Why calculated elapsed time of function 'ar' is different with function 'tic toc' and 'timeit'?

I want to compare computation complexity of Yule-Walker Method and other algorithm.

At firtst, I write the 2 codes below,

1) Calculate elapsed time using 'tic toc' function

tic;

ar(Test_Signal,N,'yw');

toc;

2) Calculate elapsed time using 'timeit' function

ff = @() ar(Test_Signal,N,'yw');

timeit(ff)

and results shows huge difference

1) 0.754736s

2) 0.0060s

Which result is correct? and how can I use this results with metric of computational complexity of algorithm?

[Alex Mcaulley]

Yes, It's a huge diference. How many outputs have your ar function?

timeit function is a better way to measure the elapsed time in the execution of a function, because the result is a median of multiple measurements.

https://es.mathworks.com/help/matlab/matlab_prog/measure-performance-of-your-program.html

[GeunHwan Kim]

Thank you Alex.

I just used Matlab's ar function.

So, Is it okay to use timeit function output as metric when I write some paper? I'm not sure that it reflect pure computational complexity of algorithm.

[Walter Roberson]

You should probably use timeit(ff, 0) but that should not matter much.

[GeunHwan Kim]

Thank you Walter!