|
|
Longitudinal Missing Data |
|
Message/Author |
|
Chance Bell posted on Friday, October 20, 2017 - 11:04 am
|
|
|
I have a dataset with 1500+ cases and 12 time points (over 8 years). Clients at a clinic completed a measure at the beginning of treatment and various time points during treatment. Some completed more measures than others because they were in treatment longer. I have two types of missing data. (1) Those who completed measures for time 1 and time 2, but then terminated therapy and do not have any of the other time points. [labeled -9999] (2) Those who completed a measure but omitted a question. [labeled 999] I do not want Mplus to run calculations for those that did not complete a measure, but I do for those that completed the measure but omitted a question. After a certain number of time points the percentage of non completed surveys is quite high (>75%). Is it possible for Mplus to skip/ignore values for the first type of missing data (-9999), and run calculations for those with the second type of missing data (999)? If I understand this issue inaccurately please let me know. Thanks, Chance |
|
|
The subjects in (1) won't contribute to the estimation of parameters at the time points where they are not present. But I think you want them to contribute to the estimation of parameters when they are present. Default missing data handling will take care of both (1) and (2) if you let -9999 and 999 both be missing data flags. |
|
Back to top |
|
|