In: Statistics and Probability
First National Bank employs three real estate appraisers whose job is to establish a property’s market value before the bank offers a mortgage to a prospective buyer. It is imperative that each appraiser values a property with no bias. Suppose First National Bank wishes to check the consistency of the recent values that its appraisers have established. The bank asked the three appraisers to value (in $1,000s) three different types of homes: a cape, a colonial, and a ranch. The results are shown in the accompanying table
House Type | 1 | 2 | 3 |
Cape | 424 | 413 | 429 |
Colonial | 528 | 549 | 539 |
Ranch | 387 | 403 | 378 |
a. Construct an ANOVA table. (Round intermediate calculations to at least 4 decimal places. Round "SS", "MS", "p-value" to 4 decimal places and "F" to 3 decimal places.)
b. If average values differ by house type, use Tukey’s HSD method at the 5% significance level to determine which averages differ. (If the exact value for nT − c is not found in the table, use the average of corresponding upper & lower studentized range values. Negative values should be indicated by a minus sign. Round your answers to 2 decimal places.)
a)
applying 2 way ANOVA without replication":
Source of Variation | SS | df | MS | F | P-value |
Rows | 36978.6667 | 2 | 18489.3333 | 133.337 | 0.0002 |
Columns | 120.6667 | 2 | 60.3333 | 0.435 | 0.6746 |
Error | 554.6667 | 4 | 138.6667 | ||
Total | 37654.0000 | 8 |
b)
MSE= | 138.667 | ||
df(error)= | 4 | ||
number of treatments = | 3 | ||
pooled standard deviation=Sp =√MSE= | 11.776 |
critical q with 0.05 level and k=3, N-k=4 df= | 5.04 |
Tukey's (HSD) =(q/√2)*(sp*√(1/ni+1/nj) = | 34.265 |
Lower bound | Upper bound | differ | |||
(xi-xj)-ME | (xi-xj)+ME | ||||
μ1-μ2 | -150.93 | -82.40 | significant difference | ||
μ1-μ3 | -1.60 | 66.93 | not significant difference | ||
μ2-μ3 | 115.07 | 183.60 | significant difference |