# Thread: Panel model with FE minimum number of observations per group

1. ## Panel model with FE minimum number of observations per group

Hello dear forum members,

I am estimating a panel model with fixed effects using yearly data from 2009 to 2013 (N=1848). Whereas 80% of the data encounter less than 5% of missing values, the remaining 20% encounter 75% of the missing values. These "problematic" 20% of the data are review-based rating scores, which (due to their nature) are not given every year to every entity in the data set. For example, an entity may be reviewed (and given a rating) say once or twice in 5 years (and not necessarily in consecutive years), thus resulting in missing values.

Stata's -xtnbreg, fe- estimation results in a total N=751 with 2,049 total observations, i.e., min observations per group 2, max = 5, and avg = 2.7

Does this relatively low average number of observations per group (i.e., 2.7 out of 5) bias the estimates substantially? Or poses any other problems?

Anton

2. ## Re: Panel model with FE minimum number of observations per group

I guess if the missing values were proven to be random, then there is a chance estimates are not biased. Otherwise, they are. Hm... having that many missings even MI will not help, or will it?

Talking to myself on a stats forum -- time for a break )

3. ## Re: Panel model with FE minimum number of observations per group

Originally Posted by kiton
These "problematic" 20% of the data are review-based rating scores, which (due to their nature) are not given every year to every entity in the data set. For example, an entity may be reviewed (and given a rating) say once or twice in 5 years (and not necessarily in consecutive years), thus resulting in missing values.
the way you phrase this makes it sound as if it were "structurally missing" data. if that is the case, then it falls under the Missing Completely At Random (MCAR) umbrella term and you can use the naive strategies (listwise/pairwise deletion) to take care of missing data.

but me being me, i would find a way to estimate if through full information maximum likelihood because...FIML.

4. ## The Following User Says Thank You to spunky For This Useful Post:

kiton (06-08-2015)

5. ## Re: Panel model with FE minimum number of observations per group

Originally Posted by spunky
the way you phrase this makes it sound as if it were "structurally missing" data. if that is the case, then it falls under the Missing Completely At Random (MCAR) umbrella term and you can use the naive strategies (listwise/pairwise deletion) to take care of missing data.

but me being me, i would find a way to estimate if through full information maximum likelihood because...FIML.
Dear spunky, thank you for your suggestions. I will explore FIML in more detail.

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts