This is definitely my favorite statistic. I liked his explanation of Cronbach's Alpha and what it does. I liked, too, how he explained if there were deleted items, what we should do with them, and how we should determine if items should be deleted. This was really helpful.
Like many others, I did not know that Cronbach's alpha existed prior to this video and can definitely see it's usefulness as a statistic being that Likert scales are quite common. Thank you for posting this.
Once again I found the presentation to be very useful, providing a step-by-step demonstration on how to successfully calculate and interpret the Cronbach’s Alpha. As someone who has little knowledge and understanding of such concepts, I appreciated the simple description, noting that the Cronbach's alpha us a common tool to determine the reliability when working with Likert Scales, and that it doesn’t work to determine the reliability for only one item there must be a series of items. I will definitely be referring back to this video in the future. Thank You.
Yet again, I agree! I think it was valuable for Dr. Grande to demonstrate how to increase reliability based upon the data set and desirable Cronbach's alpha.
Thank you for sharing this video Dr. Grande. I appreciate the explanation of what the cronbach's alpha is used for. It is also interesting to see you can remove an outlier to see how the data can shift.
Dr. Grande- This video was easy to understand but also confusing for me. This concept allowed me to understand more about Cronbach's Alpha. I am not familiar with this method but you were able to explain this method in detail, which was very helpful. Thank you!
Brilliant video! I am currently doing my final year at uni, and i am using pre-existing psychometric scales and I will check the validity of these scales against the literature.
Really liked the explanation, I think this was a good video. I remember using SPSS in undergrad and this was interesting to see again, I understand it a lot more.
I was very nervous about this video because I have never even heard of Cronbach's Alpha, but it was very interesting to learn about. I do not think that I 100% understood everything in this video, but with a bit of Googling I feel more confident towards this subject.
It is helpful that the test shows the level for Cronbach's Alpha to meet. I am a bit confused as to how you can just take out certain data sets. I did not do much work with Cronbach's alpha in the past so I am somewhat lost.
I found this video to be very interesting. I liked the explanation of Cronbach's Alpha and what it does. I thought that it was really helpful. I also liked how Dr. Grande explained the Cronbach's Alpha if items deleated and how we can use that to determine if items potentially should be deleted.
I have never heard of Cronbach’s Alpha before but this video provided a very good explanation. I feel confident about this concept after watching the video. I also think this video was helpful since so many people use Likert scales.
Thank you so much for your thorough and clear explanation. It is so helpful especially for me being new in research. I just wrote up analysis for my paper after watching your video. Do you have any RU-vid video or web page that is about writing up analysis?
Thank you so much Dr. Todd. I am fairly new to statistic analysis and I was wondering if you have or recommend any videos from the beginning of receiving survey results, how to enter them in spreadsheet, how to upload them into SPSS and analyze the data. I want to have a solid background by having a strong foundation from the beginning. I would really appreciate any help. Thanks 🙏
It was helpful to learn that a Conbach's Alpha only works with scales, subscales, or series of items to determine the Conbach's Alpha. And even though the level of measurement were nominal, the scale was ordinal (1-10).
Thanks for the video. Quick question: Is it only likert scale-type (ordinal) measurements that can be used to run Cronbach's alpha? How can one test reliability of instruments with nominal (e.g. yes/no) variables?
Cronbach's Alpha is pretty interesting. I would have never imagined that there was such a simple way to determine which items to use and not to use on a scale. My question is what, in particular, is it about the items that were deleted make the test more reliable once the items are deleted? I was wondering if it was due to the responses or statistics or something else entirely.
Thank you for your video, Todd! It's always very illuminating to watch and learn from your video. I am a novice in running SPSS. Currently I have a survey with four scales. I have already determined the alpha coefficient for each of the scale and found them to be good (>0.7). Now I am asked to determine if the scales are overlapping or to run the discriminant validity of the scales, but I am not sure how to do this using SPSS. Can someone help advice pl? Thanks much!
