Scribd Logo
Upload

Welcome to Scribd!

  • 113 views

    Cronbach Alpha

    Uploaded by

    Rendy Edisti
    Cronbach's alpha is a measure of internal consistency or reliability of a psychometric test. It is commonly used to assess how well a set of items measures a single unidimensional latent construct. Cronbach's alpha is calculated using a measure of the average inter-item covariance among the items measuring the construct divided by the variance of the total score. A higher alpha suggests better reliability, with values closer to 1 indicating greater internal consistency. Factors such as test length, item difficulty, and item correlations can impact the value of Cronbach's alpha.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd

    Cronbach Alpha

    Uploaded by

    Rendy Edisti
    113 views15 pages
    Cronbach's alpha is a measure of internal consistency or reliability of a psychometric test. It is commonly used to assess how well a set of items measures a single unidimensional latent construct. Cronbach's alpha is calculated using a measure of the average inter-item covariance among the items measuring the construct divided by the variance of the total score. A higher alpha suggests better reliability, with values closer to 1 indicating greater internal consistency. Factors such as test length, item difficulty, and item correlations can impact the value of Cronbach's alpha.

    Original Description:

    how to use cronbach alpha

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PPT, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Cronbach's alpha is a measure of internal consistency or reliability of a psychometric test. It is commonly used to assess how well a set of items measures a single unidimensional latent construct. Cronbach's alpha is calculated using a measure of the average inter-item covariance among the items measuring the construct divided by the variance of the total score. A higher alpha suggests better reliability, with values closer to 1 indicating greater internal consistency. Factors such as test length, item difficulty, and item correlations can impact the value of Cronbach's alpha.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    113 views15 pages

    Cronbach Alpha

    Uploaded by

    Rendy Edisti
    Cronbach's alpha is a measure of internal consistency or reliability of a psychometric test. It is commonly used to assess how well a set of items measures a single unidimensional latent construct. Cronbach's alpha is calculated using a measure of the average inter-item covariance among the items measuring the construct divided by the variance of the total score. A higher alpha suggests better reliability, with values closer to 1 indicating greater internal consistency. Factors such as test length, item difficulty, and item correlations can impact the value of Cronbach's alpha.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    of 15

    Explaining Cronbachs Alpha

    Kirk Allen Graduate Research Assistant [email protected] University of Oklahoma Dept. of Industrial En ineerin
    School of Industrial Engineering - The University of Oklahoma

    !hat is alpha and "hy should "e care#


    $ %ron&ach's alpha is the most commonly used measure of relia&ility (i.e.) internal consistency*. $ It "as ori inally derived &y Kuder + Richardson (,-./* for dichotomously scored data (0 or ,* and later enerali1ed &y %ron&ach (,-2,* to account for any scorin method. $ 3eople kno" that a hi h alpha is ood) &ut it is important to have a deeper kno"led e to use it properly. 4hat is the purpose of this presentation.

    School of Industrial Engineering - The University of Oklahoma

    Other

    types of relia&ility

    $ 4est5Re64est
    7 4he same test is taken t"ice.

    $ E8uivalent 9orms
    7 Different tests coverin the same topics 7 %an &e accomplished &y splittin a test into halves

    School of Industrial Engineering - The University of Oklahoma

    %ron&ach's

    &asic e8uation for alpha

    n Vi = , n , Vtest
    $ n : num&er of 8uestions $ ;i : variance of scores on each 8uestion $ ;test : total variance of overall scores (not <'s* on the entire test

    School of Industrial Engineering - The University of Oklahoma

    =o"

    alpha "orks

    $ ;i : pi > (,6pi*
    7 pi : percenta e of class "ho ans"ers correctly 7 4his formula can &e derived from the standard definition of variance.

    $ ;i varies from 0 to 0.?2


    pi 0 0.?2 0.2 ,6pi , 0./2 0.2 ;i 0 0.,@/2 0.?2

    School of Industrial Engineering - The University of Oklahoma

    =o"

    alpha "orks

    $ ;test is the most important part of alpha

    n Vi = , n , Vtest
    $ If ;test is lar e) it can &e seen that alpha "ill &e lar e alsoA
    7 Bar e ;test Cmall Ratio D;i5;test this small ratio from , hi h alpha Cu&tract

    School of Industrial Engineering - The University of Oklahoma

    =i

    h alpha is ood. =i h alpha is caused &y hi h variance. Eut "hy is hi h variance ood#
    $ =i h variance means you have a "ide spread of scores) "hich means students are easier to differentiate. $ If a test has a lo" variance) the scores for the class are close to ether. Unless the students truly are close in a&ility) the test is not useful.

