We would say that there is "missing data." The n will be lower for this variable. Excel can still calculate many things (descriptive statistics, correlations, and t-tests, for example) when there is missing data.
The sample size that you need depends on the effect size that you're trying to detect, the desired statistical power, and the alpha level, all determined before you start collecting data. You can use a sample size calculator to determine how much data you should collect. Here's an example: sample-size.net/correlation-sample-size/ Your sample size will be the number rows in the spreadsheet minus the number of rows at the top which do not contain data.
Here's one approach. If you don't want to go through them one by one, here's one approach: Make a frequency distribution for each of the composite variables or calculate their z scores. Look for outliers that are far from the mean (e.g, with an absolute value of z greater than 3).
That is really very useful video. What about next level of working after this step. How to get final results after this clean up , i.e. in percentage etc. Anyone can suggest any link ?
+M Ejaz Everything depends on the hypothesis you're trying to test or the statistics that you want to know. See my CZcams homepage for many more stats videos.
There are many, many different ways to analyze the data, depending on what your hypothesis is. Here's a video that shows how to make a table of correlations from this data (or a data a set that is similar): czcams.com/video/cswLPBXhSWs/video.html
Thank you this was really helpful
The justin bieber joke is underrated 😂
best of the best
Thanks for explaining this! I wonder how we should deal with the age we deleted in this case when we do the data analysis in terms of age?
We would say that there is "missing data." The n will be lower for this variable. Excel can still calculate many things (descriptive statistics, correlations, and t-tests, for example) when there is missing data.
This is very helpful. Do you have other videos
I'm glad you found this useful. I have made hundreds of other videos: czcams.com/channels/kcEWq4OAzG3-ldTBaFQ8vA.html
Hi, Thank you for your useful video. You try to find outliners by checking each rows, but how can you check if you have over 1000 obseravations?
The sample size that you need depends on the effect size that you're trying to detect, the desired statistical power, and the alpha level, all determined before you start collecting data.
You can use a sample size calculator to determine how much data you should collect. Here's an example: sample-size.net/correlation-sample-size/
Your sample size will be the number rows in the spreadsheet minus the number of rows at the top which do not contain data.
@@DavidDunaetz I mean how you can find outliners when you have over 1000 observations
Here's one approach. If you don't want to go through them one by one, here's one approach: Make a frequency distribution for each of the composite variables or calculate their z scores. Look for outliers that are far from the mean (e.g, with an absolute value of z greater than 3).
Hi Professor Dunaetz, thank you so much! I have a question, which video should I watch to learn how to deal with the reverse score? thank you again!
This one should help: czcams.com/video/nIKFJfDEddc/video.html
@@DavidDunaetz Thank you so much!!!
That is really very useful video. What about next level of working after this step. How to get final results after this clean up , i.e. in percentage etc. Anyone can suggest any link ?
+M Ejaz Everything depends on the hypothesis you're trying to test or the statistics that you want to know. See my CZcams homepage for many more stats videos.
Hello sir I cants seem to find the video where you finally analyze this survey data in your channel
There are many, many different ways to analyze the data, depending on what your hypothesis is. Here's a video that shows how to make a table of correlations from this data (or a data a set that is similar): czcams.com/video/cswLPBXhSWs/video.html