I am completely unfamiliar with Cronbach's Alpha, but this video clearly explained it as a test of reliability of a scale, and is popularly used when working with Likert scales. I also learned that Cronbach's Alpha does not work in determining reliability for one item, but rather for a series of items. Overall, very informative and clearly explained material.
I'm glad you used the concept of likert scale. Am I correct to understand that a multi-item scale would be used as interval level variable used as either the dependent or independent variable? So I could use a parametric statistical test?
hi todd! thank you for your videos. I find them very useful. however, I have a question and hope you could make time to answer it. I am currently conducting a study for my master's degree and I tried testing for internal validity of my observation checklist through spss and got a 0.833 for the Chronbach's alpha. however, I wanted some changes made to the questions and it was about 50% of it. so I reconstructed the checklist, did another pretest, and viola the validity came crashing to a low 0.33. question is, can I just use the old tool and just omit some items and add one question (with 3 follow up?)?
Thank you Dr. Todd Grande. I always look up at your videos to learn about SPSS because it is easy to follow and understand. Forgive me for asking, if my questionnaire has 3 sections and each section has 10 questions, do I run the 10 questions only or the whole 30 questions to check its reliability? I would really appreciate your response.
excellent presentation. I have a question, testing internal consistency reliability within the items or questions of an assessment tool by one respondent, makes sense to me. What is the interpretation of the coefficient Alpha for items or questions among multiple respondents? (taking into account that different respondents may have different experiences about the same construct)
This was the first time I have heard of Cronbach's Alpha. It was very interesting to see that SPSS can show you what items could be removed to make it more reliable.
Danielle Duboski I found the removal of the items interesting, too, but confusing since I don't understand why someone would other than to improve the score.
Danielle Duboski This was the first time that I heard of Cronbach's Alpha, but I think this is definitely something worth looking into. I think that if I ever have the opportunity to use SPSS this would be something to try.
Hi Todd Grande, may i know how we compute alpha reliability for test/exam papers with different types of test techniques (true/false), (written test) (Yes/no) (comprehension A,B, C, D). Thank you.
Thanks for the video. I would like to ask about the Cronbach's alpha and the Factor Analysis. I have a data set of N105 and I have been measuring motivation at workplace. I have 8 variables for the IV and 4 for the DV (constructing one over all variable for motivation) I'm getting a good Alpha for the variables a 0,7 and above, but I found it difficult to interpret the factor analysis. Should I run it for my DV or the IV? Thanks
Thank you so much. It helped me a lot. I have one question that confused from this video. I understand that there are only 8 items left to distribute the samples and not 10 items. Is it right or wrong? Please help me. Thank you.
I like that the video begins with an explanation of what Cronbach's alpha actually is. I would like another example of when to use it right off the bat though, so I can wrap my head around how to use it before I learn how to use it. I like this video because I am using a Likert scale model in my research proposal, so I kind of understand this in terms of the Quality of Life index. I don't understand why you would delete data, though. What does "good" and "questionable" measures mean? Is that reliability? I think that's what I gathered from this.
This seems to be a little bit different, although I have heard of scaling before, I have never heard of Cronbach's Alpha. I liked that the data set was only 50, it was helpful in trying to imagine this in practice.
I remember using SPSS back in the day of Windows 3.1 and dot matrix printers... I'm not actually THAT old, my school was just pretty out of date in the 90s. 😂 Research design and methods was always my favorite class though. 💙
Straight to the point and easy to understand. Knowing that the item 7 is a good Alpha, whereas 8 or 9 Cronbach's Alpha would give us a lower reliability.
Candace Fernandez You think? I questioned that. I guess it is because I want to know "Why" something is, and not just how to get there. I think I need to abandon that when it comes to statistics because then I'l never accept anything into my brain.
Thank you so much. it was very helpful but please must all the respondents be included during the calculation of the reliability or some % of the sample? especially when the sample is large, say 300. please help. if some percentage of the sample, what percentage please?
