# paired data and use one-sample t procedures

• Feb 24th 2013, 02:26 PM
asilvester635
paired data and use one-sample t procedures
For (a) the random condition will be satisfied if the seeds were obtained from a random sample?
and for (b) when I do the hypothesis would my null hypothesis be 0? And the alternative hypothesis will be less than or equal to zero since we would think that the kiln would yield an increase. The difference would be negative when you minus regular to kiln... Please help
• Feb 24th 2013, 06:03 PM
chiro
Re: paired data and use one-sample t procedures
Hey asilvester635.

Hint: You are correct for a) but can you say how you can get a random sample physically and process wise for the experiment? (In other words, what kind of things can create bias and non-randomness between treatments?)
• Feb 24th 2013, 06:33 PM
asilvester635
Re: paired data and use one-sample t procedures
oh ok.. but for part (b) am i right or should i rethink my hypothesis?
• Feb 24th 2013, 06:44 PM
chiro
Re: paired data and use one-sample t procedures
If you want to test an increase with the kiln then basically you are testing mu_kiln > mu_other for the alternative and mu_kiln < mu_other (or mu_kiln = mu_other) for the null hypothesis.

Recall that typically the null hypothesis is true if you get a "null" result with regards to what you are expecting to find if your original premise is true. In other words, the null is the antithetical situation to the original hypothesis.
• Feb 24th 2013, 07:27 PM
asilvester635
Re: paired data and use one-sample t procedures
so basically the null hypothesis will be mu_kiln = 0 meaning that theres no difference in using a regular or kiln and the alternative hypothesis is that we believe that mu_kiln will be bigger than the mu_other because using kiln increase the yield of barley therefore the alternative hypothesis will be mu_kiln > 0? Am i right?
• Feb 24th 2013, 07:31 PM
chiro
Re: paired data and use one-sample t procedures
No you are looking at the difference of population means not just the means in isolation.
• Feb 24th 2013, 07:34 PM
asilvester635
Re: paired data and use one-sample t procedures
will we use one sample t test for mu?
• Feb 24th 2013, 07:38 PM
chiro
Re: paired data and use one-sample t procedures
No: if you are using a paired t-test then you need to do a paired hypothesis test that tests H0: mu_kiln - mu_other <= 0 and H1: mu_kiln - mu_other > 0.

Remember that you are trying to see if the kiln process increases the yield so in this context the kiln population mean should be greater than the other one.