However, rather than imputing a single value for a missing data point and treating it as fixed, the multiple imputation framework accounts for uncertainty about the missing data by creating multiple plausible missing values resulting in multiple data sets.
Missing values resulted in the exclusion of three participants.
Although PROGRESS involved large patient numbers, significant levels of missing values resulted in small numbers for certain parameters.
Missing values resulted in the exclusion of four participants, thus 100 widowed mothers were included in the analyses.
Thus, if there were no missing values, the weight was 1, with missing values resulting in down-weighting; mean weights for each parameter ranged from 0.77 to 0.91.
The considerable share of missing values resulted from the high-quality requirements for data for the variable 'attendance of midwives' during labour.
The dataset was first filtered to remove microarray slides with >50% missing values followed by a second filter to remove gene models containing >50% missing values, resulting in a final set of 602 arrays.
Among the 149 patients from the validation set, 18 were discarded due to missing values, resulting in a sample size of 131, with 40 events and 91 censored observations.
306 questionnaires were excluded for further analysis: 200 from single-handed practices, 11 from respondents with less than half a year experience at the particular practice and 94 with more than 5 missing values, resulting in 615 questionnaires eligible for the study.
Because of these results, and since no individual variables had more than 10% missing values, the results reported are from the complete-case dataset.
