In: Statistics and Probability
A gambler thinks a die may be loaded, that is, that the six numbers are not equally likely. To test his suspicion, he rolled a die 150 times and obtained the data (1 - 23, 2- 26, 3- 23, 4- 21, 5- 31, 6- 26). Do the data provide sufficient evidence to conclude that the die is loaded?
here null hypothesis:Ho: p1=p2=p3=p4=p5=p6=1/6
Alternate hypothesis:Ha: at least pi is not equal to 1/6
degree of freedom =categories-1= | 5 |
for 5 df and 0.05 level of signifcance critical region χ2= | 11.070 |
applying chi square goodness of fit test: |
relative | observed | Expected | residual | Chi square | |
category | frequency | Oi | Ei=total*p | R2i=(Oi-Ei)/√Ei | R2i=(Oi-Ei)2/Ei |
1 | 1/6 | 23 | 25.000 | -0.40 | 0.160 |
2 | 1/6 | 26 | 25.000 | 0.20 | 0.040 |
3 | 1/6 | 23 | 25.000 | -0.40 | 0.160 |
4 | 1/6 | 21 | 25.000 | -0.80 | 0.640 |
5 | 1/6 | 31 | 25.000 | 1.20 | 1.440 |
6 | 1/6 | 26 | 25.000 | 0.20 | 0.040 |
total | 1.000 | 150 | 150 | 2.480 |
as test statistic is less than critical value we can not reject null hypothesis
we do not have evidence to conclude that die is loaded,