Search on google for scientific papers on stuttering, I'm sure there are statistical measures for it. One simple method I could think of is just calculating the percentage of stuttered letters/(words) .. for example if I say a sentence like "I-I'm thirsty" then stuttering percentage is 50% (one word of two). Calculating this in a scientific setup is easy since researchers can just record the subject and calculate this. Calculating it continuously for normal people during their day would require something like an AI that can detects when you stutter and calculates the metrics automatically. You can try speech to text softwares and intentionally stutter to see If they pick the stuttering parts right, afterwards you can write simple code to calculate the metric based on the text output.