Usually mechanical test were done at least 30 times for obtain good statistical results. A compare with literature is also meaningful. The variation coefficient must be lower than %20. If you are sure with your tests you can use them of course.

do many tests and check their statistics

We use Weibull statistics for brittle and quasi-brittle materials.

Don't overstress! If you have a homogeneous material, five tests - even three ones -  will be sufficient. If you have deviations of some % of course you should do some statistical analysis with some more samples.

Refer this link:https://www.friction-facts.com/equipment/full-tension-test-method

Hello Gerges.

You can calculate average square deviation. If it will be higher than 15% (engineering deviation) you need in the additional tests. If it will ba less than 15% it is ok.

Also You can use software for example Minitab. You can find free version in the Internet. You will can calculate average square deviation. Also using the software You will create different equation which will show influence input parameters on the output parameters and software automatically show deviation.

Best regards, Mikhail.  

Use alternative method to measure elongation and load. Then compare the results.

If your testing equipment is well calibrated, basically load cell and extensometer, then the variability you find is almost all due to the sample and test incertitude. So statistical analysis is ok for results.

In addition to above suggestions, you must make sure you have no rupture at the grips which indicates the test has failed and the data should be discarded. 

I think that it is important to check the relegation for each testing condition. In dynamic testing of material properties the details is crucial.

Firstly, you need to have an idea of the failure mode and see if it's consistent with the type of specimen tested? Secondly, look at the COV and check the distribution of your points

you can also check if your specimen / gripping system is well aligned (ASTM E1012 http://www.instron.com/en-gb/testing-solutions/by-material/composites/high-cycle-fatigue/nadcap-alignment-standards?region=United%20Kingdom) in order that you apply pure tension to the specimen, but if the material is homogenous, not too high variability (CV

the tensile test in the static mode.

solicitation speed should be relatively low (2 - 5 mm/min).

if the dispersion of the results is low you may consider to be reliable (SD

Tensile testing is measured according toASTM standered method.the results of 5 samples are similar or near each other .That is oky if the accurecy of preparation  rubber vulcanizate through mixing and vulcanization process  are good.The hemogenity of additives is an important factor (mixing).

Thank you all for your answers. It really helped me a lot.