getting an F statistic for an intermediate step
Locked
getting an F statistic for an intermediate step
 
|
Hi, all,
I'm trying to get spss to calculate an F statistic for an intermediate step in some syntax for Cronbach's alpha 95% CIs. I was previously doing this all in excel manually, using the FINV function for this piece, but I'm trying to make syntax for the operation. I need spss to find the F stat with df1 and df2 and input, so I can use the F in the next step of the syntax to generate the CIs. Is there a parallel to FINV in spss (i.e., that will let me just specify the df1 and df2, as they're easily obtainable elsewhere in the syntax)? Thanks, Mike ===================== 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 |
|
|
I believe that the SPSS equivalent of FINV is IDF.F, as in the following from
the v20 syntax manual: IDF.F. IDF.F(prob, df1, df2). Numeric. Returns the value from the F distribution, with the specified degrees of freedom, for which the cumulative probability is prob. For example, the F value that is significant at the 0.05 level with 3 and 100 degrees of freedom is IDF.F(0.95,3,100). -Mike Palij New York University [hidden email] On Sun, Feb 19, 2012 at 11:45 AM, Michael Parent <[hidden email]> wrote: > Hi, all, > I'm trying to get spss to calculate an F statistic for an intermediate step in some syntax for Cronbach's alpha 95% CIs. I was previously doing this all in excel manually, using the FINV function for this piece, but I'm trying to make syntax for the operation. I need spss to find the F stat with df1 and df2 and input, so I can use the F in the next step of the syntax to generate the CIs. Is there a parallel to FINV in spss (i.e., that will let me just specify the df1 and df2, as they're easily obtainable elsewhere in the syntax)? > Thanks, > Mike > > ===================== > 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 ===================== 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 |
Re: getting an F statistic for an intermediate step
Administrator
|
In reply to this post by Michael Parent
Check out the IDF.F function under COMPUTE command.
IDF.F(p,a,b)Return value q such that CDF.F(q,a,b)=p (0≤p<1; a>0; b>0).
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Re: getting an F statistic for an intermediate step
|
|
Thanks for your assistance! That definitely looks like the right function.
I'm running into a problem with implementing it though. When I run the portion of my syntax where I set up my DF and then the F statistic, COMPUTE DFONE = (50-1). COMPUTE DFTWO = (50-1)*(k-1). COMPUTE F01=IDF.F(.95,DFONE,DFTWO). (the 50 is just arbitrary while I debug this; k is defined earlier and is 10 in the debugging example I'm doing now) I get the error: Error # 12310 in column 1024. Text: (End of Command) There can be at most two items in a subscript list. This command not executed. DFONE and DFTWO come out right, because when I ask for them to print I get the right values (49 and 441, respectively, in the debugging example I'm doing now). The error replicates if I just manually type in …IDF.F(.95,49,441). into the above. I'm basing the IDF.F line on some other examples I found online that use the function (e.g., http://www.columbia.edu/~ld208/normtest.sps), though maybe I'm missing something obviously wrong in my setup. Everything else runs fine; I still get the right Cronbach's alpha and the right DFs. If anyone can help me out and let me know where I'm going wrong here, I'd appreciate it (a google search for that error code and text turned up nothing, so I've been stumped for the last hour). Thank you kindly, Mike On Feb 19, 2012, at 1:09 PM, David Marso wrote:
|
|
|
When I run the following code, I get no error message:
compute k=10. COMPUTE DFONE = (50-1). COMPUTE DFTWO = (50-1)*(k-1). COMPUTE F01=IDF.F(.95,DFONE,DFTWO). list var=f01. Where f01=1.38. So, I'm thinking that there's something in the rest of your code that's causing the problem. -Mike Palij New York University [hidden email] On Sun, Feb 19, 2012 at 3:22 PM, Michael Parent <[hidden email]> wrote: > Thanks for your assistance! That definitely looks like the right function. > > I'm running into a problem with implementing it though. When I run the > portion of my syntax where I set up my DF and then the F statistic, > > COMPUTE DFONE = (50-1). > COMPUTE DFTWO = (50-1)*(k-1). > COMPUTE F01=IDF.F(.95,DFONE,DFTWO). > > (the 50 is just arbitrary while I debug this; k is defined earlier and is 10 > in the debugging example I'm doing now) I get the error: > > Error # 12310 in column 1024.� � Text: (End of Command) > There can be at most two items in a subscript list. > This command not executed. > > DFONE and DFTWO come out right, because when I ask for them to print I get > the right values (49 and 441, respectively, in the debugging example I'm > doing now). The error replicates if I just manually type in > …IDF.F(.95,49,441). into the above. I'm basing the IDF.F line on some other > examples I found online that use the function > (e.g.,� http://www.columbia.edu/~ld208/normtest.sps), though maybe I'm > missing something obviously wrong in my setup. > > Everything else runs fine; I still get the right Cronbach's alpha and the > right DFs. > > If anyone can help me out and let me know where I'm going wrong here, I'd > appreciate it (a google search for that error code and text turned up > nothing, so I've been stumped for the last hour). > > Thank you kindly, > Mike > > On Feb 19, 2012, at 1:09 PM, David Marso wrote: > > Check out the IDF.F function under COMPUTE command. > IDF.F(p,a,b)Return value q such that CDF.F(q,a,b)=p (0≤p<1; a>0; b>0). > > Michael Parent wrote > > > Hi, all, > > I'm trying to get spss to calculate an F statistic for an intermediate > > step in some syntax for Cronbach's alpha 95% CIs. I was previously doing > > this all in excel manually, using the FINV function for this piece, but > > I'm trying to make syntax for the operation. I need spss to find the F > > stat with df1 and df2 and input, so I can use the F in the next step of > > the syntax to generate the CIs. Is there a parallel to FINV in spss (i.e., > > that will let me just specify the df1 and df2, as they're easily > > obtainable elsewhere in the syntax)? > > Thanks, > > Mike > > > ===================== > > To manage your subscription to SPSSX-L, send a message to > > LISTSERV@.UGA (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 > > > > > -- > View this message in context: > http://spssx-discussion.1045642.n5.nabble.com/getting-an-F-statistic-for-an-intermediate-step-tp5497180p5497267.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 > > ===================== 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 |
Re: getting an F statistic for an intermediate step
|
Administrator
|
My InterneTelepathy/ESP suggests that Michael is using MATRIX.
AFAIK: The IDF functions are not currently part of the MATRIX language. The CDF functions are. Some time ago I posted a solution to a question posted by Bruce Weaver regarding the IDF for t distribution. Search for that and it will suggest a solution for the F distribution. It basically involves an iterative calculation. Are you certain that the required calculation absolutely requires MATRIX as opposed to simple COMPUTES? HTH, David
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Re: getting an F statistic for an intermediate step
|
|
YES, haha. That was EXACTLY the problem.
And, yes, I just finished the syntax off by jettisoning the 95% CI calculation from the MATRIX portion, and setting it up separately. Works perfect now, even if not as pretty as I'd hoped for. Thanks for the assistance! It is greatly appreciated. Mike On Feb 19, 2012, at 4:53 PM, David Marso wrote: > My InterneTelepathy/ESP suggests that Michael is using MATRIX. > AFAIK: The IDF functions are not currently part of the MATRIX language. > The CDF functions are. Some time ago I posted a solution to a question > posted by Bruce Weaver regarding the IDF for t distribution. Search for > that and it will suggest a solution for the F distribution. > It basically involves an iterative calculation. Are you certain that the > required calculation absolutely requires MATRIX as opposed to simple > COMPUTES? > HTH, David > > > Mike Palij wrote >> >> When I run the following code, I get no error message: >> >> compute k=10. >> >> COMPUTE DFONE = (50-1). >> COMPUTE DFTWO = (50-1)*(k-1). >> COMPUTE F01=IDF.F(.95,DFONE,DFTWO). >> >> list var=f01. >> >> Where f01=1.38. >> >> So, I'm thinking that there's something in the