Cronbach’s Alpha

Cronbach’s Alpha is the most commonly used statistic for determining the internal consistency of measurements, such as items in a questionnaire, exam, or survey. It is equivalent to performing the split-half methodology on all combinations of these items.

Topics

References

Tavakol, M., Dennick, R. (2011) Making sense of Cronbach’s alpha. International Journal of Medical Education
https://www.ijme.net/archive/2/cronbachs-alpha.pdf

Gliem, J. A., Gliem, R. R. (2004) Calculating, interpreting, and reporting Cronbach’s alpha reliability coefficient for Likert-type scales
https://scholarworks.indianapolis.iu.edu/server/api/core/bitstreams/976cec6a-914f-4e49-84b2-f658d5b26ff9/content

Newsom, J. (2017) Empirical estimates of reliability: Cronbach’s alpha
https://web.pdx.edu/~newsomj/pmclass/empirical%20estimates%20of%20reliability.pdf

Trizano-Hermosilla, I., Alvarado, J. M. (2016) Best alternatives to Cronbach’s alpha reliability in realistic conditions: congeneric and asymmetrical measurements
https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2016.00769/full

617 thoughts on “Cronbach’s Alpha”

  1. Hello,
    I’m writing a paper that elaborate the result of 36 questions, however my professor would like to eliminate most of them and only take 7 most reliable questions. Can I do that by: trying manually to delete one of the question that result in a higher Cornbach’s alpha ? thanks a lot

    Regards
    Mar

    Reply
  2. Sir,

    Thanks for your efforts! I have a question about missing data and Cronbach’s alpha. I was getting lowish Cronbach’s Alpha scores, around .5, on some scales I created with GSS data. I had converted the qualitative responses to the equivalent of a Likert type of numerical scale. But when I drop all the zeros (which I were the “not applicable,” and “no answer” responses) from my dataset, leaving them blank in the columns, I am getting scores about .7 when Real Stats w/Excel computes Cronbach’s Alpha. Is this an acceptable route to take with missing data?

    Reply
    • Anthony,
      Any row with a empty cell is simply eliminated from the analysis (listwise deletion), which means that all subjects who had one or more N/A responses are dropped. If this is what you want then, the approach should be ok. I would be very cautious, however, if this results in eliminating a substantial percentage of the subjects (i.e. you need to understand why so many are making this response. If one question results in a very high number of N/A, it might be better to eliminate this question from the analysis.
      Charles

      Reply
      • Thank you. The responses that are removed are from subjects who were not asked the question. Off the 1700 or so subjects in each GSS, only 800 or so are asked the specific questions that I am using for my scales. When you download the GSS data from their website, you get ALL respondents, not just the ones who answered your subset.

        Reply
  3. hi charles..
    i have question regarding Cronbach’s alpha.. in my pilot study, one construct revealed that the value was 0.385 but in real study showed that theCronbach’s alpha value was 0.745.. should i continue the instrument due to the value of Cronbach’s alpha was showed acceptable rate in real study.. tq in advance

    Reply
    • Syahril,
      After getting such a low value (.385) in the pilot, you probably should have made some changes to your questionnaire, but if during the real study you are getting a reasonably high value (.745), it looks like your questionnaire has an acceptable level of consistency/reliability.
      Charles

      Reply
  4. Hi Charles,

    In example 4 you calculated Cronbach Alpha for whole survey, which I understand how to do it. But I don’t understand how to calculate alpha for each question in Excel (for example, Q1, Q2, Q3…). Please, help me understand it, I need that for my Master asap.

    Thank you a lot!

    Reply
    • Jelena,
      You calculate it exactly as for Cronbach’s alpha for the whole survey except that you delete one question from the survey. Nothing complicated.
      Charles

      Reply
  5. let suppose one construct have four dimensions, and all four dimensions reliability reported separately, and overall construct reliability is missing? so how we calculate construct reliability. Same is for R value? I face this problem while reporting data in meta analysis.

    Reply
  6. Hi,
    Thanks for the information on this page. I´m new to the subject and have some rather basic questions to ask, hope it´s ok…

    Which formula did you use to calculate the Var in Figure 1? It seems like there´re some different options on Excel and I´m not quite sure which one applies to my case. I tried using VARP (in row 17) but then it would be the same as the calculation done in B21, so might be wrong?

    Did you use the formula SUM() to calculate the Total in Figure 1?

