MHB Determine the confidence interval

AI Thread Summary
The discussion focuses on calculating confidence intervals for sodium and PCB content in water samples from Lake Macatawa and Lake Michigan. For the sodium content in the western basin, the sample mean was calculated as approximately 19.07, with an adjusted sample variance of about 10.60. A two-sided confidence interval for the mean sodium content at a 95% confidence level was determined to be approximately [17.90, 20.24]. Additionally, the participants discussed the necessary formulas and values for calculating confidence intervals, confirming the use of the sample standard deviation instead of the population standard deviation. The calculations and methodologies shared were generally validated by participants.
mathmari
Gold Member
MHB
Messages
4,984
Reaction score
7
Hey! 😊

(a) In winter, the roads around Lake Macatawa are salted. To study the impact of this on Lake Macatawa, students took $32$ water samples from the western basin of the lake and Sodium content (in parts per million, ppm) determined. As a result, the students have receive the following data:
1643538010837.png


(i) Calculate the sample mean.
(ii) Calculate the adjusted sample variance.
(iii) Assume a normal distribution and determine a two-sided confidence interval for the above sample for the mean sodium content with a confidence level of $0.95$. Also enter the used quantile and its (approximate) value.(b) In addition, $35$ water samples were collected from the eastern basin of Lake Macatawa and in each case the sodium content (in ppm) was measured. This resulted in a sample mean of $24.11$ and an adjusted sample variance of $24.44$. Assume a normal distribution.
(i) Determine a two-sided confidence interval for the mean sodium content with a confidence level of $0.9$. Also enter the quantile used and its (approximate) value.
(ii) Determine a confidence interval of the form $(-\infty, h]$ for the mean sodium content with a confidence level of $0.9$. Also state the one used quantile and its (approximate) value.(c) The PCB content was determined from a sample of fish from Lake Michigan (in ppm). It is known that the standard deviation is $0.8$ ppm. Assume a normal distribution. The below were measured:
1643539615613.png


(i) Calculate the sample mean and determine a two-sided confidence interval for the above sample for the mean PCB content with a confidence level of $0.99$. Also enter the used quantile and its
(approximate) value.
(ii) Determine a confidence interval of the form $(-\infty, h]$ for the above sample for the mean PCB content with a confidence level of $0.99$. Give also the quantile used and its (approximate) value.
I have done the following :

(a) (i) We add all elements and divide the result by the number of elements. So the sample mean is equal to \begin{align*}\overline{x}_{32}=& \frac{1}{32}\left(13.0 +18.5 +16.4 +14.8 +19.4+ 17.3 +23.2 +24.9 +20.8 +19.3 +18.8 +23.1 +15.2+ 19.9 +19.1+ 18.1 +25.1+ 16.8+ 20.4 +17.4 +25.2+ 23.1 +15.3+ 19.4 +16.0 +21.7 +15.2+ 21.3+ 21.5 +16.8+ 15.6 +17.6 \right )\\ & =\frac{1}{32}\cdot 610.2=19.06875\end{align*} Is that correct ? :unsure:

(ii) We have that \begin{align*}s_{32}^2&=\frac{1}{32-1}\sum_{i=1}^{32}(x_i-\overline{x}_{32})^2=\frac{1}{31}((13.0-19.06875)^2 +(18.5-19.06875)^2 +(16.4-19.06875)^2 \\ &+(14.8-19.06875)^2 +(19.4-19.06875)^2+ (17.3-19.06875)^2 +(23.2-19.06875)^2 \\ &+(24.9-19.06875)^2 +(20.8-19.06875)^2 +(19.3-19.06875)^2 +(18.8-19.06875)^2 \\ &+(23.1-19.06875)^2 +(15.2-19.06875)^2+ (19.9-19.06875)^2 +(19.1-19.06875)^2\\ &+ (18.1-19.06875)^2 +(25.1-19.06875)^2+ (16.8-19.06875)^2+ (20.4-19.06875)^2 \\ &+(17.4-19.06875)^2 +(25.2-19.06875)^2+ (23.1-19.06875)^2 +(15.3-19.06875)^2\\ &+ (19.4-19.06875)^2 +(16.0-19.06875)^2 +(21.7-19.06875)^2 +(15.2-19.06875)^2\\ &+ (21.3-19.06875)^2+ (21.5-19.06875)^2 +(16.8-19.06875)^2+ (15.6-19.06875)^2 \\ &+(17.6-19.06875)^2 )\\ & = \frac{1}{31}\cdot 328.54875\\ & \approx 10.59835\end{align*} Is that correct ? :unsure:

(iii) Do we use the formula \begin{equation*}\left [M(x)-\frac{\sigma}{\sqrt{n}}q_{1-\frac{a}{2}},M(x)+\frac{\sigma}{\sqrt{n}}q_{1-\frac{a}{2}}\right ] \end{equation*} where $M(x)$is the mean, i.e. $M(x)=19.06875$ ? Is $\sigma$ equal to the square root of the result of (b) ? The value of $q$ is related to the confidence level, right? :unsure:
 
Last edited by a moderator:
Physics news on Phys.org
Hey mathmari!

a.i and a.ii look correct to me. (Nod)
In a.iii we do not have $\sigma$, which is unknown. Instead we have $s$, which is the standard deviation of the sample. It is indeed the square root of the result of a.ii.

The value of $q$ is indeed related to the confidence interval. Can we find it? 🤔
 
Klaas van Aarsen said:
In a.iii we do not have $\sigma$, which is unknown. Instead we have $s$, which is the standard deviation of the sample. It is indeed the square root of the result of a.ii.

The value of $q$ is indeed related to the confidence interval. Can we find it? 🤔

Ahh ok!

So we use the formula $$\left [M(x)-\frac{\sqrt{V^{\star}}}{\sqrt{n}}t_{n-1,1-a/2}, \ M(x)+\frac{\sqrt{V^{\star}}}{\sqrt{n}}t_{n-1,1-a/2}\right ]$$ right?
Do we have $V^{\star}=s_{32}^2=10.59835$, $M(x)= 19.06875 $ and $t_{n-1,1-a/2}=t_{31,0.975}= 2,037$ ? :unsure:

So we get \begin{equation*}\left [19.06875-\frac{\sqrt{10.59835}}{\sqrt{32}}\cdot 2.037, \ 19.06875+\frac{\sqrt{10.59835}}{\sqrt{32}}\cdot 2.037\right ]\approx \left [17.89646, \ 20.24104\right ]\end{equation*} right? :unsure:
 
Last edited by a moderator:
mathmari said:
Ahh ok!

So we use the formula $$\left [M(x)-\frac{\sqrt{V^{\star}}}{\sqrt{n}}t_{n-1,1-a/2}, \ M(x)+\frac{\sqrt{V^{\star}}}{\sqrt{n}}t_{n-1,1-a/2}\right ]$$ right?
Do we have $V^{\star}=s_{32}^2=10.59835$, $M(x)= 19.06875 $ and $t_{n-1,1-a/2}=t_{31,0.975}= 2,037$ ?

So we get \begin{equation*}\left [19.06875-\frac{\sqrt{10.59835}}{\sqrt{32}}\cdot 2.037, \ 19.06875+\frac{\sqrt{10.59835}}{\sqrt{32}}\cdot 2.037\right ]\approx \left [17.89646, \ 20.24104\right ]\end{equation*} right?
Looks right to me. (Nod)
 
Back
Top