strange behaviour with c# lambda expression using random

I want to filter a list of double values I want number of items in this list that are greater or equal than a random value

When I use random function inside the lambda expression the number of items founds doesn't correspond to the amount expected. When I put random function outside lamda expression, the code is working right here is not working version vs working version

not working code:

List<double> vecteur = new List<double> { 0.45, 0.5, 1 };
List<int> lstcompteurs = new List<int> { 0,0,0};
Random r = new Random();
for (int i = 0; i < 1000; i++) {                
     int index = vecteur.FindIndex(a => a > r.NextDouble());
     lstcompteurs[index]++;    
}
foreach (int cpt in lstcompteurs) {
   Console.WriteLine(cpt);
}
Console.Read();

output : 448 288 264

we notice that 288 represents much more than 5% of the sample.

working code:

List<double> vecteur = new List<double> { 0.45, 0.5, 1 };
List<int> lstcompteurs = new List<int> { 0,0,0};
Random r = new Random();
for (int i = 0; i < 1000; i++) {
   double b = r.NextDouble();
   int index = vecteur.FindIndex(a => a > b);
   lstcompteurs[index]++;    
}
foreach (int cpt in lstcompteurs) {
    Console.WriteLine(cpt);
}
Console.Read();

output: 443 48 509

As you see 48 represents indeed 5% of the sample.

I do not understand what is going on with the first version of my code