    Our questionnaire has some empty rows (that were not filled out). Is it necessary to exclude empty rows?

    Thank you.
    Mayra

    Reply
    • Myra,
      I used VARP (or VAR.P) for the variance in row 17 as well as in cell B21. You can use VAR (or VAR.S) instead, as long as you use this function in all the variance cells.
      Yes, I used SUM(…) in cells M16 and M17.
      Yes, you should exclude any row that contains one or more cells.
      Charles

      Reply
  7. I asked 05 questions with rating from 1-5. I entered 413 responses.
    My Alpha is -0.245.
    I used Example-2 and its individual question Alphas is as under:-
    -0.41, -0.30, -0.17, -0.10, -0.07 and -0.18
    What i should to do to set alpha .7?

    Reply
    • Imran,
      Obviously, the values for Cronbach’s alpha are far from acceptable. How to address this is described on the referenced webpage. In particular by
      1. Making sure that you are testing one concept only, and if not then you should create separate Cronbach alphas for questions that are associated with different concepts
      2. Correcting for reverse coding
      3. Rephrasing questions to eliminate ambiguity
      Charles

      Reply
    • Lee,
      As described on the referenced webpage, you should first make sure that the low value for Cronbach’s alpha isn’t due to one of the following:
      1. questionnaire that is measuring more than one concept
      2. reverse coding
      If not, then you might want to explore the following corrective actions:
      1. reword some questions
      2. eliminate some questions
      Charles

      Reply
  8. Dear Charles,

    Great information.
    I have 3 questions:
    1. May I know if there any difference of the Cronbach’s alpha using excel and using SPSS in terms of result?

    2. If using observation questionnaire by assessor/researcher, can the Cronbach’s alpha is applied to check the reliability of the item?

    3. If the questionnaire on knowledge with 4 scales likert (1 Strongly not agree to 4 Strongly agree) and with positive and negative items, should I apply a reverse coding?
    (5 minus a recorded score?)

    Thank you.

    Lynn

    Reply
    • Lynn,
      1. I don’t use SPSS and so I can’t comment about SPSS, although it is very likely that the results from SPSS are the same as those calculated using the Real Statistics Excel software.
      2. Cronbach’s Alpha is used to check the reliability in terms of internal consistency of the whole questionnaire, not individual items. It can be used to test whether dropping any particular item dramatically increases internal consistency.
      3.You don’t need to apply reverse coding as long as all the questions use the same coding (with the same meaning). In fact, you should get the same value for Cronbach’s alpha if you reverse code all the questions.
      Charles

      Reply
  9. Hey there,
    great tutorial. Not many good ones around otherwise.
    I did stumble over the formulas depicted in Figure 4. For cell B42, which is supposed to be the variance of totals w/o Q1, you use a formula that calculated the variance of Q1.
    Shouldn’t it be something like “=VARP(M4:M15) – B17”?
    Anyway, keep up the great work.
    Best,
    Sang-Min

    Reply
    • Sang-Min,
      Thanks for your kind remarks.
      The data in column B consists of the totals for all the questions minus the score for Q1, and so the formula in cell B42 doesn’t calculate the variance of Q1.
      Also note that var(x-y) is not equal to var(x) – var(y).
      Charles

      Reply
  10. Hi Charles,

    I’ve conducted a pretest with only 13 responses that asks about categories like knowledge, behaviors, experience, etc. I have a 5-point Likert scale (strongly disagree to strongly agree) for each category, while each category only has about 5 to 7 statements. I grouped all of these together to compute Cronbach alpha, which yielded 0.725. However, if I calculated each category separately, I got a Cronbach alpha of -1.268 for one of the categories alone, for instance. Am I doing this right? Am I able to group all categories together or do I have to compute each category separately? If so, what does the negative mean?

