You are on page 1of 5

One way and two way ANOVA

1. Consider an experiment with four groups, with eight values in each. For the ANOVAsummary table
below, fill in all the missing results. Also test whether averages of four groups are significantly
different?
Sources df SS MS F
Among groups (i) ? (ii)? 80 (iii)?
Within groups (iv)? 560 (v)?
Total (vi)? (vii)?
Ans: (i) 3 (ii) 240 (iii) 4 (iv) 28 (v) 20 (vi) 31 (vii) 800
Solution:
Ho: 1=2=3=4
H1: 1≠2≠3≠4
 = 0.05
F3,28,0.05 = 2.92
Since, Fstat = 4 is greater than 2.92, Ho is rejected. Hence, averages of four groups are significantly
different.

2. The prices of certain commodity in 4 different cities are given. There are 4 shops, 3 shops, 4 shops and
4 shops in city 1, city 2, city 3 and city 4 respectively. Complete the output table.Also test whether
average prices of four cities are significantly different?
Sources df SS MS F
Among cities (i) ? (ii)? (iii)? (iv)?
Within cities (v)? (vi)? 11.76515
Total (vii)? 307.7333
Ans: (i) 3 (ii) 178.3167 (iii) 59.43889 (iv) 5.052114 (v) 11 (vi) 129.4167 (vii) 14
3. A supermarket that has a chain of 5 stores is concerned about its service quality reputation perceived
by its customers during Monday to Friday. Complete the output table. Also test whether service
qualities of 5stores are significantly different?
Source SS df MS F
Rows (Days) (i)? (ii)? (iii)? 8.737051
Columns (Stores) (iv)? (v)? 115.44 (x)?
Errors (vi)? (vii)? 17.665
Total (viii)? (ix)?

Solution: (ii) r-1 = 5-1 = 4 (v) c – 1 = 5 -1 = 4 (vii) = (r – 1) (c – 1) = 4*4 = 16


(ix) rc – 1 = 5*5 – 1 = 24 (iii) 8.737051*17.665 = 154.34
(ii) = 154.34*4 = 617.36 (iv) = 115.44*4 (vi) = 17.665*16
(viii) 617.36 + 461.76 + 282.64 = 1361.76 (x) 115.44/17.665 = 6.5349556.
Ho: 1=2=3=4=5
H1: 1≠2≠3≠4≠5
 = 0.05
F4,16,0.05 = 3.01
Since, Fstat = 6.53 is greater than 3.01, Ho is rejected. Hence, service qualities of 5 stores are significantly
different.
Source SS df MS F
Rows (Days) 617.36 4 154.34 8.737051
Columns (Stores) 461.76 4 115.44 6.5349556
Errors 282.64 16 17.665
Total 1361.76 24
4. A marketing manager of company producing tires was interested in knowing the comparative picture
of average life of various 4 brands of tires. The experiment was done in 4 cities to take care of road
conditions. Fill all the remaining results.
Source SS df MS F
Rows (City) 325.5 (i)? (ii)? (iii)?
Columns(Brand) (iv)? (v)? (vi)? 9.678344
Errors 157 (vii)? (viii)?
Total (ix)? (x)?
Ans: (i) 3 (ii) 108.5 (iii) 6.219745(iv) 506.5 (v) 3(vi) 168.8333 (vii) 9 (viii) 17.4444 (ix) 989 (x) 15

5. An important consideration in deciding which database management system to employ is the mean
time required to learn how to use the system. A test was designed involving three systems and four
users. Complete the output table.

Source SS df MS F
Rows (Systems) (i)? (ii)? (iii)? 0.705882
Columns (Users) (iv)? (v)? (vi)? (vii)?
Error (viii)? (ix)? 22.66667
Total 210 (x)?

Ans: (i) 32 (ii) 2 (iii) 16 (iv) 42 (v) 3 (vi)14 (vii) 0.617647 (viii) 136 (ix) 6 (x) 11
6. Complete two way analysis of variance table of mean sales due to (i) four different
salespersons and (ii) three different districts, where following information are given:
Source SS df MS F
Between salespersons 12.66667 (ii) r-1=? (vi)? (ix)?
Between districts (i)? (iii) c-1=? (vii)? (x)?
Error 20.66667 (iv) (r-1)(c-1)=? (viii)?
Total 41.66667 (v)n-1=?
Also test whether the average sales is same for different district.
Ans: (i) 8.333333 (ii) 2 (iii) 3 (iv) 6 (v) 11 (vi) 4.2222 (vii) 4.16667 (viii) 3.444444
(ix) 1.225 (x) 1.2096
Regression (simple and multiple)
1. Partial output of regression analysis of sales revenue (in lakh) and number of advertisement on
TV is given below:

(i) Write down the regression model.


(ii) See whether the independent variable(no. of adv on TV) is significant or not? If yes, why?
(iii) What will be the sales revenue when advertisement on TV telecasted 20 times per week?
2. Partial output of regression analysis of sales revenue (in lakh) and number of advertisement on
TV is given below:

(i) Is the regression model best fitted? If yes, why?


