Missing data: No access to NA values, consecutive data points only

[Mohammad Khaleghi]

Hi there

In a dataset, participants were supposed to provide reports twice daily over 90 days, while only minimum of 60 observations per person needed. If we do not have access to the raw timestamps or explicit missing data (NA values) and enter only 60 consecutive responses per participant into GIMME, any problem?

My question is: does it create a serious problem for the modeling that missing responses were not represented as NA values (i.e., I treated the 60 points as a continuous series)? I understand that this might especially affect lagged relations, but contemporaneous relations might be more robust.

I would really appreciate your guidance on whether this approach is acceptable or whether it could distort the results. Any suggestion? 

Thanks

[Katie Gates]

Hi Mohammad, 

Typically, people put a row of "NA"s for one or multiple time points of missingness. Sounds like you don't have this option. I think it's still fine. 

You are absolutely right that in this situation it would impact the lagged relation estimates. This in turn might have some influence on the contemporaneous estimates, since lagged effects may not be totally controlled for. But, given that in most EMA studies the contemporaneous effects are typically far stronger than lagged, to me this is OK. 

Curious to hear what others might think. 

Best, 

Katie