Mixed Model and Test Subcommand

[Elanor]

Hi there,

I have a question regarding a statistical analysis I am conducting at present. I have a 2 x 2 mixed design, with one factor being between subjects and one factor within subjects. I am expecting an ordinal interaction, with the BS factor making a difference only for WS1 but not for WS2.

Now I would like to run a hypothesis test with specified contrasts and test the one group that is supposed to differ against all other groups. I have tried using the MIXED Command and define this contrast via the /test subcommand. However, when I do this SPSS always gives me an error for the hypothesis test, telling me that "the L Matrix cannot be estimated". Does anyone has an idea what went wrong? Happy for any ideas!

Cheers,
Selma

That's the syntax I am currently using:

MIXED DV BY WS BS
  /FIXED=WS*BS WS BS | SSTYPE(3)
  /PRINT=DESCRIPTIVES
  /REPEATED=WS | SUBJECT(id) COVTYPE(AR1)
/TEST = 'Kontrast' WS*BS -1 1/3 1/3 1/3.

[Ryan]

Hint: First figure out how to estimate the means using the TEST statements. Then subtract the means of interest.

Ryan

On Nov 1, 2013 5:04 AM, "Elanor" <[hidden email]> wrote:

Hi there,

I have a question regarding a statistical analysis I am conducting at
present. I have a 2 x 2 mixed design, with one factor being between subjects
and one factor within subjects. I am expecting an ordinal interaction, with
the BS factor making a difference only for WS1 but not for WS2.

Now I would like to run a hypothesis test with specified contrasts and test
the one group that is supposed to differ against all other groups. I have
tried using the MIXED Command and define this contrast via the /test
subcommand. However, when I do this SPSS always gives me an error for the
hypothesis test, telling me that "the L Matrix cannot be estimated". Does
anyone has an idea what went wrong? Happy for any ideas!

Cheers,
Selma

That's the syntax I am currently using:

MIXED DV BY WS BS
  /FIXED=WS*BS WS BS | SSTYPE(3)
  /PRINT=DESCRIPTIVES
  /REPEATED=WS | SUBJECT(id) COVTYPE(AR1)
/TEST = 'Kontrast' WS*BS -1 1/3 1/3 1/3.




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Mixed-Model-and-Test-Subcommand-tp5722841.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Elanor]

Dear Ryan,

thank you for your advice! I have trying to figure out how to do it by looking at older threats. Is this the kind of command line I am looking for? (as I said, it's a 2 x 2 design with one repeated factor), and am trying to show that the combination MSreg is different from the other three groups (MSrej, DPreg and DPrej).

In this case, would the last  test be the one I report?

Cheers,

Selma

MIXED Anzahl_Erinnerung BY  MortalitySalience Index1

 /FIXED = MortalitySalience*Index1 MortalitySalience Index1

 /REPEATED = MortalitySalience*Index1 | SUBJECT(id) COVTYPE(UN)

/TEST 'mean(MSreg)' intercept 1 MortalitySalience 1 0 Index1 1 0  MortalitySalience*Index1 1 0 0 0

 /TEST 'mean(MSrej)' intercept 1 MortalitySalience 1 0 Index1 0 1 MortalitySalience*Index1 0 1 0 0

 /TEST 'mean(DPreg)' intercept 1 MortalitySalience 0 1 Index1 1 0 MortalitySalience*Index1 0 0 1 0

 /TEST 'mean(DPrej)' intercept 1 MortalitySalience 0 1 Index1 0 1 MortalitySalience*Index1 0 0 0 1

  /TEST 'MEAN(MSrej, DPreg, DPrej)'  intercept 3 MortalitySalience 1 2 Index1 1 2 MortalitySalience*Index1 0 1 1 1 DIVISOR=3

  /TEST 'MSreg; revised with divisor 3'  intercept 3 MortalitySalience 3 0 Index1 3 0 MortalitySalience*Index1 3 0 0 0 DIVISOR=3

  /TEST 'MEAN(MSrej, DPreg, DPrej ) MINUS MSreg; revised with divisor 3' intercept 0 MortalitySalience -2 2 Index1 -2 2 MortalitySalience*Index1  -3 1 1 1 DIVISOR=3. 

Am 01.11.2013 um 22:37 schrieb Ryan Black [via SPSSX Discussion]:

Hint: First figure out how to estimate the means using the TEST statements. Then subtract the means of interest.

Ryan

On Nov 1, 2013 5:04 AM, "Elanor" <<a href="x-msg://215/user/SendEmail.jtp?type=node&amp;node=5722866&amp;i=0" target="_top" rel="nofollow" link="external">[hidden email]> wrote:

Hi there,

I have a question regarding a statistical analysis I am conducting at
present. I have a 2 x 2 mixed design, with one factor being between subjects
and one factor within subjects. I am expecting an ordinal interaction, with
the BS factor making a difference only for WS1 but not for WS2.

Now I would like to run a hypothesis test with specified contrasts and test
the one group that is supposed to differ against all other groups. I have
tried using the MIXED Command and define this contrast via the /test
subcommand. However, when I do this SPSS always gives me an error for the
hypothesis test, telling me that "the L Matrix cannot be estimated". Does
anyone has an idea what went wrong? Happy for any ideas!

Cheers,
Selma

That's the syntax I am currently using:

MIXED DV BY WS BS
  /FIXED=WS*BS WS BS | SSTYPE(3)
  /PRINT=DESCRIPTIVES
  /REPEATED=WS | SUBJECT(id) COVTYPE(AR1)
/TEST = 'Kontrast' WS*BS -1 1/3 1/3 1/3.




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Mixed-Model-and-Test-Subcommand-tp5722841.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
<a href="x-msg://215/user/SendEmail.jtp?type=node&amp;node=5722866&amp;i=1" target="_top" rel="nofollow" link="external">[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Mixed-Model-and-Test-Subcommand-tp5722841p5722866.html

To unsubscribe from Mixed Model and Test Subcommand, click here.
NAML