    Thanks,

    Kay

    Reply
    • Kay,
      This sounds strange. Cronbach’s alpha for all questions should be lower than for some subset of these questions. If you send me an Excel file with your data and calcualtions I will try to figure out what is going on. You can find my email address at Contact Us.
      Charles

      Reply
  11. Hi Charles,

    I’ve read over the remedy for negative cronbach’s alpha. However, for some reason I am still ending up with a negative alpha. I even reverse coded the relevant items but I still end up with a negative. My question is “Regarding project objectives; to what extent does your organisation consider the following project objectives important to the success of construction projects?” I used a five point Likert Scale. Each of the items or objectives in the questionnaire are relevant to one another and are testing importance of objective to success of a construction project. What can I do? It’s really throwing off the results for my dissertation

    Reply
    • Hi Shafaq,
      One possible issue is that the questions are not measuring the same concept. If you are measuring different concepts, then you need to perform Cronbach’s Alpha separately on each group of questions which correspond to one of these concepts. In this case, you may have a very low value of Cronbach’s Alpha for the whole questionnaire, but perfectly acceptable values for each subgroup of questions.
      Charles

      Reply
  12. hi
    I am doing project on assessment of QOL. For measuring QOL Iam taking yes and no options for each question.Can I use cronbachs alpha score for my project.

    Reply
    • You can use Cronbach’s Alpha for yes/no questions. If there is a correct answer, you should use 1 for correct and 0 for incorrect.
      Charles

      Reply
      • Hello Charles, I would like to know whether the sample size affects the Cronbach’s alpha ? I had done a pilot survey for my research and i got alpha as.82 which is considered good, but later when I carried my final survey after my final data collection it reduced. forpilot survey my sample size was 80 and now the sample size is 320? I have used likert 5 point scale.

        Reply
  13. Dear Charles,
    Thanks for such a good explanation of Cronbach’s coefficient alpha. Request your assistance in resolving my query. As part of my research paper I have developed a questionnaire & this questionnaire was responded by 100 participants. There are around 10 close ended questions where I have asked if a particular factor has influenced participant’s activity. Response is in the form of ‘Yes’ or ‘No’ which has been coded with 1=Yes & 2=No. Kindly have a look at the sample Q –

    Q. Which of the following factors influence your xxxxx…activity?
    Factor 1 Yes/No
    Factor 2 Yes/No
    Factor 3 Yes/No
    Factor 4 Yes/No

    All these are independent factors & I have received response as Yes or No for these factors from 100 participants.

    My problem is when I am running Cronbach alpha on this I am getting negative alpha.
    Request your assistance on below mentioned queries –
    Q. – Can I use Cronbach alpha on such data? Primarily I want to check consistency & reliability of my response data?
    Q. If Cronbach alpha is not correct measure which one I can use for such a data?

    Thanks for all your help.
    Tanish

    Reply
    • Tanish,
      Getting a low value (and a negative value is clearly a low value) for Cronbach’s Alpha may indicate that the reliability/consistency of your questionnaire is low, but it also can mean that you are testing multiple concepts. Cronbach’s Alpha is only meaningful when the questions are all testing the same concept. If not you need to partition your questionnaire based on the concepts being tested and calculate a separate Cronbach’s Alpha for each group of questions. You can use Factor Analysis to discover the underlying concepts being tested and to partition the questionnaire.
      Charles

      Reply
  14. Hi Charles,

    Thank you for this. But I have a question, I am testing the reliability of my survey questionnaire, the first 2 parts uses likert scale on the scale of 1 to 5 (Strongly Disagree to Strongly Agree) and one additional question that’s answerable on the scale of 1 to 10 (if your familiar with NPS). Is the process still the same?

    Thanks.

    Reply
    • Reynel,
      You can calculate two separate Cronbach’s Alpha values, especially if the two parts measure different concepts.If you want one value, perhaps you can map the 1 to 5 Likert values into 1 to 10 values (e.g. by doubling the value).
      Charles

      Reply
      • Got it Charles. Thanks.

        My problem now is the 2nd and 3rd part which are single-item scale. Based on my reviews, I need to do a test-retest reliability as SPSS cannot cater these single-item questions.

        Reply
  15. Hi Charles,
    Thanks for excellent answers that clarify the concepts. My question : “Is it ok to assign 1 for strongly agree and 5- for strongly disagree in a 5-point likert scale? Will the Cronbach alpha value differ if I assign 5-strongly agree and 1- strongly disagree ? Also, for 100 nos. responses to say 36 question parameters, will the cronbach alpha value remain same for either of above mentioned scale designs? Is it sufficient to ensure that cronbach alpha is more than 0.7 to ensure that the questionnaire is both consistent/reliable and construct-valid?”

