t-test or Wilcoxon test in R I have a very small data sets of web traffic to compare the effect of performing advertising over five days or seven days. Yes, running a test over 7 days would definitely give me more traffic but I would like to know if the traffic is significantly higher and worth my consideration.
I run each test for 2 weeks. 
This is my dataset:
5day advertising web traffic    7day advertising web traffic
week1   week2                     week1    week2
184418  179650                    301978    308019

I ran the t-test in R and got this value
> a<-c(184418,179650)
> b<-c(301978,308019)
> t.test(a,b)

t = -31.9557, df = 1.898, p-value = 0.001307
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -140407.5 -105521.5
sample estimates:
mean of x mean of y 
 182034.0  304998.5 

Is my dataset too small and does this mean that there is a significance difference? 
When I run the Wilcoxon test (via wilcox.test), I get a different output:
data:  a and b
W = 0, p-value = 1
alternative hypothesis: true location shift is greater than 0

Thanks
 A: Given what you are trying to do, I am not sure a t-test is what you want.
I am guessing that advertising for 7 days costs more than advertising for 5. So, let's look at cost per day:
Week 1: 184,418/5 = 36,883 per day
        202,316/7 = 28,902 per day

Week 2: 179,650/5 = 35,930
        196,395/7 = 28,056

Then what impresses me is that the difference is nearly the same for the 2 weeks. The question of how to test this is, I think, not so important. 
In fact, if 5 days is M-F and 7 is every day, you might consider trying only the weekend!
Of course, cost may not be the same per day, in which case the above would have to be modified. 
A: Here is what I got when I used your data. I would go with the results of the wilcoxon test since your sample size is so small. 
x1<- c(184418, 179650)
x2<- c(202316, 196395)

t.test(x1,x2)

    Welch Two Sample t-test

data:  x1 and x2
t = -4.557, df = 1.913, p-value = 0.0488
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -34409.9541   -233.0459
sample estimates:
mean of x mean of y 
 182034.0  199355.5 

wilcox.test(x1,x2)

data:  x1 and x2
W = 0, p-value = 0.3333
alternative hypothesis: true location shift is not equal to 0

