Hello Bobbie, I have three types of questions in my survey, Positive, negative and even. For positive questions I have assigned the likeret scale as 1=Strongly disagree and 5= strongly agree; for negative 1=Strongly agree and 5= Strongly disagree. So what should i code for the even question? Is it 1=Strongly agree, Strongly disagree; 3=Agree, Disagree; and 5= undecided? Can i use this for reverse coding for even questions in the set?
Nirzara Bandekar For the even questions, you could make undecided in the middle of your scale and make it a 3. Making undecided a 5 would make it like they strongly agree. What do you mean by even? The number of scale options?
@@bobbiedirr6116 Thanks for your quick response. I mean for the even type of questions, where both the answers agree and disagree are correct, how should i assign the likert scale and include them to calculate the cronbach's alpha calculation?
@@bobbiedirr6116 Example of an even statement in safety culture survey: "My organisation is ready to deal with something if it goes wrong". How should I assign the likert scale for this for 5 option- Strongly agree, Agree, Undecided, Disagree, Strongly disagree.
You are completely correct. Cronbach’s does not tell you anything about retest reliability. But it was made to kind of replace retest reliability when a researcher didn’t have people take two tests. I have a different video that explains it a bit more. The correction to the average interitem correlation makes the cronbach’s alpha somewhat comparable to retest reliability. But it doesn’t mean that your Cronbach’s alpha will always be same as the retest reliability. The reason for the correction and not just using the average interitem correlation comes from making it similar to retest reliability.
Hello Bobbie I am trying to validate my questionnaire I ollow what you did but I got weird results on some cells like this.. #DIV/0! let me know if exist a possibility to send my results. Also how I can do reverse coding because I have a lot on negative numbers...hope you can help me! Thanks WIlfred
Harmsy Rai I would need a little more information. But if the correlations that are negative are very small it isn't a big deal. If it is mostly correlations with one item, then you can consider taking that item out.
This is the standardized alpha, so the variance of a standardized variable (i.e., a variable that has been converted to z-scores) is 1. I didn't standardize the variables before hand, but correlations are the standardized form of covariance. Are you talking about the diagonal of 1's in the table?
![Gucci Mane & Sexyy Red - You Don't Love Me [Official Music Video]](http://i.ytimg.com/vi/j7IBqyFKT4c/mqdefault.jpg)
Very easy to understand ! Really helps, Thank you.
What does it mean if I have negative numbers and already double checked the coding- could it be if the number of responders are low?
Awesome ... very useful ... you explained in v simple way ... thanks
Thank you for posting your video! It's very valuable. Regards
Glad it helped!
what do you call of that formula?
Have you tried AVERAGEIF or AVERAGEIFS?
Eg. =AVERAGEIFS(B2:K11;B2:K11;"
No, but that would make it easier!
Bobbie Dirr yes :-)
But the function allows you to do it faster.
Zdzisław Kes thanks for the tip! :)
Hello Bobbie, I have three types of questions in my survey, Positive, negative and even. For positive questions I have assigned the likeret scale as 1=Strongly disagree and 5= strongly agree; for negative 1=Strongly agree and 5= Strongly disagree. So what should i code for the even question? Is it 1=Strongly agree, Strongly disagree; 3=Agree, Disagree; and 5= undecided? Can i use this for reverse coding for even questions in the set?
Nirzara Bandekar For the even questions, you could make undecided in the middle of your scale and make it a 3. Making undecided a 5 would make it like they strongly agree. What do you mean by even? The number of scale options?
@@bobbiedirr6116 Thanks for your quick response. I mean for the even type of questions, where both the answers agree and disagree are correct, how should i assign the likert scale and include them to calculate the cronbach's alpha calculation?
Nirzara Bandekar could you provide an example of an even question that you have?
@@bobbiedirr6116 Example of an even statement in safety culture survey: "My organisation is ready to deal with something if it goes wrong". How should I assign the likert scale for this for 5 option- Strongly agree, Agree, Undecided, Disagree, Strongly disagree.
I thought Cronbach's alpha was a measure of internal consistency reliability, not test-retest reliability.
You are completely correct. Cronbach’s does not tell you anything about retest reliability. But it was made to kind of replace retest reliability when a researcher didn’t have people take two tests. I have a different video that explains it a bit more. The correction to the average interitem correlation makes the cronbach’s alpha somewhat comparable to retest reliability. But it doesn’t mean that your Cronbach’s alpha will always be same as the retest reliability. The reason for the correction and not just using the average interitem correlation comes from making it similar to retest reliability.
Hello Bobbie I am trying to validate my questionnaire I ollow what you did but I got weird results on some cells like this.. #DIV/0!
let me know if exist a possibility to send my results. Also how I can do reverse coding because I have a lot on negative numbers...hope you can help me!
Thanks
WIlfred
Wilfred Matias sure email me at bobbiedrr@gmail.com
Perfect thanks I sent you an email!
Hello, I did the Reverse Coded correctly but after I correlate them it is still coming negative. Any help?
Harmsy Rai I would need a little more information. But if the correlations that are negative are very small it isn't a big deal. If it is mostly correlations with one item, then you can consider taking that item out.
Thank you so much. I hope pleasant for you.
thank you so much
why you have used average variance as 1?
This is the standardized alpha, so the variance of a standardized variable (i.e., a variable that has been converted to z-scores) is 1. I didn't standardize the variables before hand, but correlations are the standardized form of covariance. Are you talking about the diagonal of 1's in the table?