question about measure creation statistics
Locked
 
|
Hello-
I'm creating a measure that currently has 20 items. 18 of these items are multiple choice, 4 possible answer questions. 2 of the items have 7 possible answers. All of the items are covering the same broad area (knowledge of psychological disorder criteria). I ran cronbach's alpha and the score was .78. I talked to my advisor, who said this was incorrect and I should have run KR-20. I did some reading, and two things appear to be true. First, that in fact, I should be using KR-21 since the answers are not dichotomous. Second, that there is no way to do this in SPSS. Would LOVE some guidance. Michael |
Re: question about measure creation statistics
Administrator
|
SPSS can calculate just about anything. Please provide formula for KR-21 coefficient (I have not studied psychometrics for about 20+ years) and don't have the time to google it.
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: question about measure creation statistics
|
|
In reply to this post by mmiesner
Michael,
If you score the responses true/false the items are dichotomous. If not please elaborate on the scoring rules of the items. KR-21 is a special case of the reliability coefficient with the (general unrealistic) assumption of equal item difficulties (means). Cronbach's alpha and KR-20 are identical. The only distinction is that KR-20 is somewhat easier to compute, something that was useful when calculation had to be done by hand. It is perfectly valid to use the reliability command in SPSS to compute the "alpha" on dichotomous variables and call it KR-20. Paul ===================== 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: question about measure creation statistics
|
|
In reply to this post by mmiesner
Explain what you mean by the answers are not dichotomous. A multiple choice knowledge question has a right answer and several wrong answers but the scoring is dichotomous, right (=1) wrong (=0). Cronbach's alpha in this case gives exactly the same result as the KR-20. It is just a generalization of the Kr-20 to items that are not dichotomous. As the other Paul said, the KR-21 has unreasonable assumptions and was only used because at the time they were developed (1930s) simplicity of computation was of concern. In other words, your advisor is wrong.
Paul Dr. Paul R. Swank, Children's Learning Institute Professor, Department of Pediatrics, Medical School Adjunct Professor, School of Public Health University of Texas Health Science Center-Houston -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of mmiesner Sent: Thursday, November 17, 2011 6:47 PM To: [hidden email] Subject: question about measure creation statistics Hello- I'm creating a measure that currently has 20 items. 18 of these items are multiple choice, 4 possible answer questions. 2 of the items have 7 possible answers. All of the items are covering the same broad area (knowledge of psychological disorder criteria). I ran cronbach's alpha and the score was .78. I talked to my advisor, who said this was incorrect and I should have run KR-20. I did some reading, and two things appear to be true. First, that in fact, I should be using KR-21 since the answers are not dichotomous. Second, that there is no way to do this in SPSS. Would LOVE some guidance. Michael -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/question-about-measure-creation-statistics-tp5002888p5002888.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: question about measure creation statistics
|
|
In reply to this post by David Marso
KR-20 is for dichotomous scored items, asuming all have the same dificulty Does all your items have the same dificulty? Andrés Mg. Andrés Burga León Coordinador de Análisis e Informática Unidad de Medición de la Calidad Educativa (UMC) Ministerio de Educación del Perú Av.de la Arqeuología s/n (cuadra 2) Lima 41 Perú Teléfono 615-5840 - 6155800 anexo 1212 http://www2.minedu.gob.pe/umc/
SPSS can calculate just about anything. Please provide formula for KR-21 coefficient (I have not studied psychometrics for about 20+ years) and don't have the time to google it. mmiesner wrote: > > Hello- > I'm creating a measure that currently has 20 items. 18 of these items are > multiple choice, 4 possible answer questions. 2 of the items have 7 > possible answers. All of the items are covering the same broad area > (knowledge of psychological disorder criteria). > > I ran cronbach's alpha and the score was .78. > I talked to my advisor, who said this was incorrect and I should have run > KR-20. I did some reading, and two things appear to be true. > > First, that in fact, I should be using KR-21 since the answers are not > dichotomous. > Second, that there is no way to do this in SPSS. > > Would LOVE some guidance. > > Michael > -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/question-about-measure-creation-statistics-tp5002888p5003057.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 |
