Forum Discussion
#NUM! error with chi-squared test statistic
Dear All, I have been following the instructions on this YouTube video for executing the chi-squared test in Excel. It has been working well the first few times I've used it; however, now I ha...
Hi Joe,
Many thanks once again!
I realised one can run chi-squared tests in SPSS on uncoded 'word' data i.e. string data as SPSS calls it. So, I thought I'd quickly import my data to SPSS to see if a similar error message occurred. I'm pleased to report it did not, so I guess I'll continue on SPSS. Interestingly, the chi-squared statistic that SPSS returns is 1462.463 (vs 1453.99071 as per your equation above). I thought you mind be interested to know this. Obviously they are quite similar but nevertheless different...
Best wishes,
Jenny
Many thanks once again!
I realised one can run chi-squared tests in SPSS on uncoded 'word' data i.e. string data as SPSS calls it. So, I thought I'd quickly import my data to SPSS to see if a similar error message occurred. I'm pleased to report it did not, so I guess I'll continue on SPSS. Interestingly, the chi-squared statistic that SPSS returns is 1462.463 (vs 1453.99071 as per your equation above). I thought you mind be interested to know this. Obviously they are quite similar but nevertheless different...
Best wishes,
Jenny
JennyLMace wrote: "import my data to SPSS to see if a similar error message occurred. I'm pleased to report it did not"
In Excel, the "error message" per se arises when we calculate the chisq statistic based on the p-value alone. That is, CHISQ.INV.RT.
Is that the "error message" that you refer to?
What did you input to SPSS for that calculation: (1) the p-value alone; or (2) the actual and expected values, or the original data?
I am not familiar with SPSS. Even if you __think__ you did #1, is it possible that SPSS did #2 because it is aware of the related data (context of the "session")?
Just for grins, I would be interested in the result when SPSS calculates the chisq statistic without importing any data. IOW, just enter the p-value, ideally with 15 significant digits of precision.
Is that possible in SPSS?
-----
In any case, what p-value does SPSS return for your data?
Exactly zero (0.00E+00), as Excel does?
If not, please show us the SPSS p-value in Scientific notation, ideally with 5 or more decimal places (up to 15 significant digits).
Just curious. Thanks.
JennyLMace wrote: "Sorry Joe, I can't prioritise replying thoroughly right now"
I understand. And thanks; I think you answered my questions sufficiently. Since SPSS is aware of all of the data, it probably calculates the chisq statistic with SUM((actual-expected)^2/expected), which works fine in Excel as well, not with an SPSS equivalent of chisq.inv(pValue,df), which is "numerically-challenged" in Excel (wink). And you confirmed that with such a large chisq statistic, SPSS hits the same numerical limitations as Excel, and it returns a zero p-value.
Bottom line: In Excel, do not use CHISQ.INV.RT(pValue, df) to calculate the chisq statistic, if we have the data. It is unreliable. Use the SUM formula instead. Good to know. Thanks again.
- Sorry Joe, I can't prioritise replying thoroughly right now. #NUM! (or equivalent error in SPSS) was the error I was referring to. One doesn't have to input any formulae in SPSS. You just put in the data and then ask it to work its magic (i.e. select which test you want done). I can confirm it did state "0.00" in SPSS for the p-value. I read elsewhere that this should still be taken as p<0.001 as absolute zero is impossible - as you also said.
Many thanks again 🙂
