Multi-Value QCA with Multiple Outcome Values

[Simon Kolbeck]

Hi All,

After reviewing the literature on multi-value QCA, I noticed that the focus tends to be on multiple outcomes. However, I did not see any cases where the outcome value takes on multiple values.

Is it possible to use mvQCA on a dataset in which the outcome has 4 conditions? Or, is there another form or extension of QCA that is more suited for such situations. 

An example of an outcome with 4 conditions would be a nominal variable with 4 values. 

Thanks,

Simon 

[Simon Kolbeck]

On Friday, March 27, 2020 at 2:02:19 PM UTC-4, Simon Kolbeck wrote:

Sorry, I meant to say that the focus tends to be on multiple conditions, not outcomes, in the first sentence.

[Adrian Dușa]

Dear Simon,

Yes, the focus tends to be on multi-value conditions, and with a very good reason: the algorithm is called "Boolean" minimization. Which means the outcome value should always be Boolean, or at least converted to Boolean in the truth table. We need to compare the configurations where the outcome is "present" vs where the outcome is "absent". There can be no other state (except for the unknown remainders).

Having said that, your dataset can obviously contain fuzzy outcomes, and it gets recoded to Boolean for the truth table configurations. What it might not be so obvious is that outcomes from your dataset can also be multi-value, but you will have to specify which levels should be considered equal to 1.

For instance, suppose you have an multi-value outcome Y with the following values: 0, 1, 2 and 3.

When creating the truth table, you need to specify the outcome, suppose: outcome = "Y". This works well in case of binary crisp and fuzzy set data, but for multi-value the truth table function will ask you to specify what level is intented to be considered as <present>.

For multi-value outcomes, you might want to specify: outcome = "Y{3}"

This is equivalent to a manual recoding of your outcome column in the data, where 3 becomes equal to 1 and all other values (0, 1, and 2) would be recoded to 0, this creating a binary crisp outcome.

Or you might want to specify: outcome = "Y{1,3}"

This would be similarly equivalent to a recoding of values 1 and 3 into 1, while values 0 and 2 would be recoded to 0. Either way, you will obtain a binary crisp outcome, which is absolutely mandatory.

Hope this helps,
Adrian

> --
> You received this message because you are subscribed to the Google Groups "QCA with R" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to qcawithr+u...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/qcawithr/04ad61b6-72ec-4c7f-89f7-ec7903d24cca%40googlegroups.com.

—
Adrian Dusa
University of Bucharest
Romanian Social Data Archive
Soseaua Panduri nr. 90-92
050663 Bucharest sector 5
Romania
https://adriandusa.eu

[Simon Kolbeck]

Hi Adrian,

Thank you. This is very helpful and clears a lot up.

Best,

Simon

On Friday, March 27, 2020 at 2:37:33 PM UTC-4, Adrian Dușa wrote:

Dear Simon,

Yes, the focus tends to be on multi-value conditions, and with a very good reason: the algorithm is called "Boolean" minimization. Which means the outcome value should always be Boolean, or at least converted to Boolean in the truth table. We need to compare the configurations where the outcome is "present" vs where the outcome is "absent". There can be no other state (except for the unknown remainders).

Having said that, your dataset can obviously contain fuzzy outcomes, and it gets recoded to Boolean for the truth table configurations. What it might not be so obvious is that outcomes from your dataset can also be multi-value, but you will have to specify which levels should be considered equal to 1.

For instance, suppose you have an multi-value outcome Y with the following values: 0, 1, 2 and 3.

When creating the truth table, you need to specify the outcome, suppose: outcome = "Y". This works well in case of binary crisp and fuzzy set data, but for multi-value the truth table function will ask you to specify what level is intented to be considered as <present>.

For multi-value outcomes, you might want to specify: outcome = "Y{3}"

This is equivalent to a manual recoding of your outcome column in the data, where 3 becomes equal to 1 and all other values (0, 1, and 2) would be recoded to 0, this creating a binary crisp outcome.

Or you might want to specify: outcome = "Y{1,3}"

This would be similarly equivalent to a recoding of values 1 and 3 into 1, while values 0 and 2 would be recoded to 0. Either way, you will obtain a binary crisp outcome, which is absolutely mandatory.

Hope this helps,
Adrian

> On 27 Mar 2020, at 20:02, Simon Kolbeck <sgko...@gmail.com> wrote:
>
> Hi All,
>
> After reviewing the literature on multi-value QCA, I noticed that the focus tends to be on multiple outcomes. However, I did not see any cases where the outcome value takes on multiple values.
> Is it possible to use mvQCA on a dataset in which the outcome has 4 conditions? Or, is there another form or extension of QCA that is more suited for such situations.
> An example of an outcome with 4 conditions would be a nominal variable with 4 values.
>
> Thanks,
>
> Simon
>
> --
> You received this message because you are subscribed to the Google Groups "QCA with R" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to qcaw...@googlegroups.com.