# Central limit theorem - num random variables vs. sample size?

Does the Central Limit Theorem require the number of random variables to increase to a sufficiently large number or the number of samples of each random variable to increase to a sufficiently large number?

If the number of random variables is 1 for the chi-square statistic (as an example), then the degrees of freedom is 1, and the distribution would not represent a normal distribution, no matter how many samples. However, if the number of random variables is large, it seems we would still require more than a few samples for the chi-square distribution to resemble a normal distribution.

Note: I am starting to self learn statistics, and may need some contextual info in answers.

Cross Validated Asked by imagineerThat on November 18, 2021

Your question shows some confusion! A sample is usually represented by a sequence of random variables iid $$X_1, X_2, dotsc, X_n, dotsc$$, where each random variable is a function from some sample space $$omega mapsto X(omega)$$, the argument $$omega$$ often omitted from the notation. So a realization is only one value, always. So $$n$$ in the CLT refers to the number of observations, each observation represented with one random variable.

So in your chisquare example, $$n$$ has nothing to do with the number of degrees of freedom $$nu$$. For the iid CLT, the only assumption is that expectation and variance exists as finite numbers. Since that is the case for the chisquare distribution, the CLT applies and when the number of observations $$n toinfty$$, the standardized mean converges to a normal distribution.

Answered by kjetil b halvorsen on November 18, 2021

## Related Questions

### Statistical line comparison

1  Asked on October 6, 2020 by mobeus-zoom

### In a 2 class problem how do I compute the normalization constant for finding the posterior distributions?

0  Asked on October 2, 2020 by anonymous

### Using percentage change with different sample sizes, which is more significant?

0  Asked on September 30, 2020 by katie-fenton

### Python Random Forest Prediction Probabilities Reliability, Overfitting?

0  Asked on September 24, 2020 by rkhan8

### Euclidean distance score and similarity

4  Asked on September 24, 2020 by navige

### How could I understand the Self-critical sequence training (SCST) model?

1  Asked on September 22, 2020

### Looking for feedback on my approach to split data into validation and test set?

0  Asked on September 22, 2020 by s_am

### How to set up first differences model?

1  Asked on September 21, 2020 by fabio

### Precision and recall for clustering?

3  Asked on September 20, 2020 by learner

### Mixed Effects Model

1  Asked on September 20, 2020 by seydou-goro

### Which Regression scoring to use when have small dataset?

0  Asked on September 19, 2020 by user293111

### Is my logistic regression model correct?

2  Asked on September 15, 2020 by mustapha-hakkou-asz

### Gibbs entropy and Shannon entropy

0  Asked on September 14, 2020 by alhayer

### Why small values produce undulating densities when ploting logarithm of a loguniform prior (in R)?

1  Asked on September 13, 2020 by prolix

### Can we maximize the log of the odds instead of log of the probability?

1  Asked on September 9, 2020

### Computing GLM Relativities from Spline Regression

1  Asked on September 8, 2020 by jordan

### Comparing a Bayesian model with a Classical model for linear regression

1  Asked on September 7, 2020 by student_r123

### Tuning parameters of SVM in tune function

1  Asked on September 5, 2020 by siegfried