Re: question about measure creation statistics
|
|
Sorry for the last post, I've mixed 20 and 21. Anyway, Both are for dichotoimous items, and KR-20 is mathematically equivalent to Cronbachs alpha Michael, you said: > First, that in fact, I should be using KR-21 since the answers are not > dichotomous. If your answres are not dichotomous you should not be using any KR reliability estimate. Cronbacdh alpha is both for dichotomous and polytomous scored items. Andrés Mg. Andrés Burga León Coordinador de Análisis e Informática Unidad de Medición de la Calidad Educativa (UMC) Ministerio de Educación del Perú Av.de la Arqeuología s/n (cuadra 2) Lima 41 Perú Teléfono 615-5840 - 6155800 anexo 1212 http://www2.minedu.gob.pe/umc/
That’s the KR-21 that does that. KR-20 assume that all items have the same relation to the true score. Dr. Paul R. Swank, Children's Learning Institute Professor, Department of Pediatrics, Medical School Adjunct Professor, School of Public Health University of Texas Health Science Center-Houston From: SPSSX(r) Discussion [[hidden email]] On Behalf Of ANDRES ALBERTO BURGA LEON Sent: Friday, November 18, 2011 2:27 PM To: [hidden email] Subject: Re: question about measure creation statistics KR-20 is for dichotomous scored items, asuming all have the same dificulty Does all your items have the same dificulty? Andrés Mg. Andrés Burga León Coordinador de Análisis e Informática Unidad de Medición de la Calidad Educativa (UMC) Ministerio de Educación del Perú Av.de la Arqeuología s/n (cuadra 2) Lima 41 Perú Teléfono 615-5840 - 6155800 anexo 1212 http://www2.minedu.gob.pe/umc/
SPSS can calculate just about anything. Please provide formula for KR-21 coefficient (I have not studied psychometrics for about 20+ years) and don't have the time to google it. mmiesner wrote: > > Hello- > I'm creating a measure that currently has 20 items. 18 of these items are > multiple choice, 4 possible answer questions. 2 of the items have 7 > possible answers. All of the items are covering the same broad area > (knowledge of psychological disorder criteria). > > I ran cronbach's alpha and the score was .78. > I talked to my advisor, who said this was incorrect and I should have run > KR-20. I did some reading, and two things appear to be true. > > First, that in fact, I should be using KR-21 since the answers are not > dichotomous. > Second, that there is no way to do this in SPSS. > > Would LOVE some guidance. > > Michael > -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/question-about-measure-creation-statistics-tp5002888p5003057.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 |
Re: question about measure creation statistics
Administrator
|
In reply to this post by ANDRES ALBERTO BURGA LEON
It is KR-21, not KR-20, that assumes all items are equally difficult. See Paul O's earlier post:
http://spssx-discussion.1045642.n5.nabble.com/question-about-measure-creation-statistics-tp5002888p5003728.html
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
Re: question about measure creation statistics
|
|
In reply to this post by ANDRES ALBERTO BURGA LEON
Cronbach's alpha serves as a reliability coefficient both for dichotomous
and scaled-score items (not, "polytomous" items). I gather that both KR-20 and KR-21 are used for dichotomies, where KR-21 is a simpler formula for hand calculation, usable under an unlikely assumption. No one would use it these days, since no one bothers with the tedium and inaccuracy of hand-calculations. KR-20 matches one version of Cronbach's alpha. "Polytomous" describes categorical items when there are more than two categories. If they are "ordered" categories, I would always say "ordered" or "scaled". -- Rich Ulrich Date: Fri, 18 Nov 2011 16:06:43 -0500 From: [hidden email] Subject: Re: question about measure creation statistics To: [hidden email] Sorry for the last post, I've mixed 20 and 21. Anyway, Both are for dichotoimous items, and KR-20 is mathematically equivalent to Cronbachs alpha Michael, you said: > First, that in fact, I should be using KR-21 since the answers are not > dichotomous. If your answres are not dichotomous you should not be using any KR reliability estimate. Cronbacdh alpha is both for dichotomous and polytomous scored items. Andrés [snip, previous] |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |