ANOVA with missing case

[brett_a_baxter]

Hi All,

I was wondering if anyone could shed some light on a current situation I have encountered. I have a study set-up with 13 participants that I am examining over 5 time-points, hence, I intend to conduct a one-way ANOVA. However, on one of the time-points, a participant was unable to attend the situation so I now have a missing case but given there only being 13 participants I am reluctant to remove the participant from the data set. Could anyone please advise what to do in this situation? My data is laid out in wide-format e.g. every row is a participant and each measure has 5 columns such as strengthday1, strengthday2 etc.

Thank you in advance,

Brett

 

[fed2]

probably one-way repeated measures anova since you are testing the effect of time within subject?

the missing observation is not a deal breaker. you would need to use mixed-model though. this is an option in most common software packages.

 

[brett_a_baxter]

probably one-way repeated measures anova since you are testing the effect of time within subject?

the missing observation is not a deal breaker. you would need to use mixed-model though. this is an option in most common software packages.

Thank you for your response. Would I be correct in thinking that this is a random effects model and the subject ID is entered as the random variable?

 

[fed2]

yes. random intercept.

 

[brett_a_baxter]

yes. random intercept.

Brilliant, thank you so much.

 

[noetsi]

Missing cases are less than ideal. You might want to look at the literature on multiple imputations as a way of filling in the value.

 

[brett_a_baxter]

Missing cases are less than ideal. You might want to look at the literature on multiple imputations as a way of filling in the value.

Thank you for you advice, this is something I will look into, if you have any tutorials/papers to recommend I would be extremely grateful.

 

[noetsi]

Thank you for you advice, this is something I will look into, if you have any tutorials/papers to recommend I would be extremely grateful.

Most of what I know is in SAS and I am not sure how useful that would be to you (ignoring that I have not worked with this in a long time, my data never has missing values).

You might look at the Paul Allison Sage monograph from 2001 called Missing Data. I have not read it, I use other sources, but those monographs are usually pretty good and he is a good author.

I glanced through this, it might be a good place to learn the basics. The real trick to me is learning the coding. There are many ways to do what they suggest and they all make different assumptions as does the software.

Missing Data Analysis Using Multiple Imputation: Getting to the Heart of the Matter (nih.gov)