Which t test?

I have 5 machines that each produced 100 cars. I have data showing how long each toy car lasted. We think one machine is malfunctioning. I need to form a hypothesis and do a t-test. Do I do a single sample t test or independent or paired sample?

I think it should be a single sample?

My hypothesis is machine number 1 is malfunctioning. Should I add up the data for machines 2-5 and then compare the mean to that of machine 1? Or is there a better way to do this?


TS Contributor
I would suggest using ANOM (Analysis of Means). The null hypothesis is that each mean = the overall group mean. The alternate hypothesis is that at least one mean differs from the group mean.