    Reply
    • Sanjay,
      1. I don’t think the value of Cronbach’s alpha will differ, but it is easy enough for you to try it both ways and compare the results.
      2. More than 0.7 is often considered to be sufficient to consider the questionnaire to be consistent/reliable. This does not mean that the questionnaire is valid. The threshold of 0.7 is not universally agreed upon.
      Charles

      Reply
  16. Dear Charles
    On your Example 1 – how do I calculate the EVar in cell M17? Can you please point me in the right direction as to how excel can achieve this?
    Regards Richard

    Reply
  17. One of my dimensions of a scale has 2 items which yielded an internal consistency of .42. How could I solve such a problem? Can I report the inter-item correlation which is .36?

    Reply
  18. Hello. I am using Cronbach’s Alpha in a psychometric study. What kind of figure should i use to illustrate alpha? Maybe a covariance table or covariance 3d figure ? Thanks

    Reply
    • Giorgos,
      Sorry, but I don’t understand what you mean by “illustrate alpha”. Generally, alpha is set to 0.05.
      Charles

      Reply
  19. Dear Charles,
    I am working with 251 subjects using 32 questions (including 7 questions on the demographic characteristics). The questionnaire is more of rating scale style (agree, strongly agree, disagree and strongly disagree). please, I need assistance on how to calculate the reliabilty (formula) using Half split method

    Best Regards.

    Christy.

    Reply
    • Christy,
      You probably don’t want to include the questions regarding demographic characteristics in the calculation since they are of a completely different nature from the other questions. The referenced webpage shows how to calculate Cronbach’s alpha for questions using Likert scales. Here, for example, you can use the coding 4 = strongly agree, 3 = agree, 2 = disagree and 1 = strongly disagree. Even better is to use the scale 5 = strongly agree, 4 = agree, 3 = neutral, 2 = disagree and 1 = strongly disagree.
      You can also perform the split half method using this coding as described on the webpage
      Split Half Method
      Charles

      Reply
  20. Hello Sir

    I’m thinking of computing Cronbach’s alpha to test the reliability of the multiple choice tests that i prepare for my students’ assessments in the same way you presented in Example 1 ( with the codes: 1 = correct answer and 0 = incorrect answer)
    but i was wondering if there is a way to calculate Cronbach’s alpha if i can provide the total number of students who got the correct answer for each question only

    best regards
    thanks in advance for your valuable reply

    Reply
  21. Hi Sir,
    Please how can i get a simulated data for 42 variables with good chronbach alpha as well to satisfy average variance criteria >.7 or =.7.
    Thank you Sir.

    Reply
        • Andrew,
          I can probably come up with an example of the type you requested, but I can’t think of a quick way to do this.
          Charles

          Reply
          • Andrew,
            I just sent you an example. The example consists of a range B2:AO101 containing 0’s and 1’s. The formula for each cell in the first column is =RANDBETWEEN(0,1). The formula in cell C2 (the first cell in the second column) is =IF(RAND()>.2,B2,1-B2). I copied this formula into all the other cells in the range. The result is Cronbach’s alpha of .725515.
            I then copied the whole range and pasted values so that the cell values wouldn’t change.
            I can’t guarantee that this approach will work in general, but it seemed to make some sense when I did it.
            Charles

  22. hi,
    My thesis’ Cronbach alpha is below 0.700, how can I increase reliability coefficient cronbach alpha? i try to remove question or so on..it still under 0.700…what i can do to increase it?may i forward my data to seek you advice..thanks

    Thank you

    Reply
    • Ting,
      The usual approach to increasing Cronbach’s alpha is to change the wording of some questions before you give out your questionnaire to a large group of respondents.
      Just because Cronbach’s alpha is less than .700 doesn’t make the questionnaire bad; some would use a cutoff of .600 anyway. If removing a question substantially increases Cronbach’s alpha then I might remove that question.
      Charles

      Reply
        • Ting,
          This value is generally considered to be low. This doesn’t mean that the questionnaire is bad, just that there is some kind of internal inconsistency. The low Cronbach’s value may be caused by many things, including that it is measuring more than one thing — this is not necessarily a bad thing.
          Charles

          Reply
    • Hello,
      I also have the same problem, cronbach alpha value of my data is 0.44, how could it be increased? If i continue with the same data, will it be ok? Or what kind of problems i ll have in regression anaysis?
      Thanks

      Reply
      • Noma,
        As described on the referenced webpage, you should first make sure that the low value for Cronbach’s alpha isn’t due to one of the following:
        1. questionnaire that is measuring more than one concept
        2. reverse coding
        If not, then you might want to explore the following corrective actions:
        1. reword some questions
        2. eliminate some questions
        Cronbach’s Alpha measures internal consistency, but a low value is not necessary for many statistical tests.
        Charles

