Error with choiceRT_DDM command

[Aleksander Nero]

Hello

I am currently running into a problem, where when I run the choiceRT_ddm command, the code does say that model fitting is complete, it completes with 50 warnings which all say
50: In mean.default(parVals[[x]][, i]) : argument is not numeric or logical: returning NA

Also when the model is fitting, I am getting errors like  Chain 1: Rejecting initial value: Chain 1: Error evaluating the log probability at the initial value. Chain 1: Exception: modeldc887bc65382__namespace::log_prob: tau[8] is 0.0510793, but must be greater than or equal to 0.100000 (in 'string', line 53, column 2 to column 49) I have updated R to the newest version and reinstalling the newest version of all packages and Rtools, but that doesn't seem to have made a difference.

Also as a separate question, is it possible, to run the command choiceRT_ddm_single, even if the subject has only chosen one option (the correct one) throughout the entire experiment Thank you!
Best,
Nero

[Jeongyeon Shin]

Hello Nero,

I took a closer look at your data and noticed that some participants seem to have always chosen the same option (always selecting 2). In such cases, it's difficult for DDM to estimate the parameters properly because the model needs some variation in both choices and RTs. This is because DDM relies on the interplay between choice and response time to distinguish between different parameters (e.g., drift rate vs. bias), and when choices don’t vary, the model becomes mathematically underdetermined.

One option is to exclude those participants from the analysis. Another option is to manually set initial values for tau to help the model start properly, though this is only a temporary solution and may not fully address the issue.

Also, for the same reason, it's difficult to apply choiceRT_ddm_single to subjects who consistently chose only one option.

Please let me know if you have any further questions!

Best,

Jeongyeon

2025년 4월 22일 화요일 오후 9시 3분 50초 UTC+9에 aleksan...@gmail.com님이 작성: