Simple Explanation of Chi-Squared
HTML-код
- Опубликовано: 11 сен 2024
- An explanation of how to compute the chi-squared statistic for independent measures of nominal data.
For an explanation of significance testing in general, see cit.evc.edu/psy...
There is also a chi-squared calculator at cit.evc.edu/psy...
![Chi-Square Test [Simply explained]](http://i.ytimg.com/vi/rpKzq64GA9Y/mqdefault.jpg)
![[DOKKAN BATTLE] Worldwide Campaign Announcement Video Part 2!](http://i.ytimg.com/vi/JpT5Voak6WA/mqdefault.jpg)
The fact that you decided to watch this video shows that you must be very intelligent!
Thank you for making me know that am intelligent. It's my first time to learn this topic n I feel I know what it talks about n i can do chi-square on my own .
Well, to determine that we'd need to know the number of people that were presented with the ability to watch the video, and the number of those who did or did not, as well as whether each of those people passed some intelligence test.
Then we'd make predictions based on the numbers and percentages we got, subtract that from our actual observed numbers, square it, divide by the number of observed cases in each instance, and eventually add them all together.
I know this now because I have become intelligent from watching this video.
(Thanks for making this video!)
Thank you
Or im just too dumb that i had to search about this so i could learn 🤣
nope. I just don’t understand this lesson
My prof. uses 2+ hours. You use 5 mins. RUclips University needs to start handing out degrees.
+OnlineNamesAreStrange You come out of class all confused, but primed to learn with all that crap kicking around in your head. After your mind has had time to chew over the info, you come up with specific questions and look them up on RUclips and suddenly figure it out. It's not your Professor's fault that you have your epiphanies on RUclips; it's just how our brains work.
Same here. The profs suck. I paid so much tuition for those profs/instructors who have trouble making you understand the concepts. I doubt if those profs really understand the subject himself. University education is a big scam. It just like a trader who trades you a degree.
greg i beg to differ i have taken same class twice or more with profs to understand asked them Qs multiple times but still doesnt feel satisfied or think got the ans or cleared the topic. while there have been times where i just used books or youtube or googled a topic without attending any class on topic n had fair understanding of the topic n performed well in later classes n in exam. There is something about the method a teacher uses in class to make it look heavy for self-satisfaction i dont know they but its true for many a time if not all the time. also they never get this deep they assume a lot on students behalf n skip over underlying concepts too. this is understandable given limited amount of time they cant but then again the same things r done neatly in a youtube vdo in shorter time.
You don't blame your prof. because this man solved a problem. (just example) i guess your prof gave full explanation, starting from how the formula used was derived.
@@jameson32 I get where you're coming from, but I'll politely disagree with you. In many cases, individuals google the keywords from lecture notes as they have ABSOLUTELY no idea what their instructors were speaking about. Eventually, after watching several videos such as this they can make their own opinion as a result of the information from several explanations.
I tried asking my professor how to use this shit 3 times, he lost his patience with me and just recommended I get a tutor. I'm watching this and this is beginning to make a little sense.
I hate when they do that. It makes me feel dumb sometimes when that happens
Your prof does not know this stuff.
Thank you for your clear explanation on this topic. I didn't really understand anything in my 3 hrs lecture but I understood instantly after your 5 minutes video. You just saved me from failing my statistics class
I am a med student from Russia working on a research project and this video really put it across for me. Thank you so much!
My professor has been going over this for 2 days and I was completely lost however after your video It makes A LOT more sense now.
Thank you so much for this. My professor's powerpoints are useless.
Same here. He sucks but you can't complain about him. It just like a big scam. You paid so much tuition but it ends up making you confused and miserable. Do they care? No, they don't because they have already taken your tuition.
Absolutely awesome, thought I would NEVER understand Chi Square, thank you
This video has been beyond helpful!! Thank you so much for uploading such a simplified explanation; I finally understand it now!
i dont
Is there a part 2? the video cut off at the end... :(
Thanks for the short and simple examples. It's been many years since I had to think about chi-squared tests, and this was a fantastic refresher.
Very straightforward explanation of the chi-squared. I will recommend to my medical students.
Plz help me
Best explanation of chi-squared test I have ever seen. So simple. Thanks.
For those who compare RUclips tutorial videos to your real professors, it may be a good practice to collect your data and do a Chi Square test on that hypothesis. Thanks for the great video by the way.
You are the man. For years I have wondered what df means and who even cares, today you just showed my why it is important.
omg that helped me out. I can easily do any chi tests now. and also that I scored 38 out of 40 THANK YOU SO MUCH
same for me!