(ii) Interpret the value of R2
(iii) Interpret the value of standard error.
Answer:
(i) In ANOVA table, p-value (significance F) is less than  = 0.05, regression model is best fitted.
(ii) 59.5 % of the variation in sales revenue is explained by the variability in no. of advertisement
on TV.
(iii) There is average variation 6.16 lakh in the estimation of sales revenue from actual values.
3. Partial output of regression analysis of satisfaction based on experience and income:

(i) Is the regression model best fitted? If yes, why?


(ii) Interpret the value of R2
(iii) Interpret the value of standard error.

4. Partial output of regression analysis of blood pressure and age is given below:

(i) Fit regression model.


(ii) Does independent variable age significantly explain the dependent variable BP? If yes, why?
(iii) How do you interpret the regression coefficient (1.199)?
(iv) Estimate BP of a respondent whose age is 50 years.
5. Partial output of regression analysis of house rent(per month in Rs) based on size of
apartment(no. of rooms) and the distance from down town:
(i) Is regression model best fitted? If yes, why?
(ii) Which independent variable is significant in the model? Why?
(iii) Which independent variable is insignificant in the model? Why?
(iv) Estimate the rent, if we want a 4-room apartment at distance 4 km far from the downtown.
(v) How do you interpret the regression coefficient of the independent variable No. of rooms?
(vi) How do you interpret the regression coefficient of the independent variable Distance(km)?
Solution:
(i) In ANOVA table, p-value(significance F) is less than  = 0.05, regression model is best fitted.
(ii) In coefficient table, as p-value(0.0045) of independent variable no. of rooms is less than
(0.05), it is significant in the model.
(iii) In coefficient table, as p-value(0.196) of independent variable distance is greater than
(0.05), it is insignificant in the model.
(iv)regression model Y = 1256.32 + 2629.31X1 – 279.31X2
The estimated rent for 4 roomed apartment at 4 km distance from the downtown is Rs 10656.32
(v) When no. of room is increased by 1, the rent will increase by Rs 2629.31
(vi) When distance is far away from downtown by 1 km, the rent will decrease by Rs 279.31
6. Partial output of regression analysis of satisfaction based on experience and income:

(i) Is regression model best fitted? If yes, why?


(ii) Which independent variable is more significant in the model? Why?
(iii) Which independent variable is less significant in the model? Why?
(iv) Estimate the satisfaction, if respondent’s experience is 20 years and has income Rs 600000.
(v) How do you interpret the regression coefficient of the independent variable experience?
(vi) How do you interpret the regression coefficient of the independent variable income?

Formula for partial and multiple correlation (when sums are given)

𝑛 ∑ 𝑋1 𝑋2 −∑ 𝑋1 ∑ 𝑋2
Use 𝑟12 =
√𝑛 ∑ 𝑋21 −(∑ 𝑋1 )2 √𝑛 ∑ 𝑋22 −(∑ 𝑋2 )2
𝑛 ∑ 𝑋1 𝑋3 − ∑ 𝑋1 ∑ 𝑋3
𝑟13 =
√𝑛 ∑ 𝑋12 − (∑ 𝑋1 )2 √𝑛 ∑ 𝑋32 − (∑ 𝑋3 )2
𝑛 ∑ 𝑋2 𝑋3 − ∑ 𝑋2 ∑ 𝑋3
𝑟23 =
√𝑛 ∑ 𝑋22 − (∑ 𝑋2 )2 √𝑛 ∑ 𝑋32 − (∑ 𝑋3 )2
when ∑ 𝑋1 , ∑ 𝑋2 , ∑ 𝑋3 , ∑ 𝑋12 , ∑ 𝑋22 , ∑ 𝑋32 , ∑ 𝑋1 𝑋2 , ∑ 𝑋1 𝑋3 𝑎𝑛𝑑 ∑ 𝑋2 𝑋3 𝑎𝑟𝑒 𝑔𝑖𝑣𝑒𝑛

𝑛 ∑ 𝑥1 𝑥2
𝑢𝑠𝑒 𝑟12 =
√𝑛 ∑ 𝑥12 √𝑛 ∑ 𝑥22
𝑛 ∑ 𝑥1 𝑥3
𝑟13 =
√𝑛 ∑ 𝑥12 √𝑛 ∑ 𝑥32
𝑛 ∑ 𝑥2 𝑥3
𝑟23 =
√𝑛 ∑ 𝑥22 √𝑛 ∑ 𝑥32
when ∑ 𝑥1 , ∑ 𝑥2 , ∑ 𝑥3 , ∑ 𝑥12 , ∑ 𝑥22 , ∑ 𝑥32 , ∑ 𝑥1 𝑥2 , ∑ 𝑥1 𝑥3 𝑎𝑛𝑑 ∑ 𝑥2 𝑥3 𝑎𝑟𝑒 𝑔𝑖𝑣𝑒𝑛

𝑤ℎ𝑒𝑟𝑒, 𝑥1 = 𝑋1 − 𝑋̅1 , 𝑥2 = 𝑋2 − 𝑋̅2 𝑎𝑛𝑑 𝑥3 = 𝑋3 − 𝑋̅3

You might also like