In: Statistics and Probability
Listed below are paired data consisting of amounts spent on advertising (in millions of dollars) and the profits (in millions of dollars). Determine if there is a significant negative linear correlation between advertising cost and profit . Use a significance level of 0.01 and round all values to 4 decimal places. Advertising Cost Profit 3 18 4 22 5 16 6 29 7 24 8 31 9 22 10 29 11 25 Ho: ρ = 0 Ha: ρ < 0 Find the Linear Correlation Coefficient r = Find the p-value p-value = The p-value is Less than (or equal to) α Greater than α The p-value leads to a decision to Do Not Reject Ho Reject Ho Accept Ho The conclusion is There is a significant negative linear correlation between advertising expense and profit. There is a significant linear correlation between advertising expense and profit. There is a significant positive linear correlation between advertising expense and profit. There is insufficient evidence to make a conclusion about the linear correlation between advertising expense and profit.
The given data is:
Advertising Cost | Profit |
3 | 18 |
4 | 22 |
5 | 16 |
6 | 29 |
7 | 24 |
8 | 31 |
9 | 22 |
10 | 29 |
11 | 25 |
We want to determine if there is a significant negative linear correlation between the advertising cost and profit.
Hence, the alternative hypothesis is that: r<0 and the null hypothesis is that r>=0
where r refers to the linear correlation coefficient. This coefficient ranges between -1 and 1. -1 signifies a strong negative correlation and +1 indicates a strong positive correlation.
In order to determine r, we first normalize both Advertising Cost (X) and Profit (Y). This is done by using the formula:
where X_bar and Y_bar refer to the mean of X and Y. Sigma(X) and Sigma(Y) refers to the standard deviation of X and Y respectively.
For the given data, we have:
Using these and the table above, we get the normalized X and Y to be:
Advertising Cost (X) | Profit (Y) | Z_X | Z_Y |
3 | 18 | -1.4606 | -1.1767 |
4 | 22 | -1.0955 | -0.3922 |
5 | 16 | -0.7303 | -1.5689 |
6 | 29 | -0.3652 | 0.9806 |
7 | 24 | 0 | 0 |
8 | 31 | 0.3652 | 1.3728 |
9 | 22 | 0.7303 | -0.3922 |
10 | 29 | 1.0955 | 0.9806 |
11 | 25 | 1.4606 | 0.1961 |
Next, we take the products of Z_X and Z_Y. We get:
Advertising Cost (X) | Profit (Y) | Z_X | Z_Y | (Z_X*Z_Y) |
3 | 18 | -1.4606 | -1.1767 | 1.7187 |
4 | 22 | -1.0955 | -0.3922 | 0.4297 |
5 | 16 | -0.7303 | -1.5689 | 1.1458 |
6 | 29 | -0.3652 | 0.9806 | -0.3581 |
7 | 24 | 0 | 0 | 0 |
8 | 31 | 0.3652 | 1.3728 | 0.5013 |
9 | 22 | 0.7303 | -0.3922 | -0.2864 |
10 | 29 | 1.0955 | 0.9806 | 1.0742 |
11 | 25 | 1.4606 | 0.1961 | 0.2864 |
Next, we solve for r. The formula for r is:
Summing the last column, we get:
The number of entries in the table are n= 9.
Hence, we have:
Now, there are n=9. Hence, df = 9-2 = 7.
Now, our alternate hypothesis is one-tailed. Hence, we have to multiply the significance level by 2 and our new significance level becomes 0.01*2 = 0.02.
Using a correlation coefficient, we get the critical value to be 0.7450.
To reject the null our r should be r<-0.7450 to have a negative correlation. And for the positive correlation, r>0.7450.
Clearly, that is not the case. Hence, we cannot reject the null hypothesis and conclude that the correlation is not significant.