to enhance claims of representativeness. Since these surveys are often based on a nonrandom selection of respondents, model-based approaches are used to correct for nonresponse and coverage bias in online surveys (Valliant & Dever, 2018). Recent evidence suggests that inference from nonprobabilistic online surveys highly depends on the models as well as the variables used in these model-based approaches (Schonlau, van Soest, and Kapteyn 2007). However, the auxiliary variables frequently used for weighting adjustments are often limited to a small subset of sociodemographic variables, such as age and level of education (ZITAT). Some research has carried out that using these variables to calculate sampling weights hardly contributes to minimize undercoverage and nonresponse bias in online surveys (ZITAT). One proposed explanation for these weights often not performing well in reducing nonresponse and noncoverage bias is the poor coherence of the variable used to calculate s adjustment weights with the survey variables on the one hand, and noncoverage or nonresponse on the other (Little 1986; Little & Vartivarian 2005; Valliant & Dever 2018; Valliant, Dever & Kreuter 2018). However, there are often no other variables with known distributions in the target population available to calculate survey weights that are also collected in surveys. A recent study by Peytchev, Presser & Zhang (2017) claims the deliberate inclusion of substantive survey questions that have dependable external benchmark estimates (Peytchev, Presser & Zhang, 2017). The results of their study show a substantial reduction of nonresponse bias when adjustment weights are calculated using information on voting and volunteering (Peytchev, Presser & Zhang, 2017). The present study builds on these findings. It aims at answering the research question whether using COVID-19 vaccination status, a substantive survey question for that an external benchmark exists, improves the effectiveness of adjustments weights in reducing nonresponse and coverage biases in online surveys. Data for this study is collected by the GLES (German Longitudinal Election Study) Study. One survey of this study relies on a nonprobabilistic selection of respondents from an opt-in online panel, based on crossed quotas for age (categorized, 5 groups), gender and eductaional status (categorized, 3 groups). . It yields to capture political attitudes and voting behaviourfor the German federal election in September 2021. Information on the COVID-19 vaccination status of the respondents is collected in September/October 2021 (completely vaccinated, partial vaccinated and not vaccinated) of the respondents. The benchmark data on COVID-19 vaccination status in September/October 2021, which we used in calculating the adjustment weights, was provided by the Robert Koch Institute, the government’s central scientific institution in the field of biomedicine (www.rki.de). Notably, the information on vaccination status in the population is available federal state and age group. We hypothesize that including COVID-19 vaccination status in calculating the weighting factors helps reducing nonresponse and coverage biases in our sample and improves the accuracy of point estimates. This assumption is based on the result of initial data analyses. These analyses indicated that vaccination status correlates with several variables on voting behavior as well as political attitudes. Therefore, one criterion, namely the correlation of the weighting variable with variables of interest is met. Moreover, initial data analysis also showed a correlation of vaccination status and the respondents’ intention to cast a vote. Since non-voters are commonly underrepresented in election studies (ZITAT), we assume an (at least) indirect relation of vaccination status and survey participation. However, the validity of this assumption has not been tested empirically so far. To evaluate whether the calculated weights that include vaccination status as weighting variable helps to reduce noncoverage and nonresponse bias in our sample and to improve the accuracy of point estimates, we estimate the outcome of the 2021 federal election in Germany, an external population benchmark. We compare the estimates of the unweighted sample, the weighted sample with weights calculated based solely on sociodemographic variables and the weighted sample with the vaccination weights, with and without additional sociodemographic variables. We suppose that by adding vaccination status to the adjustment weights, the estimates for the outcome of the 2021 federal election in Germany move toward the benchmark in comparison to either the unweighted or sociodemographic weighted sample. We use the MSE (Mean Squared Error) as measure for this comparison (Peytchev, Presser & Zhang 2017). The particularity of this study lies in its ability to evaluate the reduction of bias of the augmented weights using population benchmarks.
dcterms:rights
dcterms:type
dcterms:creator
osf:contains

://osf.io/j3ykg

osf:File
more details...
dcterms:identifier
https://osf.io/j3ykg
osf:fileName
vac_weight.do
rdf:type
type
dcterms:created
2021-09-26
osf:isContainedBy
osf:filePath
/Archive of OSF Storage/vac_weight.do
osf:hasFileVersion

://osf.io/j3ykg
?revision=1

osf:FileVersion
more details...
dcterms:requires
dcterms:creator
dcterms:modified
2021-09-25
rdf:type
type
dcterms:format
application/octet-stream
dcterms:created
2021-09-25
osf:versionNumber
1
osf:storageRegion
dcterms:extent
0.020 MB
dcterms:modified
2021-09-26
dcterms:isVersionOf

://osf.io/w3djp

osf:Project
VacWeight
more details...
dcterms:publisher
dcterms:identifier
https://osf.io/w3djp
https://doi.org/10.17605/OSF.IO/W3DJP
dcterms:created
2021-08-28
owl:sameAs
sameAs
rdf:type
type
dcterms:creator
dcterms:title
VacWeight
rdf:type
type
prov:qualifiedAttribution
qualifiedAttribution
(blank node)
prov:agent
agent
dcat:hadRole
owl:sameAs
sameAs
osf:hostingInstitution
foaf:primaryTopic
primary topic