This video is very informative. However, I did a reliability test cronbach's alpha on my pilot data but two variables came up with a negative value and at 0.00 respectively. All other variables in the instrument were reliable. I tried reverse coding the items to solve the problem but it seems not to work out. Please I need help. Thanks
Hi, do you have any resources on how to write up a results section for a lab report on a cronbach's alpha and justify a deletion of an item? Specifically for a dichotomous scale. Thank you.
If Item deleted in reliability analysis, does it effect other analysis? do i have to re-run compute mean, descriptive analysis, correlation pearson analysis, multiple regression analysis?
If we have 8 variables in our research. Shall we measure the coefficient alpha separately for each variable or we must conduct it on all items of the questionnaire?
Good to know that Cronbach's Alpha is a popular method to determine reliability when working with Likert scales. I also learned how to interpret the data to determine which items can/should be deleted to increase the reliability of an instrument (trying to get the alpha above 0.7).
I would like to have an explanation of how the mathematical formula ties to the concept. Also, if I have 5 items, and 100 subjects, for example, does the formula involve the number of items or the number of subjects, or both? Is there a formula that looks at relationship of covariance to variance? Is covariance the same as inter-item correlation? What does it mean to have item-total score correlation mean...in terms of variance and co-variance. As you can see, I want to understand chronbach's alpha in terms of covariance and variance, and link this to concepts such as item-total score correlation. But I need to understand the fine distinctions. All help is most appreciated... thanks..jes
Hi Dr. Todd Grande, Thanks for the video. I want to check with you that i have 12 dimensions to test for the reliability analaysis but there are over 7 dimensions with different items are very low. if the Cronbach's Alpha is very low which is below 0.4 then should i just get rid of the items and the dimensions?
hi, thanks for the video. but would like to know what do you do if you have few negative items and the scores are reversed from 5 to 1 on a five point rating scale. do you just continue in this same process
I`m not so sure, but I think you have to transform the reversed items. What I mean is that the scores of a scale must measure in the same direction. Let`s say you measure happiness for example and most of your items are formulated in similar way like: "I feel fulfilled", "I sleep well", "I eat normally", "I laugh everyday", and you score them from 1 to 4. 1 will be the minimum and 4 the maximum. And let`s assume in the same scale you have 2 items formulated in an opposite fasion like "I feel depressed" and "I feel exhausted" also scored from 1 to 4. Then you have to transform them in the direction most of the other subscales are oriented so 4 becomes 1, 3 becomes 2 etc. When your subscales measure in the same direction then you can test internal reliability with alpha. I hope I`m not mistaken :D
One of my questions to measure the variable was a ranking of 4 options. (from 1 to 4) These show up in my dataset as 4 separate questions. The other questions are likert scale. I set it on scale (not ordinal). Can I still use Cornbach's alpha the same way?
Hi! how to check the manipulation question? my advisor told me to use Cronbach alpha, but how to know the question is manipulated or not from the result of Cronbach alpha?
Hi Dr., I have a really tough question bothering me a lot. So, regarding the second point at the very end of the video, what should we do if this whole questionnaire (say these 8 items of scale measurements) is to measure the self-efficacy level, and you ve got another sheet of questionnaire to measure employee engagement, and finally, you want to get another sheet of data (one column of the self-efficacy level), and another sheet of data (one column of their engagement level), in order to analyse if they are correlated or not. Then HOW TO GAIN THAT DATA FROM EACH QUESTIONNAIRE? Or it is not the way to do the correlation? Bit urgent, could you please try to reply to me as soon as you could? Many many thanks, :)
This is a vey useful vedio Dr. Grande. I have a question regarding assessing the psychometric property of a scale. I used a published scale for job satisfaction including 4 options ( satisfied, very satisfied, dis satisfied and very dissatisfied ) after piloting my survey, participants suggested to add option nuteral , how I can assess the psychometric propority of the new scale? is adding that option change it?
So you just get rid of the items (10 and 7) to gain a .7 alpha? Are they no longer important to the data? Very clear demonstrations but confused about why to delete participant items because of the effect on the reliability on the scale. Thank you for the video Dr. Grande.