I didn't understand my lecture notes, but this really cleared it up!
Thanks a million!
Awesome video! You taught 4 days of lecture in 5 minutes!
i missed the lecture on chi square and you just made me understand in 5 minutes very clearly. thank you very much
Fantastic video. You broke it down into simple terms that made it easy to understand. Thank you, sir.
If you read my comment after 10 years how are you and what are you doing I mean Job ?
This explanation of degrees of freedom just changed my perception of degrees of freedom.
You should have showed how you looked up 2.0 in the Chi-square table. I couldn't find 0.16, then I was lost for the rest of it
You look up the Chi squared value either lower or equal to the one you calculated, not just the one closest to it and that is your p value. We work with a p value of 0.05 as reference, so any p value lower or equal to that is seen as significant and any p value higher/ greater than that is seen as insignificant.
in chi table for 1 df and 0.05 significant level i can find 3.841...
what is 0.16? im confused
@@shashikaaman3901 exactly my sentiments.
@@shashikaaman3901 If you solve for Chi sq. value you get 2.025, which is less than 3.841 so null hypothesis is not rejected.
0.16 is the percentile value for 2.025, so you either compare 2.025 with 3.841 or 0.16 to 0.05, both correspond to each other for 1 degree of significance.
My next AP Bio class will be much easier, because of this video. Thank you so much for making this understandable!
The clearest explanation that I came across so far! So my questions is: if Ho is "There is significant difference between man and woman behaviour", why when Ho is true, there is no difference?
+Cynthia33c H0 stands for the *null* hypothesis -- H0 true means "there is no (null) difference" H1 is the alternate hypothesis: that there is a difference.
Thank you, David. I see now. So p-value is the first line (starting with 0.99, 0.95... 0.01), or short for ‘percentage' (the line aligned with degrees of freedom)? Say if p-value is increased, it is to the right or the left of the table of critical values (say using this table passel.unl.edu/Image/Namuth-CovertDeana956176274/chi-sqaure%20distribution%20table.PNG)
+Cynthia33c "p" stands for "probability" -- not "percentage." It's the probability that your result happened purely by chance, given that H0 is true. A lower p-value means that the probability that your result is just a fluke is lower.
Thanks! Excellent examples! Better than khan academy, and your voice is cool! 📝
You explained it better than my professor. Thank you!
That statement about degrees of freedom is brilliant !
Thanks for this. At grade 7 and my teacher being complicated af, this helped so much thanks man
Thank you. Your video helped me a lot. After watching many other, yours is the one that finally helped me understand the Chi Square Test.
Books and Teachers, simplify things like this man! Please!!!
Hey just wanted to say thanks for making this video. I don't know if you'll read this since this is 6 years after you made it but it helped me understand the basics of Chi-Squared a lot better than how my professor went about it. So thanks again! :)
Thanks; I do check in from time to time to read the comments, and I'm glad you liked the video.
Apperently this is impossible to explain outside youtube ^^
thanks a bunch you might just saved my future!
This just might get me to pass my Stats final exam today.
Thank-you
Ilovekirby Did you pass?
Bro you still replying some comments. Your legend. 10 years from now. Nice video dude
I find it ironic that people who are good at statistics use explanation that make statistics seem difficult, thanks for not falling into that trap and explaining things in an accessible way. Feynman would be proud.
Thank you; that’s a great compliment.
the best video yet on how to compute this. thanks so much
why do lectures have to be 2 hrs when they could make it 5 mins like this guy
2 and a half hours before my bio exam. thank you so much!
This is a good explanation video, despite the sound (or the lack of...).
Using the formula (O-E)**2 ÷ E on each cell, the terms are (32-24)**2÷24 + (38-36)**2÷36 + (20-30)**2÷30 +(8-16)**2÷16 + (22-24)**2÷24 + (30-20)**2÷20. That becomes 64÷24 + 4÷36 + 100÷30 + 64÷16 + 4÷24 + 100÷20. That works out to 2.67 + 0.11 + 3.33 + 4 + 0.17 + 5, which adds up to 15.28
one of best teacher ever
@Jonisto Saw your comment before it was removed; the p-value comes from a table (usually in the back of your statistics book) or from a computer program that calculates chi squared. Yes, I should have made that more explicit.
Do a search on Wikipedia for "Yates's correction for continuity"; in some cases (for small amounts of data) it is useful. The article states that "Unfortunately, Yates' correction may tend to overcorrect."
You're a good teacher! But. May I know how you get the probability of the result of the first problem to be 0.16? Does anyone know?? please answer.....