        Reply
  23. hai i am nur hashida

    which type of realiability suitable for testing the realibility of parameter estimation method? is it cronbach alpha suitable to use? Tq

    Reply
    • i am working with 267 simulated of data to estimate the drift and diffusion parameter for three types of one-dimensional linear stochastic differential equation models. we use three type of parameter estimation method and compare the reliability of method. Can you suggest suitable type of reliability to be used in our master research. Tq so much

      Reply
      • I understand that you have three different methods of estimating parameters. What do you mean by reliability in this case? Do you want to dtermine whether the three methods yield the same results?
        Charles

        Reply
        • Dear Charles ,
          Thank you so much for this great information. I would like to know if the calculated Cronbach’s Alpha was 1.000 what does it mean? Should I remove some of the questions?

          Regards

          Reply
          • Susan,
            The measurement indicates 100% reliability/consistency, which is a bit suspect. This indicates either that the questions are somewhat redundant or that the number of people who answered the questionnaire/exam is very small. Removing questions typically increases the value of Cronbach’s alpha (which is not possible in this case) not reduce it.
            Charles

    • Cronbach’s alpha is not used for parameter estimation.
      In this case, you may be looking for “goodness of fit” – e.g. chi-square goodness of fit test. The specific test to use depends on which parameter you are referring to.
      Charles

      Reply
  24. Good Evening-I am using an inventory that measures two factors: knowledge of cognition and regulation of cognition. There are 12 items, 6 of which measure knowledge of cognition and 6 which measure regulation of cognition. As long as I know which questions are measuring what factor, can I keep these in one inventory and separate by item number when it comes time to run the Alpha Chronbach? Thank you!

    Reply
    • Susanne,
      Yes, but you need to calculate two separate Cronbach alphas, one for the 6 knowledge of cognition questions and another for the other 6 question.
      Charles

      Reply
  25. Hello, we are trying to measure the reliability of a pain scale, where the nurses valuate 5 different features of the newborns. The problem is that we have different nurses(around 15) that valuate 5 different features(it has to be valuated between 0 and 2) on about 15 newborns. It seems like 3 different variables, how do we write our data into the SPSS and calculate the Cronbach alpha of the whole scale? It would be clear if there was only 1 newborn

    Reply
    • Emma,
      I don’t completely understand the scenario you are describing. If there are 5 different features, why are there only 3 variables?
      Why have you decided to use Cronbach’s alpha? Are you trying to see whether the nurses agree in their evaluations?
      Charles

      Reply
  26. I have a construct consist of three items and Cronbach alpha is 0.564, how can I increase reliability coefficient cronbach alpha

    Reply
    • Tahir,
      As described on the referenced webpage, first of all make sure that the Cronbach’s alpha value isn’t being reduced due to problems with reverse coding and/or multiple factors. Also check to see what happens if one of the questions is removed (as described on the referenced webpage). If removing one of the questions results in a much higher value for Cronbach’s alpha, then consider rewording or eliminating that question.
      Charles

      Reply
  27. Hi Charles! What if my questionnaire is a 9-item quiz-type one and for every question, it corresponds to either of the three: easy, average, and difficult, the scoring of which are 1 for easy, 2 for average, and 3 for difficult? Is Cronbach’s alpha applicable to this type of questionnaire? If not, can you suggest other reliability test that I may use to know its validity? Thanks in advance!

    Reply
    • Pat,
      You can use Cronbach’s alpha to measure reliability if for each question you use code 1 for the correct answer and 0 for the incorrect answer. I’m not sure how best to integrate this approach with easy, average, difficult. Also, kind in mind that Cronbach’s alpha is used measure reliability, not validity.
      Charles

      Reply
  28. Hi
    I counted the total number of participation in 5 enrichment activities (piano, art, sport, etc) that are based on yes-no questions (took a lesson = yes, didn’t take that lesson = no). The score ranges from 0 to 5 with more than 50% for 0. Cohen’s alpha is really low like .40. To me, theoretically, it doesn’t seem to make to sense to run cohen’s alpha because internal consistency is not important (a student who takes a piano activity may not take a sport activity, but at the same time, the total number of activities show students’ degrees of engagement in those activities. Should I still run cohen’s alpha? I would really appreciate your advice.