rest of your code >> that's causing the problem. >> >> -Mike Palij >> New York University >> mp26@ >> >> >> On Sun, Feb 19, 2012 at 3:22 PM, Michael Parent <michael.parent@> >> wrote: >>> Thanks for your assistance! That definitely looks like the right >>> function. >>> >>> I'm running into a problem with implementing it though. When I run the >>> portion of my syntax where I set up my DF and then the F statistic, >>> >>> COMPUTE DFONE = (50-1). >>> COMPUTE DFTWO = (50-1)*(k-1). >>> COMPUTE F01=IDF.F(.95,DFONE,DFTWO). >>> >>> (the 50 is just arbitrary while I debug this; k is defined earlier and is >>> 10 >>> in the debugging example I'm doing now) I get the error: >>> >>> Error # 12310 in column 1024.� � Text: (End of Command) >>> There can be at most two items in a subscript list. >>> This command not executed. >>> >>> DFONE and DFTWO come out right, because when I ask for them to print I >>> get >>> the right values (49 and 441, respectively, in the debugging example I'm >>> doing now). The error replicates if I just manually type in >>> …IDF.F(.95,49,441). into the above. I'm basing the IDF.F line on some >>> other >>> examples I found online that use the function >>> (e.g.,� http://www.columbia.edu/~ld208/normtest.sps), though maybe I'm >>> missing something obviously wrong in my setup. >>> >>> Everything else runs fine; I still get the right Cronbach's alpha and the >>> right DFs. >>> >>> If anyone can help me out and let me know where I'm going wrong here, I'd >>> appreciate it (a google search for that error code and text turned up >>> nothing, so I've been stumped for the last hour). >>> >>> Thank you kindly, >>> Mike >>> >>> On Feb 19, 2012, at 1:09 PM, David Marso wrote: >>> >>> Check out the IDF.F function under COMPUTE command. >>> IDF.F(p,a,b)Return value q such that CDF.F(q,a,b)=p (0≤p<1; a>0; b>0). >>> >>> Michael Parent wrote >>> >>> >>> Hi, all, >>> >>> I'm trying to get spss to calculate an F statistic for an intermediate >>> >>> step in some syntax for Cronbach's alpha 95% CIs. I was previously doing >>> >>> this all in excel manually, using the FINV function for this piece, but >>> >>> I'm trying to make syntax for the operation. I need spss to find the F >>> >>> stat with df1 and df2 and input, so I can use the F in the next step of >>> >>> the syntax to generate the CIs. Is there a parallel to FINV in spss >>> (i.e., >>> >>> that will let me just specify the df1 and df2, as they're easily >>> >>> obtainable elsewhere in the syntax)? >>> >>> Thanks, >>> >>> Mike >>> >>> >>> ===================== >>> >>> To manage your subscription to SPSSX-L, send a message to >>> >>> LISTSERV@.UGA (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 >>> >>> >>> >>> >>> -- >>> View this message in context: >>> http://spssx-discussion.1045642.n5.nabble.com/getting-an-F-statistic-for-an-intermediate-step-tp5497180p5497267.html >>> Sent from the SPSSX Discussion mailing list archive at Nabble.com. >>> >>> ===================== >>> To manage your subscription to SPSSX-L, send a message to >>> LISTSERV@.UGA (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 >>> >>> >> >> ===================== >> To manage your subscription to SPSSX-L, send a message to >> LISTSERV@.UGA (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 >> > > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/getting-an-F-statistic-for-an-intermediate-step-tp5497180p5497612.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 > ===================== 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 |
Re: getting an F statistic for an intermediate step
|
Administrator
|
Would you mind posting your 'final' result? I'm sure others might benefit! Maybe even assist in refinement.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Re: getting an F statistic for an intermediate step
|
|
Uncreatively, I just cannabilized the code for 95% CIs from
Iacobucci, D., & Duhachek, A. (2003). Advancing alpha: Measuring reliability with confidence. Journal of Consumer Psychology, 13, 478-487. They do require manual input of a correlation matrix, which is a bit of a pain/potential error source, so I'll probably fix that. Otherwise it's the same. Mike
On Feb 19, 2012, at 5:39 PM, David Marso wrote:
|
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |