Question

You wish to examine if there is a difference between two dog training techniques. At your dog training academy you use a strictly voice control method that has been reliable but you read that a treat-reward method produce more lasting results and takes less time (i.e. you can make more money, more quickly). You rely on word of mouth recommendations to aquire new customers so you want the best training method possible.

When designing your experiment would you use an alpha of .05 or .01

When designing your experiment would you use a 1 or 2 taiied test?   

Answer 1

A lower alpha level is more stringent since the probability of making an error reduces. Since we reject the null hypothesis only when the p-value of the test <significance level of alpha

So a lower significance level means we want a lower p-value. Since we want to ensure that our results are correct we should choose a lower alpha level of 0.01 here ie a stricter test.

Here we would use a one-tailed test with the alternative hypothesis that whether using a treat reward method produces better results. (Ie of the type u1>u2)