Hi

I have performed an independent samples t-test to evaluate the difference between two experimental groups which represent different intervention conditions. The groups have quite different standard deviations, so I was planning on reporting Glass's delta (rather than Cohen's d) but I have two queries about this:

1) how does SPSS decide on which is the control group? I'm aware that Glass's delta uses just the st dev for the control group so it matters which group is which
2) is it still OK to use this test even though in a sense neither of my groups is really a control. For this test, I am trying to figure out whether intervention A or B is better not A with respect to no intervention at all.

Any thoughts or advice would be much appreciated!

Thanks

Steph

------------------------------
Stephanie Ainsworth
------------------------------

#SPSSStatistics

Hi.

1. According to the Algorithms manual, the second group is the "control" group. I verified that with a calculator on an example run through SPSS.

2. If you have no "control" group, then of course you will have to defend yourself against those who will challenge your design and the choices made therein, but those arguments are methodological, not statistical.

------------------------------
Rick Marcantonio
Quality Assurance
IBM
------------------------------

Original Message:
Sent: Tue September 27, 2022 06:04 AM
From: Stephanie Ainsworth
Subject: How does SPSS decide the control group for Glass's delta?

Hi

I have performed an independent samples t-test to evaluate the difference between two experimental groups which represent different intervention conditions. The groups have quite different standard deviations, so I was planning on reporting Glass's delta (rather than Cohen's d) but I have two queries about this:

1) how does SPSS decide on which is the control group? I'm aware that Glass's delta uses just the st dev for the control group so it matters which group is which
2) is it still OK to use this test even though in a sense neither of my groups is really a control. For this test, I am trying to figure out whether intervention A or B is better not A with respect to no intervention at all.

Any thoughts or advice would be much appreciated!

Thanks

Steph

------------------------------
Stephanie Ainsworth
------------------------------

#SPSSStatistics

Thanks Rick! This is really helpful.

Best Wishes

Steph

------------------------------
Stephanie Ainsworth
------------------------------

Original Message:
Sent: Tue September 27, 2022 11:26 AM
From: Rick Marcantonio
Subject: How does SPSS decide the control group for Glass's delta?

Hi.

1. According to the Algorithms manual, the second group is the "control" group. I verified that with a calculator on an example run through SPSS.

2. If you have no "control" group, then of course you will have to defend yourself against those who will challenge your design and the choices made therein, but those arguments are methodological, not statistical.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Tue September 27, 2022 06:04 AM
From: Stephanie Ainsworth
Subject: How does SPSS decide the control group for Glass's delta?

Hi

I have performed an independent samples t-test to evaluate the difference between two experimental groups which represent different intervention conditions. The groups have quite different standard deviations, so I was planning on reporting Glass's delta (rather than Cohen's d) but I have two queries about this:

1) how does SPSS decide on which is the control group? I'm aware that Glass's delta uses just the st dev for the control group so it matters which group is which
2) is it still OK to use this test even though in a sense neither of my groups is really a control. For this test, I am trying to figure out whether intervention A or B is better not A with respect to no intervention at all.

Any thoughts or advice would be much appreciated!

Thanks

Steph

------------------------------
Stephanie Ainsworth
------------------------------

#SPSSStatistics