Its interesting to think we could replace a teacher with this video and have kids learn better. Thanks for the help!!!
Not gonna lie, my teacher was kinda confusing. You're much better :D Thank you so much!!
Hi David. Thank you for this simple example that really has alleviated severe anxiety in attempting to understand Nonparamteric Statistics on a doctoral level (when I have never taken a stats course ever before). Please explain the P=0.05, P>0.05, P
Thanks. This really explained when we should use Chi-square test and how at the same time.
Dear Mr. Eisenberg,
Thanks for the video. It helped me understanding how chi-square tests are calculated. But, what I am still unsure of is if chi-square value itself retains any significance or not, like, p < 0.01 means it has a chance of happening in less than 1 in 100 cases.
FYI, I came across this while doing my thesis, with a poor insight of it. I would really appreciate you guide me to any clearer conception. TY.
p < 0.01 means that if you did the experiment many, many times, and the null hypothesis is indeed correct, you would expect results like the ones you got only one out of a hundred times. I wrote a "visual novel" about hypothesis testing; see the link in the video description.
Go to a chi-squared calculator (such as the one at evc-cit.info/chisquared). Enter the female data in the first column, and the male data in the second column. Personally, I think a Wilcoxon signed-ranks test is more appropriate, but since you have been asked to use chi-squared, do that.
thanks sir I miss the lecture of chi-squared on some case but I easily understand from this lecture....from Sindh pakistan
Great video man! I'm sending this video to a bunch of friends who, like me, know oh so little of the magic kingdom Statistics.
Yes, it can, but the formula works *only* for a 2x2 table. That's why I gave the long explanation of how and why chi-squared works. The formula I give works for any number of rows and columns.
thankyou very much i had an assignment at uni to do with statistics this helped me alot!!!!!!!!!!!!!!!
@lennongirl5 Thanks for the kind words. Glad to be of assistance.
Thanks for the video explanation, it's very concise. I'd like to chime in that against the naysayers, I actually enjoy that sort of rawness in your voice.
I'm really bad when it comes to math but this helped clear things up a little thank you
@Jonisto You can try a calculator like the one Texas A&M University's site; do a google search for "chi-squared calculator tamu"
Thank you so much! This video was much clearer than how my professor was explaining this. :)
I'm not sure. You'll have to ask someone with a better grasp of statistics than I have. My best guess: you can take interval data (like height in cm.) which would be appropriate for a t-test and then break it into categories (short, medium, tall) and treat it as nominal data. They may have taken a t-test and then treated it as categorical. Is the paper online where I could look at it in context?
Thank you J. David! That was tremendously helpful!
I thought that if the value of chi squared exceed the critical value, the null hypothesis should be rejected? Please enlighten me..Thank You!
A chi-squared is *not* a t-statistic. You use a t-test with interval data, to determine if the means of two groups are identical or not. Chi-squared is used with nominal (categorical) data, and is used to see if the distribution of frequencies among the categories is equal or not. Calculating the mean for nominal data doesn't make sense. Instead, you look at the expected values and see which frequencies are higher or lower than the expected values.
Thank you so much for this video! Super clear and easy to understand! :)
wow thank you so much! my prof explained the expected and observed stuff but didn't explain why we were using them...
You made that so simple! thank you a million times!
thanks for this upload ! audio could have been better but the concept was so simply explained ! :)
Thank you! Very simply put and well explained. Saved my test!
A very simple and understandable explanation. Thank you.
Great video, makes it easier to understand. Thanks!
I can't tell how you got 15.26 unless you send me the exact calculations you did. You can calculate the p value by looking it up in a table, or most statistics packages will calculate it for you. The larger the value of χ², the further the actual value is from the value that would be expected by chance, so there is less probability that your result was "pure luck."
I read a paper: "we obtained a t-statistic that is modeled as a Chi-square distribution
with 2k degrees of freedom where k represents the number of subjects". What does this then mean?
Thank you so much for explaining this so clearly! I understand this much better finally after your explanantion
What if you have 3 groups (Men, Women, and Children). How do you know which group is contributing to the significance?
Thank you; thank you so much! I still have a question - can the same test tell which group (between the males and the females) litters more than the other?
Thanks for the video David. I am a little confused about your explanation for why to retain Ho in the litter example. You got a chi square of 2.0 - I would think you then look up the critical value at df1, is 3.84 for p=.05, and 6.63 for p=.01. Now because 2 is less than each of these, we say that the chi square is not significant. i.e the whether you litter or not does not appear related to being male or female.
what is the answer of last question? is it like this? As P value is less than 5% we reject the null hypothesis.
