Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Missing values that are left out analysis

Hey!

I run into a problem with the repeated measures anova. I have done a pilot, collecting data 5 days in a row. They had to solve the same problems each day in order to train them on the problems. Of course, I am interested in seeing if their reaction time for solving the problems lowered during their training. My data is organized in the following way: I have a column with subject number, a column with RT_1, for the reaction time for each problem individually for session one, RT_2, for session 2 and so on. The thing is, RT is only interesting when problems are solved correctly. So, when a problem was solved incorrect, the value is empty.

Now, Jasp only included the problems that are solved correctly over all 5 sessions, because when for a problem (one for each row), one value is missing, it also excludes the RT in the other sessions when the problem was solved correctly. Can I force Jasp to include all of the values, independent of how they were solved in another session?

I cannot find this exact question in other topics considering missing values, but sorry if this question already has been answered elsewhere. I just could not find it.

Kind regards ;)

Comments

  • Hi Merel,

    I don't think you can, at least not right now. If the five conditions were between-subjects you'd just have different sample sizes in each of the conditions. But here you have a within-design, and this complicates things. From a Bayesian perspective, you would just estimate the missing values as if they were parameters about which you are unsure.

    Please let me know if you find an acceptable solution somewhere, and maybe we can implement it. Perhaps I am mistaken and the problem is straightforward.

    Cheers,
    E.J.

  • Okay, thanks for your response. For now I will try to do the anova using averages per block or problem (I still have to figure out which will answer my question best ;) ). Then, I lose the missing values and I will have the same sample sizes for each condition (training session in my case). I think that will be closer to the truth than just getting rid of all trials where only 1 measure lacks, as in now the case.

    Have some nice holidays!
    Merel

Sign In or Register to comment.