jamovi

Hello

While attempting a reliability analysis with my dataset, I encountered two errors. The first error states "Item X has no variance," while the second error indicates "Missing value where True/False needed."

After investigating further, I identified a pattern related to the second error. It occurs when a variable lacks a factor level that appeared in previous variables but is absent in the specific variable being used in the calculation. To address this, I attempted to manually add the missing factor level to the variable via the setup function in "Variables". Unfortunately, this did not resolve the error.

For example, including "Car4_Level 1" in the reliability analysis triggered this issue, along with "Car_Level 1," "Car2_Level 1," and "Car3_Level 1." Only including the last three variables and leaving out "Car4_Level 1" works fine.

I am wondering if there is a way to resolve this issue, as it seems to stem from the missing factor level (5). There should be 6 in total, ranging from 1 to 6.

The second error, "Item X has no variance," occurs when calculating the reliability of "Car_Level4," "Car2_Level4," "Car3_Level4," and "Car5_Level4," with X representing "Car3_Level4." Sadly, I have not yet determined the underlying cause for this error. However, when I conduct an exploration using the "Descriptives" function in "Exploration", the results display a variance.

The missing factor level is also a problem when I do a principal component analysis. I am guessing the problem here could be fixed if I am able to fix the problem of "Missing value where True/False needed"? Yet I am unaware on how that could be done.

If anyone can provide some insight into the possible reasons for these errors, I would be very grateful. Perhaps there is a statistical reason, I am sorry if that is the case. My knowledge is rather limited in this area. I have attached the jamovi file (.omv) to this message for reference.

Best regards,
Deniz

hi,

check that you're using the latest version. there is a problem with your data, but i agree the error messages jamovi is producing here aren't that helpful. let me attend ravi to this. he normally takes a few days to respond, so he may not get here until tuesday.

jonathon

Thanks for the quick reply. This sounds like a good plan. I only have a quick question, what do you mean by "there is a problem with your data"? Is this referring to the missing values in some rows/colums?

an update, ravi won't get to this until friday.

missing values in rows/columns is generally OK ... this is more likely to do with "empty cells in the design", or "few observations in cells of the design". mz might chime in on this.

jonathon

jonathon wrote: ↑Tue Jun 13, 2023 1:28 am mz might chime in on this.

Here I am.

I took a look on the fly.
By inserting the 4th variable 'Car 4_Level 1' the refresh of the data exits the (car wash ) with a single observation...

Cheers,
Maurizio

Hi Deniz,

Maurizio is indeed right that in your first example ("Car_Level 1", "Car2_Level 1", "Car3_Level 1", and "Car4_Level 1") only 1 row/case remains after deleting all the missing values (jamovi uses listwise deletion in the reliability analysis), and as you can imagine it's impossible to return any reliability results for only one data point.

For your fist example ("Car_Level 1", "Car2_Level 1", "Car3_Level 1", and "Car5_Level 1"), you can see in the screenshot below that after deleting all the missing values, `Car3_Level 1` doesn't have any variance because the only values remaining are three 4's. I think it's also not possible to do a reliability analysis if one of the variables included doesn't have any variance.
Hope this at least makes it a bit more clear why these analyses don't work for your data. Having said that, the errors that are being displayed could be clearer of course, so we'll look at that.

Thank you very much for the answers. I have also come to the conclusion that the missing values will be problematic for further analysis. Fortunately, I have found a solution that meets the requirements I need to fulfil.