20211202 Class Activity 5

Name: 潘氏春瓊 (Quinn)

Student ID: 11096020

Appendix 5.8: Econometric LAB #3

Step 1: Specify the Model

Equation: ^
LIFEEXPECT = β0 + β1MEDINC1i + β2UNINSURED2i + β3SMOKE3i +
β4OBESITY4i + β5TEENBIRTH5i + β6GUNLAW6i + β7METRO7i + ɛi

Step 2: Hypothesize the Signs of the Coefficients

Equation: ^
LIFEEXPECT = β0 + β1MEDINC1i - β2UNINSURED2i - β3SMOKE3i -
β4OBESITY4i - β5TEENBIRTH5i + β6GUNLAW6i ± β7METRO7i + ɛi

I expect the relationship between LIFEEXPECT and those above coefficients are:

- MEDINC: positive sign, because the more money people earn, the more
investment in health care services which leads to the better health people can be.
It means if the income increase, people are able to spend more money on
maintaining healthy lifestyle (eating healthy foods, checking health regularly,
meeting better doctors), their standard of living is also in turn gone up, lengthened
life;
- UNINSURED: negative sign, because the more percentage of non-insurance, the
more risk people get when facing any health problems;
- SMOKE: negative sign, because smoking causes health problems such as lung
diseases, heart disease as well as cancer and vice versa;
- OBESITY: negative sign, since some research shown that overweight is capable
to increase the death rate of people, which reduces life expectancy;
- TEENBIRTH: negative sign, because giving birth when being a teenager may
increase the level of mental health problems. The higher rates of depression, the
less longevity;
- GUNLAW: positive sign. As the government gives more strict laws to control
gun violence growth which results in decreasing negative psychological effects
(suicide with gun or shooting people). So I expect that gun laws increase life
expectancy;
- METRO: both signs. On the one hand, according to the impact of air population,
the crime rates, the higher percentage of the population lived nearby metro stops,
the higher negatively affection on life expectancy. On the other hand, it is
beneficial for people living nearby metro stations, especially the elder, to move
wherever they want by metro without traffic congestion (that can cause skin
problems) as well as traffic accident (that may harm the longevity rate).

Step 3: Summary Statistics

Step 4: Estimation

Step 5 and 6: Hypothesis Testing (t-statistics and p-values)

 tc = t0.05, 43 = 1.68107

MEDINC
Step 1: State null and alternative hypotheses
H0: β1 ≤ 0
HA: β1 > 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

β^1 −0.0108538
t1 = = = -0.4879143 ~ t(43)
^
se ¿ ¿¿ ¿ 0.0222453

Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

Rejection rule for a upper tail test: Reject H0 if tk ≥ tα, df
t1 = -0.4879143 < 1.68107
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α
p1 = 0.628089 > 0.05

Step 5: Make conclusion

 Do not reject H0, there is not enough evidence to infer HA: β1 > 0

UNINSURED
Step 1: State null and alternative hypotheses
H0: β2 ≥ 0
HA: β2 < 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

^
β2 0.0459375
t2 = ^ = = 1.24624262 ~ t(43)
se (β ¿¿ 2) ¿ 0.0368608
Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

Rejection rule for a lower tail test: Reject H0 if tk ≤ -tα, df
t2 = 1.24624262 > -1.68107
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α
p2 = 0.219422 > 0.05

Step 5: Make conclusion

 Do not reject H0, there is not enough evidence to infer HA: β2 < 0

SMOKE
Step 1: State null and alternative hypotheses
H0: β3 ≥ 0
HA: β3 < 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

β^3 −0.221999
t3 = = = -4.41759165 ~ t(43)
^
se ¿ ¿¿ ¿ 0.0502534

Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

Rejection rule for a lower tail test: Reject H0 if tk ≤ -tα, df
t3 = -4.41759165 > -1.68107
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α
p3 = 6.63996e-005 < 0.05

Step 5: Make conclusion

 Reject H0, there is enough evidence to infer HA: β3 < 0

OBESITY
Step 1: State null and alternative hypotheses
H0: β4 ≥ 0
HA: β4 < 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