    Reply
    • Hi,
      I assume that you are referring to Cronbach’s alpha and not Cohen’s alpha.
      You are not forced to run Cronbach’s alpha. If internal consistency is not important then you don’t need to use Cronbach’s alpha.
      Charles

      Reply
  29. Hi Charles,

    Thank you so much for this wonderful explanation!
    I did have one question, though.

    I am doing a questionnaire about purchase intentions of genuine and counterfeit bags with a lot of logos or no logos. If a question had two pictorial choices with one showing preference for a bag with a lot of logos and the other with no logos (I hope this makes sense…), how would you code a question with these two answer choices? Would it be 0 and 1 or 1 and 2? or 1 and 7 (my other questions were 7-point Likert scales)? Would this even be eligible for a Cronbach’s Alpha test?

    If you could help me out, that would be greatly appreciated.

    Many thanks!
    Lily

    Reply
    • Lily,
      I believe that you will get the same value for Cronbach’s alpha whether you use a 0-1 coding, a 1-2 coding or a 1-7 coding.
      Cronbach’s alpha can handle a 7-point Likert scale, although you will need to be careful about mixing questions from this scale with questions using a 0-1 or 1-2 scale.
      Charles

      Reply
    • Vahid,
      Yes, Cronbach’s alpha can be used with Yes/No questions. If these questions test someone’s knowledge, then code the question as 1 if the answer is correct and 0 if the answer is incorrect (irrespective of whether it is Yes or No).
      Charles

      Reply
  30. Dear Charles,

    Thanks for this great explanation. I was actually already familiar with the contant, but I used it to again check my data, because I have a problem I can’t seem to solve. My alpha keeps being 1,01 (which is theoretically impossible, right?). I know it’s very difficult to give advice if you don’t see my dataset and used formulas, but I’ll give it a try: do you have any advice for me on which things to check? I already tried calculating things manually, but I keep coming up with the same number.
    (My dataset looks exactly like figure 1 and I used the formulas from figure 2.)

    Thanks very much in advance!

    Reply
  31. I need to know that the Cronbach’s Alpha Test is applied only on the questions of independent variables??? or it include questions both dependent and independent variables?Please reply as soon as possible.

    Reply
    • Sardar,
      You can perform Cronbach’s alpha on any data that you like. Whether something is an independent variable or a dependent variable often depends on the context.
      The real question is why do you want or need to perform Cronbach’s alpha at all?
      Charles

      Reply
  32. Please I want to know if i can check the reliability of a “writing test” using this technique or it is just used for questionnaires!!!

    Reply
  33. pls am doing my project in which I want how to analysis it am having 20 questionnaire and it contains 17 questions each and respondent has to choose from strongly Agree or Agree or strongly disagree or disagree in their responses. how I calculate it using alpha

    Reply
    • Abimbola,
      You can calculate Cronbach’s alpha with Likert scale items as shown in Example 4 of the referenced webpage. Alternatively you can use the Real Statistics capabilities described on that webpage.
      Charles

      Reply
  34. Can we use Cronbach’s Alpha to see if there is an agreement in the ratings (of features of some products) of two groups (gender groups or age groups for example)? If yes, how much sample size is enough for 6 features (rating 1 to 5).

    Reply
    • Learner,
      Cronbach’s alpha is not used for this purpose. To determine whether there is agreement in rating, you could use Cohen’s kappa, Fleiss’s kappa, ICC, Kendall’s W, etc., depending on the specifics of what you are trying to do. I suggest that you look at the website for more details about each of these approaches and see which one best fits with your needs. See the webpage
      Reliability
      Charles

      Reply
      • Many thanks Professor. Earlier I found use of kappa coefficient and I’ve tried different measures like Cohen’s kappa, Fleiss’s kappa and Krippendorff’s Alpha, Kendall’s tau etc. using R, all gave me similar results regarding agreement in the ratings (of features of some products) of two groups (gender groups or age groups for example).
        But I was wondering, what is the fundamental difference (difference between Cohen’s kappa, Fleiss’s kappa is clear to me). How to choose between these coefficient?
        I found your way of explaining things is the best.

