whomds 1.2.0

• Fixed error in rasch_mds() / rasch_model() (“Can’t use NA as column index in a tibble for assignment”) caused by eRm >= 1.0-7 adding a Std.Error column to person.parameter()$theta.table. The person ability columns in df_score are now selected by name instead of by position, so NAgroup and Interpolated are labeled correctly again
• New function helper_disability_cat() categorizes a disability score into ordered levels, by default No, Mild, Moderate and Severe with cut-offs at the mean minus one SD, the mean and the mean plus one SD of the score, which is how disability_cat in df_adults was created. Explicit cut-offs and labels can be supplied. Documented in the c6_after_rasch vignette (#10)
• rasch_mds() and rasch_mds_children() now also return the standard error of each person’s ability estimate in a new column person_pars_se, next to person_pars, in the returned data and in Data_final.csv (#14). It is on the same logit scale as person_pars. This requires eRm >= 1.0-7
• Fixed fig_LID() failing with “adjacency matrix contains NAs” when two items have no respondents in common (for example items split by age group in rasch_mds_children(), or by split_strategy in rasch_mds()), which broke print_results = TRUE. Such item pairs are now treated as having no local dependence
• Fixed table_basicstats() giving wrong results when group_by_var is a character variable rather than a factor: households with no members in a group were omitted instead of being counted as 0
• Added examples to every exported function. The examples for the rasch_*() building blocks show the data preparation that rasch_mds() and rasch_mds_children() do before calling them, and explain what each result means
• Removed dependencies on plyr, readr, scales and stringr, replacing them with base R equivalents and small internal helpers. Output CSV files are now written with utils::write.csv(), so character values are quoted
• Fixed remaining deprecation warnings: as_tibble() on unnamed matrices and bare column selections in rasch_DIF() and rasch_split_age(), the size aesthetic for lines in fig_density() (now linewidth, requiring ggplot2 >= 3.4.0), and aes_string() in fig_dist(). Also fixed a spurious “no non-missing arguments to min” warning from fig_density() when cutoffs is NULL
• Fixed tidyselect deprecation warnings in rasch_recode(), rasch_rawscore(), rasch_DIF(), rasch_split_age() and rasch_rescale(), and replaced a deprecated matrix-to-tibble conversion in rasch_rawscore()
• Removed an internal na.rm = TRUE from the survey_mean() call in table_weightedpct(); it had no effect on the grouped proportions and triggered a warning from srvyr. The function’s arguments are unchanged
• Added unit tests for the data-preparation functions, input validation in rasch_mds(), and the helper functions

whomds 1.1.1

• Fixed bug with related to printing of DIF results
• Create new function SepRel_101() which is a copy of eRm::SepRel() v1.0-1
• Fixed bug with anchored model output in rasch_quality_children_print()
• Updated use of max_NA in rasch_mds_children() so that the function only looks for missing values in the items relevant to the respondent’s group
• Fixed documentation issue with whomds-package alias as requested by CRAN

whomds 1.1.0

• Reduced number of packages in Imports by moving some to Suggests and removing others
• Updated version of R required to v3.6.0 or newer
• Fixed some typos in documentation (#7)
• Fixed labels in output of fig_LID() (#5)
• Fixed setting of x-axis labels in fig_density() (#2)
• Changed some syntax that had been deprecated or superseded in tidyverse functions (#6, #11)
• Updated broken URLs
• Updated example used in documentation for table_basicstats()

whomds 1.0.1

• Updated DESCRIPTION to refer to license correctly
• Updated README and vignettes to refer to installation of package from CRAN

whomds 1.0.0

• First release on CRAN! 🎉