 ^
β4 −0.126588
t4 = ^ = = -2.5161098 ~ t(43)
se ( β 4 ) 0.0503110

Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

Rejection rule for a lower tail test: Reject H0 if tk ≤ -tα, df
t4 = -2.5161098 < -1.68107
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α
p4 = 0.0156791 < 0.05

Step 5: Make conclusion

 Reject H0, there is enough evidence to infer HA: β4 < 0

TEENBIRTH
Step 1: State null and alternative hypotheses
H0: β5 ≥ 0
HA: β5 < 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

^
β5 −0.0781766
t5 = ^ = = -4.24119093 ~ t(43)
se (β ¿¿ 5)¿ 0.0184327

Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

Rejection rule for a lower tail test: Reject H0 if tk ≤ -tα, df
t5 = -4.24119093 < -1.68107
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α
p5 = 0.000115874 < 0.05

Step 5: Make conclusion

 Reject H0, there is enough evidence to infer HA: β5 < 0

GUNLAW
Step 1: State null and alternative hypotheses
H0: β6 = 0
HA: β6 ≠ 0
Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

Restriction:

b[gunlaw] = 0

Test statistics: F(1, 43) = 3.75135, with p-value = 0.0593533

( RSS M −RSS)/ M (21.1225−19.4276)/1

F(1, 43) = = = 3.75135
RSS/(n−k −1) 19.4276/43

Step 4: Decision rule

1. Critival value approach: choose the level of significance is: α = 0.05

F0.025, 1, 43 = 5.39456
Rejection rule for an upper tail test: Reject H0 if Fk ≥ Fα/2, dfn, dfd
F = 3.75135 < 5.39456
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α

p6 = 0.000115874 < 0.05

3. CI approach:
If the confidence interval contains the hypothesized value “0”, do not reject
H0, otherwise reject H0
-0.0199756 < β6 < 0.988998

Step 5: Make conclusion

 Do not reject H0, there is not enough evidence to infer HA: β6 ≠ 0

METRO
Step 1: State null and alternative hypotheses
H0: β7 = 0
HA: β7 ≠ 0

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

^
β7 −0.0155073
t7 = ^ = = -2.36249514 ~ t(43)
se ( β ) 0.00656395
7

Step 4: Decision rule

 1. Critival value approach: choose the level of significance is: α = 0.05
tc = t0.0252, 43 = ± 2.01669
Rejection rule for a lower tail test: Reject H0 if | tk | > -tα/2, df
| t7 | = 2.36249514 > 2.01669
2. p-value approach:
Rejection rule using p-value: Reject H0 if p-value ≤ α

p
= Pr{t7 > 2.01669|H0} = 0.01135
2

p7 = 0.0227 < 0.05

3. CI approach:
-0.0287448 < β7 < -0.00226987

If the confidence interval contains the hypothesized value “0”, do not reject H 0, otherwise
reject H0

Step 5: Make conclusion

 Reject H0, there is enough evidence to infer HA: β7 ≠ 0

Step 7: Overall F-test

Step 1: State null and alternative hypotheses

H0: β1 = β2 = β3 = β4 = β5 = β6 = β7 = 0

HA: at least one βj ≠ 0, j = 1, 2, 3, 4, 5, 6, 7

Step 2: Set up the level of significance: α = 5% = 0.05

Step 3: Collect data and calculate the test statistic

ESS /K 119.573/ 7 17.0818571

F= = = = 37.808 ~ F(7, 43)
RSS /(n−k −1) 19.4276/43 0.45180465
 Step 4:

1. Critical value approach:

The result of Critical value of the F-Statistic is: Fc = F0.05, 43 = 2.23153

Reject H0 if F > F0.05, 43 = 2.23153

F = 37.808 > F0.05, 43 = 2.23153

2. p-value approach:
p = Pr{F > 37.808|H0} = 2.43*10-16 ≤ α

Step 5: Make conclusion:

 Reject H0, there is enough evidence to infer that at least one βj ≠ 0, j = 1, 2,

3, 4, 5, 6, 7

Step 8: Drawing Conclusions

The result testing β6 shows that to not reject H0, there is not enough evidence to infer HA:
β6 ≠ 0. So the coefficient GUNLAW is not reliable and adding this dummy variable to the
equation is not necessary. It also means passing the variable GUNLAW doesn’t affect the
life expectancy significantly than reducing the rate of SMOKE 3 points.