Analysis | Preprocessing script

HIFIS Survey 2021: Data preparation

Recoding

• Q001: -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7
  • Categorize free text answers
  • Replace answers that fit in no category with missing value
  • Replace -oth-/Other in Q001/_ with answer from free text field in Q001/other
• Q022: Replace A00x with SQ00x -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7
• Q031: Replace A000/No answer with missing value -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7
• Q040: Rename {SQ002: SQ003, SQ003: SQ002} -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7

Transformation

• free text fields -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7
• new variable for
  • developer -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/analysis/-/issues/129
  • HIFIS -> https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/analysis/-/issues/128

Update values for individual cases

• Q038_comment: Manual screening -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7
• Q045: Manual screening -> done in https://gitlab.hzdr.de/hifis/overall/surveys/survey-2021/data/-/merge_requests/7

Validity checks

• Q005/Q006: not(Q005/_ == A002/Yes and Q006/_ == A006/Non-developer)
  - [ ] Q007/_: 0 < value < 100
  - [ ] Q022/_: check if answer of Q022 is in list of checked subquestions in Q021
  - [ ] Q014: 1-3 checked sub-questions
  - [ ] Q025: 1-5 checked sub-questions
• Q025/Q005: Q005/_ == A002/Yes
  - [ ] Q027/Q026: check if sub-questions of Q027 were not checked in Q026 -> probably done in consulting
  - [ ] Q028/Q026: check if sub-quesitons of Q028 were checked in Q026 -> probably done in consulting

Exclusion

• Q001/_ == -oth-/Other
• Q002/_ == A006/I have never worked for a Helmholtz centre
• incomplete cases: missing values in Q001-Q014:
  • all: Q001, Q002, Q003, Q005, Q006, Q009, Q010, Q011, Q012, Q013, Q014
    • research_personnel (Q003/_ == A001 or Q003/_ == A002): all + Q004
    • potential_developer (Q006/_!=A006): all + Q008
    • realized_developer (Q005/_==A002): all + Q007
• complete cases: all_cases - (incomplete cases + invalid cases)

Export options

• as CSV
• all cases including incomplete ones
• question codes and answer codes instead of text
• use expression manager to avaoid [] as separator