Lecture 20

HypothesisHypothesisTestingTestingforforMeansMeans--SmallSmallSamplesSamples

Remark

For small samples, the t-distribution is used to test hypotheses about the population mean. The t-distribution is used because the population standard deviation is unknown and we have to estimate it using the sample standard deviation. Moreover we have to assume that the population is normally distributed since the Central Limit Theorem does not guarantee normality of the distribution of sample means for small samples when the population is not normal.

Example 1

Do US bathroom scales underestimate a person`s true weight? A $75 \mathrm{~kg}$ test weight was placed on end of $8$ randomdly selected bathroom scales. The readings were as follows:$$\begin{array}{8*{c}} 76.1 & 74.8 & 74.9 & 73.6 & 75.4 & 72.8 & 74.1 & 74.0 \end{array} $$
For the population of bathroom scales test $H_0: \mu=75$ versus $H_1: \mu<75$ at $\alpha=0.05$ level of significance. Report the $p$-value and draw a conclusion in the context of the problem.

Solution

$ \begin{aligned} \bar{x}&=74.475 \quad s=1.063 \end{aligned} $$$ t=\frac{74.475-75}{1.063/\sqrt{8}}=-1.40 \quad \quad (df=n-1=7)$$$0.10 \leq p-\text{value} \leq 0.125$.

Fail to reject $H_0$. This sample does not provide enough evidence to claim that the bathroom scales undersestimate weight.

Example 2

Dichluro-Diphenyl-Toichloroethane (DDT) is still used in some regions of the world as an agricultural insecticide. DDT causes cancer, reproductive problems, and liver damage.

A study of $6$ ponds in south-east Asia found the following readings for the concentration of DDT (in ppm)$$ \begin{array}{6*c} 9.5 & 9.6 & 9.3 & 9.5 & 9.7 & 9.2\end{array}$$At $\alpha=0.05$ level of significance test $H_0: \mu=9.0 $ versus $H_1: \mu>9.0 \mathrm{~ppm} $

Solution

$ \begin{aligned} \bar{x}&=9.467 \quad s=0.1862 \end{aligned}$$$t=\frac{9.476-9}{0.1862/\sqrt{6}}=6.14 \quad \quad (df=n-1=5)$$$0.0005 \leq P-value \leq 0.005$.

Reject (strongly) $H_0$. The concentration of DDT in this region exceeds $9 \mathrm{~ppm}$.

;