    School of Industrial Engineering - The University of Oklahoma

    !hat makes a 8uestion FGoodG or FEadG in terms of alpha#


    $ C3CC and CAC "ill report Falpha if item deletedG) "hich sho"s ho" alpha "ould chan e if that one 8uestion "as not on the test. $ Bo" Falpha if item deletedG means a 8uestion is ood &ecause deletin that 8uestion "ould lo"er the overall alpha. $ In a test such as the C%I (.H items*) no one 8uestion "ill have a lar e deviation from the overall alpha.
    7 Usually at most 0.0. in either direction

    School of Industrial Engineering - The University of Oklahoma

    !hat

    causes a 8uestion to &e FEadG# Iuestions "ith hi h Falpha if deletedG tend to have lo" inter6item correlations (3earson's r*.

    School of Industrial Engineering - The University of Oklahoma

    How Negative Correlations affect alpha


    0 02$ R2 = 0 !"2" 0 02 Change in Alpha (positive=good) 0 0#$ 0 0# 0 00$ 0 -0 2 -0 # -0 00$ -0 0# -0 0#$ -0 02 Average Inter-Item Correlation 0 0# 02 0%

    School of Industrial Engineering - The University of Oklahoma

    !hat causes lo" or ne ative inter6item correlations#


    $ !hen a 8uestion tends to &e ans"ered correctly &y students "ho have lo" overall scores on the test) &ut the 8uestion is missed &y people "ith hi h overall scores. $ 4he F"ron G people are ettin the 8uestion correct.

    Iuantified &y the F apG &et"een correct and incorrect students $ %orrect studentsA avera e score ,2.0 $ Incorrect studentsA avera e score ,?.2 $ Gap : ,2.0 $ ,?.2 : ?.2

    School of Industrial Engineering - The University of Oklahoma

    Change in Alpha vs. "Gap"


    0 02$ 0 02 Change in Alpha (positive=good) 0 0#$ 0 0# 0 00$ 0 -$ -0 00$ -0 0# -0 0#$ -0 02 core of Correct !in"s core of Incorrect 0 $ #0 #$ R2 = 0 &'!!

    School of Industrial Engineering - The University of Oklahoma

    If a 8uestion is F&adG) this means it is not conformin "ith the rest of the test to measure the same &asic factor (e. .) statistics kno"led e*.
    $ 4he 8uestion is not Finternally consistentG "ith the rest of the test.

    3ossi&le causes (&ased on focus roup comments*


    $ Ctudents are uessin (e. .) 8uestion is too hard*. $ Ctudents use test6takin tricks (e. .) correct ans"er looks different from incorrect ans"ers*. $ Iuestion re8uires a skill that is different from the rest of the 8uestions (e. .) memory recall of a definition*.

    School of Industrial Engineering - The University of Oklahoma

    =o" does test len th FinflateG alpha# 9or eJample) consider dou&lin the test len thA
    $ ;test "ill increase &y a po"er of H &ecause variance involves a s8uared term. $ =o"ever) D;i "ill only dou&le &ecause each ;i is Kust a num&er &et"een 0 and 0.?2. $ Cince ;test increases faster than D;i (recall that hi h ;test is ood*) then alpha "ill increase &y virtue of len thenin the test.

    School of Industrial Engineering - The University of Oklahoma

    References

    Kuder + Richardson) ,-./) F4he 4heory of the Estimation of 4est Relia&ilityG (Psychometrika v. 2 no. 3* %ron&ach) ,-2,) F%oefficient Alpha and the Internal Ctructure of 4estsG (Psychometrika v. ! no. 3* %ortina) ,--.) F(hat is coefficient al)ha* +n e,amination of theory and a))lications- (". of Applie# Psych. v. $% no. p. &%' ()* Ctreiner) ?00.) FCtartin at the Ee innin A An Introduction to %oefficient Alpha and Internal %onsistencyG (". of Personality Assessment v. %( no. p. &&' (3*

    School of Industrial Engineering - The University of Oklahoma

    You might also like