# Release Notes for DZHW PhD Panel 2014 (phd2014)

All notable changes to this data package will be documented in this file.
The format is based on Keep a Changelog.

## [4.0.0] 2020-10-30
### Added
- Nothing
### Changed
- Nothing
### Removed
- Nothing
### Fixed
- Data:
    - variable pid: Variable pid was changed in the individual data record because it differed between individual data and spell data from version 3.0.0.
	

## [3.0.1] 2020-10-23
### Added
- Documents related to data package
    - Added data and methods report (DE)
    - Added citation guideline (DE/EN)
    - Added data package overview (EN)
### Changed
- Nothing
### Removed
- Nothing
### Fixed
- Nothing


## [3.0.0] 2020-07-01
### Added
- Data: Added on-site SUF, remote-desktop SUF, download SUF, download CUF (wave 3 to 5) (DE)
- Metadata
- Documents related to data package
### Changed
- Nothing
### Removed
- Nothing
### Fixed
- Nothing


## [2.0.0] 2018-05-17
### Added
- Nothing
### Fixed
- Nothing
### Changed
- Data: 
   - variable bocc14: During the data editing of the second wave, the variable was 
     mistakenly changed in 109 cases. This concerns especially persons on parental leave, 
     who stated in bocc14 that they were currently not in employment, but who were assigned 
     a current employment due to the indication of a current employment in the calendar 
     (change from value 2 or value 3 to value 1). These changes have been revoked and the stated 
     information of the surveyed persons has been restored.
   - attrition weight for wave 2: The attrition weight for wave 2 has been adjusted. In order 
     to predict participation in the second wave, the industry sector (wave 1) will additionally 
     be used. By this means, varying participation probabilities of persons from different industry 
     sectors will be compensated for.
   - trimming of the weights: The initially calculated weights exhibit a small proportion of 
     outlying weighting factors. In order to remove them, all weights were subjected to a trimming 
     according to Potter 1990 (see also Valliant et al. 2013, pp. 338). The procedure is based on the 
     assumption that the weights conform to a probability distribution (beta distribution). All 
     those weights that lie above the 99 percent quantile are truncated to this limit. Excess on the 
     other side of the truncation is distributed among the remaining weights.
- Documents:
   - In the Data and Methods Report, the chapter on weighting has been accordingly adjusted.
   
   
## [1.0.0] 2017-06-01
### Added
- Data: Added on-site SUF, remote-desktop SUF, download SUF, download CUF (wave 1 and 2) (DE)
- Metadata
- Documents related to data package
### Changed
- Nothing, initial release
### Removed
- Nothing, initial release
### Fixed
- Nothing, initial release