resquin 0.1.1
CRAN release: 2025-06-27
- Changed the way the mahalanobis distance is calculcated in
resp_distributions(), if missing values are allowed. Now within respondent mean imputation is used. Before missing values were turned to a value of 0, which is wrong in almost any case and would have skewed the results of respondents with missing values. Mean imputation is not an ideal solution, but it allows the observed data to influence the value of the mahalanobis distance value under missing data. Because within respondent mean imputation is not ideal, a new section in the description is added to call for caution when interpreting the mahalanobis distance values produced byresp_distributions()ifmin_valid_responsesis smaller than 1. - Polished documentation (fixing typos etc.)
- Added vignette on flagging respondents with
flag_resp(). - Added a disclaimer for handling of missing data in
resp_nondifferentiation. - Added documentation on s3 methods.
- Fixed bug in s3 print function which would crash if
resp_styles()was used on even numbered response scales. - Added tests for functions added since 0.0.2.
- Now depends on R version being >= to R version 4.1.
- Added example data set
nepfrom the GESIS panel.
resquin 0.1.0
- Added
resp_patterns()andresp_nondifferentiation()as a new function. - Added
idcolumn to all outputs to make it easier to identify respondents or merge function outputs to data frames.idis eitherTruefor an integer id,Falsefor noidcolumn, or a vector of unique integer or character values identifying each respondent. - Added
flag_resp()function to quickly create and compare different flagging strategies based on response quality indicators. - Added s3 types to outputs of
resp_*()andflag_resp()functions. - Added s3 print, summary and plot methods for outputs of
resp_*()functions. - Added s3 summary method for
flag_resp()output.