iSixSigma

reliability

Six Sigma – iSixSigma Forums Old Forums General reliability

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #48528

    madiha
    Participant

    the values in °F are exactly equal to 1.8 times °C plus 32.
    if we make to columns for temperature in °C and one its corresponding values in °F and the run a reliability analysis for the two tables, value is never equal to 1, please explain this to me.

    0
    #164040

    qualitycolorado
    Participant

    Madiha, Good morning!You indicated that your are running “a reliability analysis” on your temperature conversion situation — that covers a lot of territory!! — what specific test are you running? That may help us troubleshoot this with you …
    Best regards,
    QualityColorado

    0
    #164052

    madiha
    Participant

    well its not any project, i just wanted to clear my concept of reliability and wanted to know why the reliability co-effecient was not constant when readings are calculated using the same formula.
    If you can shed some light on it i would be grateful.

    0
    #164053

    Reliability
    Participant

    The Fahrenheitscale is a linear transformation of Celsius. The correlation is therefore 1. However, in a split-half reliability, the Cronbach alpha is a lower estimate of the reliability coefficient. Depending on the splits, the reliability estimates will be different even though you are using the same statistical coefficient to estimate the reliability. I think that this is what you are referring to.

    0
    #164067

    fake accrington alert
    Participant

    I  f  you  drive  your  car daily without troubles,then  your  car  is  reliable 

    0
    #164088

    madiha
    Participant

    thanku very much or giving me an inview regarding reliability of both scales. i really helps clear some confusion.

    0
    #164089

    madiha
    Participant

    Can you please shed some light on tis case too
    The following are scores obtained by four persons on two alternate versions of a test. Based on these scores, the alternate forms reliability of a single observed score is .83 (that is, rX = ICC(A, 1) = .83). It was later discovered that Aleksandra’s score on Version B was actually 41 rather than 45 as recorded in the table. This change increases the level of unsystematic error. Does this change affect the level of systematic bias?

    Aleksandra 43 Version A 45 Version B Beatrix 50 Version A 52 Version B Colleen 52 Version A 50 Version B Danna 55 Version A 61 Version B Mean 50 Version A 52 Version B

    0
    #164100

    Reliability
    Participant

    First of all, there is the difference between systematic measurement error and selection bias in selection.
    Systematic measurement error is defined as those sources of error that “consistently affect an individual’s score because of some particular characteristic” (gender etc.). Systematic error is estimated via validity studies because it is part of the true score variance. My take is that based on your research design, you cannot determine if systematic error went up or down. You have an estimate of the error component, but not of the true score component which consists of variance in the relevant and irrelevant characteristics:
    True score = Relevant (R) + Irrelevant Characteristics (I)
    Total score = R + I + Error
    Validity studies always require an external criterion, which you don’t have in this data.
    Selection bias is estimated via a regression of the minority group vs. majority group (male vs. female … don’t ask me which is the majority as opposed to the minority :-).
    Jensen defines unbiased tests as follows: (a) the regression lines are the same for both groups, (b) the standard errors of prediction are the same for both groups. It’s impossible to determine this based on the data because the comparison group is missing. So, no determination of selection bias is possible.  
     

    0
    #164107

    faux accrington alert
    Participant

    Every day I drive my car I have trouble.  I hit a parked car once, I hit a telephone pole once, I fell asleep at the wheel and drove into a pond – my car is very unreliable, nothing but trouble every day I drive it!  Thank you for the clear explanation – I will go right to the dealer and complain about my unreliable car!

    0
    #164121

    New ATI
    Participant

    I  believe  you  are  not OK,your  car  is  reliable and  fine.
    Instead you  have to  visit  a special clinic  to  check  your  unreliable  personality.
    My 2  cents  for  free
    Have  a  nice  day 

    0
Viewing 10 posts - 1 through 10 (of 10 total)

The forum ‘General’ is closed to new topics and replies.