Feature request: create new variable in Reliability Analysis
Posted: Mon May 06, 2019 5:55 am
Reliability Analysis is on the face of it much better than the SPSS version. I especially love the ability to note which variables are reverse scaled rather than having to recode them separately. However, the benefit of this feature is diminished when the user realises that they still need to recode the variables and compute the average in order to create the scale variable, the mean and standard deviation of which jamovi can already tell them.
Also, while I prefer the mean, some standardised instruments ask for the sum instead so it would be great if:
1) There was a text box where the name of the computed variable could be entered.
2) There was a check box to allow the sum to be used instead of the mean. The scale statistics should, of course, then reflect the sum rather than mean if this option is chosen.
3) I'm not sure what the current behaviour is for missing values, but it might be that the mean should be the mean of the available values, but the sum should impute the mean across participants for any holes in the data.
4) One final option would be to switch to using Z scores instead of raw scores. I don't often use this, but recommend it if some items have different response categories -- though I also recommend avoiding this situation.
Cheers,
Wake
Also, while I prefer the mean, some standardised instruments ask for the sum instead so it would be great if:
1) There was a text box where the name of the computed variable could be entered.
2) There was a check box to allow the sum to be used instead of the mean. The scale statistics should, of course, then reflect the sum rather than mean if this option is chosen.
3) I'm not sure what the current behaviour is for missing values, but it might be that the mean should be the mean of the available values, but the sum should impute the mean across participants for any holes in the data.
4) One final option would be to switch to using Z scores instead of raw scores. I don't often use this, but recommend it if some items have different response categories -- though I also recommend avoiding this situation.
Cheers,
Wake