        Reply
        • Learner,
          A key difference relates to the type of input data. E.g. Cohen’s kappa supports two raters with categorical ratings, Fleiss’s kappa supports two or more raters with categorical ratings, Weighted kappa supports two raters with ordered ratings, ICC supports any number of raters with interval/continuous ratings.
          Charles

          Reply
          • Thanks a ton Professor. So simple explanation!!
            What I could understood is that ICC is used to measure internal consistency of the raters (within each group) and the measures mentioned by you (Cohen’s kappa, Fleiss’s kappa and weighted kappa) measures agreement between
            two or more group of raters. Please correct, if I am wrong.

  35. I have spent all day reading psychometric testing books and this is the most helpful thing I’ve read on Cronbach’s alpha. Thank you so much – very clear.

    Reply
  36. Hello,
    I know that if you have several dimensions, you calculate Cronbach alpha for each dimension. But can you calculate it for the whole test? In case you want to know reliability for the test as whole. Or do you rather conclude on the whole reliability based on reliability of each dimension?
    Thank you very much

    Reply
    • If there are several dimensions you should use the Cronbach’s alpha for each dimension. Cronbach’s for the whole test won’t be that meaningful. If you need one value, you could use the smallest of the Cronbach’s alpha values, but it is better to simply use Cronbach’s for each dimension.
      Charles

      Reply
  37. I have two groups of questions; First is Functional form of question (Positive side) , for example: If a door is closing how do you feel. One answer be selected by the user 1. I like 2. It must be that way, 3. I am Neutral, 4. I can live that way, 5. I dislike it that way
    Second question: Dysfunctional (negative): If a door is not closing how do you feel. The answer be selected by the user 1. I like 2. It must be that way, 3. I am Neutral, 4. I can live that way, 5. I dislike it that way. How to can I certainly use Cronbach’s alpha

    Reply
    • Shri,
      If some of the questions reverse the scale then you should reverse the coding (i.e. 1 becomes 5, 2 becomes 4, 4 becomes 2 and 5 becomes 1), but I can’t tell from your example whether this is really a case of reverse coding or not.
      Charles

      Reply
  38. I would like to know the answer for this question. Anyone who knows answer this.
    Question: If you are provided with questionnaire consisting of 20 questions and you have the response in the form of a likert scale value(1 to 5) obtained form 25 respondents, then how will you find the factor loadings value?

    Reply
  39. Hi Charles,

    I want to know the minimum sample size required to safely calculate the test reliability (cronbach alpha) in a 4 choice mcq test consisting of 3 sections with 20 questions in each section. Each section measures different skills.

    Also, kindly explain the method so I can determine the same for different variants of the test.

    Regards

    Reply
  40. I am new to this site/blog but it is exciting.
    I am involved in HIV research at operation level. So at time we wonder how to measure internal consistence. For example we have this question:
    In the last 12 months, have you experienced any of the following feelings because you are living with or affected by HIV? (Tick as many boxes as apply ) (read out)

    1. I feel ashamed
    2. I feel guilty
    3. I blame others
    4. I blame my self
    5. I have low self esteem
    6. I feel i should be punished etc.

    All these are items that measure what is termed as Internal HIV stigma.
    Kindly help guide me how to measure the internal consistency. This will guide whether I should drop some items.

    Reply
    • Essentially you have 6 questions with a response of ticked or not ticked. You can assign the code 1 for the ticked items and 0 for non ticked items. This is the coding assuming that all the ticked items represent Internal HIV stigma; if some of the items are reverse coded (i.e. not ticked represents Internal HIV stigma) then simply interchange the roles of 0 and 1 for those items. You then calculate Cronbach’s alpha as described on the referenced webpage.
      Charles

      Reply
  41. Hi charles,
    im doing an experiment about voice perception.. i use likert scale in in 5 question. There are five voices. The questions are do u find the voice trusthworthy dominant aggressive etc. Can this method help me solve this?pls. A student from philippines

    Reply
    • It depends on what you mean by “Can this method help me solve this?”. What is it that you want to solve?
      You can certainly use Cronbach’s alpha with a Likert scale.
      Charles

      Reply
  42. Hi Charles,

    I am questioning if I did my Cronbach’s Alpha test correct. I have one cell with a score of -7.113. The rest of the scores fall between 2.25 and 0.22.

    Should/can there be negative numbers? And should I be concerned about my one cell that has a score of 7?

    I realize this may be an impossible question for you to answer without my work (most questions are a Likert scale with a range of 1 – 7).

    Thank you in advance.

    Reply

Leave a Comment