Is the first test a chi squared test for homogeneity and the second one a test for independence? Please answer someone who knows or @J David Eisenberg
That's correct. Chi squared is not significant, which means there is no significant difference in littering behavior between men and women. That's why we retain the null hypothesis: because our findings ("no big difference between men and women") agree with what the null hypothesis says ("there's no difference").
I didn't "choose" one degree of freedom; that's what the method requires. For chi squared, degrees of freedom is (# of rows-1) times (# of columns-1), so (2-1) times (2-1) = 1. I have a video about degrees of freedom as well, by the way.
Why so many people including me find that some videos on RUclips are clearer and simpler to understand than university classes?
Wow very good tutorial! Simplified approach for a beginner like me. Thanks for sharing sir ;)
It's a two-tailed test. You'll usually use those. I have a tutorial about significance testing,which I reference in the description of the video (I just added it). However, the explanation doesn't go into "rejection region".
how do you find the value of 2 in the chi square chart? there isnt one
At the secodn example you divided the row total by the grand total, but at the first example you just multiplied the two row totals. Why is that? wasn't it possible to use the grand total for the first part as well?
The grand total in the first example is 100, so everything works out as a percentage; dividing by 100 only moves the decimal point.
keep in mind instead of row total * column total / full total, you could do the same thing differently but have a better grasp on the concept: row total over total * column total over total, youre finding the probability of being in that square !!!
Man thanks a gazillion. This is so easy!
Thank you !
Question though on the first example : How do you go from P=0,16 to 'the H0 is true' ... ?
Thank you very much. Do you have any recommendations for other non parametric tests or other videos of K-S one sample tests, one sample runs tests, McNemar tests, Sign Tests, Wilcoxin Tests, K sample tests, Correlational Tests, Spearman r tests, Kendall Rank and Partial Rank Correlation Coefficients, Walsh Tests, Fisher Tests. Chi Squared 2 sample Tests: uhh...basically an intro level couse for neophytes who need a "Rosetta Stone" to understand the language of nonparametric stats?
Thanks. I found it useful. And don't worry about the microphone.
Hi David. Just have a quick question as to why you chose 1 degree of freedom for the litter example? Thanks
I understood this perfectly well, yet when my lecturer explains this to me I have no idea what he is talking about; it's as if he is speaking a foreign language.
Nice explanation! This helped me quite a bit.
Excellent explanation of chi square. Thanks.
No. It's quite lengthy (and would be pretty boring in a video). It should be in the back of your statistics textbook. If not, do a Google search for "chi squared table" and there it is.
Thank you for the excellent lecture.
Hi Mr. Eisenberg: I am working on an assignment using the chi-square. How would I fill in the chart with data of the following: sales force of 500 people that are divided into 4 regions; during the last 3 months, half of the sales reps in each region was given a software to help manage contacts. How would I set up the table for that data?
I can't say, since I don't know what you are trying to measure. Are you trying to find out if the people with the software had more sales than those without? If so, that's a plain independent samples t-test, not a chi-squared. I need to know what your dependent variable is.
@jdeisenberg Thank you, I figured it out in the end :) One more question though, If the table is not big enough, for example, the one im using in the exam only starts at .10 so if the chi-squared is less than 2.706 what would you do?
By the way this is easily the best explanation of hypothesis testing on youtube (I've been looking around all day!) Big thumbs up from me!
Hi, thanks for this good explanation! However, there is something I still don't understand. In the last example (flyers of different colors), you conclude that the results are significant, meaning that you reject the null hypothesis, meaning that there is a difference between the colors. But can you conclude on what kind of difference there is, based on the data in the contingency table? Since 32 people took the white flyers vs. an expected 24, and that 30 people did not take the pink flyers vs. 20 expected only, can we ALSO conclude, statically speaking, that white flyers are more attractive than pink flyers ? Basically, that White is better than Light Blue which is better than pink. Or do you need further tests to assess that? If yes, what kind of tests? Thanks!
+Charles Vestur I'm not sure if you can make that conclusion statistically or not. I suppose you could do a one-dimensional chi-square test of the color totals for those who took flyers. When you do this, the p value comes out to .061, so it's just barely missing significance. For a better viewpoint on this, find a statistics professor at a local college/university and ask.
HI im quite confused on where the 0.16 came from when the value stated in the table is 3.841.
Your table says that 3.841 is required for 5%. But since our calculate chi square is less, that means our probability is more than 5%; my table shows 10% is 2.71, 20% is 1.64, so a value of 2 is somewhere between 10 and 20%; in this case, about 16%