Вы находитесь на странице: 1из 8

STATISTICAL METHODS FOR

DECISION MAKING
POST-RESIDENCY ASSIGNMENT
By

Arunkumar Suriyanarayanan, PGXPM-13

Arunkumar.pgxpm13@greatlakes.edu.in
MANAGERIAL REPORT
Wentworth Medical Centre Case
As part of a long-term study of individuals 65 years of age or older, sociologists and physicians at the
Wentworth Medical Center in upstate New York investigated the relationship between geographic
location and depression. A sample of 60 individuals, all in reasonably good health, was selected; 20
individuals were residents of Florida, 20 were residents of New York, and 20 were residents of North
Carolina. Each of the individuals sampled was given a standardized test to measure depression. The
data collected follow; higher test scores indicate higher levels of depression. These data are contained
in the file Medical1.

A second part of the study considered the relationship between geographic location and depression for
individuals 65 years of age or older who had a chronic health condition such as arthritis, hypertension,
and/or heart ailment. A sample of 60 individuals with such conditions was identified. Again, 20 were
residents of Florida, 20 were residents of New York, and 20 were residents of North Carolina. The
levels of depression recorded for this study follow. These data are contained in the file named Medical2

Case Solution Discussion


Question 1: Use descriptive statistics to summarize the data from the two studies. What are
your preliminary observations about the depression scores?
MEDICAL 1 MEDICAL 2
Florida New York North Carolina Florida New York North Carolina
Mean 5.55 Mean 8 Mean 7.05 Mean 14.5 Mean 15.25 Mean 13.95
Standard Error 0.478346934 Standard Error 0.492041932 Standard Error 0.634428877 Standard Error 0.708965146 Standard Error 0.923024206 Standard Error 0.65884668
Median 6 Median 8 Median 7.5 Median 14.5 Median 14.5 Median 14
Mode 7 Mode 8 Mode 8 Mode 17 Mode 14 Mode 12
Standard Deviation 2.139232523 Standard Deviation 2.200478417 Standard Deviation 2.837252192 Standard Deviation 3.170588522 Standard Deviation 4.127889737 Standard Deviation 2.946451925
Sample Variance 4.576315789 Sample Variance 4.842105263 Sample Variance 8.05 Sample Variance 10.05263158 Sample Variance 17.03947368 Sample Variance 8.681578947
Kurtosis -1.062192818 Kurtosis 0.626431669 Kurtosis -0.904925496 Kurtosis -0.340799481 Kurtosis -0.0301367 Kurtosis -0.592052134
Skewness -0.27355709 Skewness 0.625687389 Skewness -0.056188269 Skewness 0.280721497 Skewness 0.525352494 Skewness -0.041733773
Range 7 Range 9 Range 9 Range 12 Range 15 Range 11
Minimum 2 Minimum 4 Minimum 3 Minimum 9 Minimum 9 Minimum 8
Maximum 9 Maximum 13 Maximum 12 Maximum 21 Maximum 24 Maximum 19
Sum 111 Sum 160 Sum 141 Sum 290 Sum 305 Sum 279
Count 20 Count 20 Count 20 Count 20 Count 20 Count 20
CV 0.385447302 0.275059802 0.402447119 CV 0.218661277 0.270681294 0.211215192

Based on the descriptive statistics performed on data from Medical 1 and Medical 2, following are the
key observations.

 Regardless of central tendency used, the Florida depression scores are lower compared to New
York and North Carolina amongst reasonably healthy people above 65 years of age. However,
North Carolina has better depression scores (i.e. lower depression levels) amongst people above
65 years and suffer from Chronic illness.

 Combined observation based on descriptive statistics is that people older than 65 years of age
and suffering from chronic diseases such as arthritis, heart disease, and hypertension tend to
have higher depression levels.

 Based on Coefficient of Variance, it can be observed that variation spread of depression scores
in medical 1 (i.e. healthy people above 65 yrs old) is lower for New York whereas the variation
spread of depression scores in medical 2(i.e. among sample population of people above 65 yrs
with chronic illness) is lower for North Carolina, closely followed by Florida.

 To determine the influence of geographic location on depression scores across both sets of
observations, we move on to perform inferential statistics as described in next section.
NOTE: Solutions for Questions 2, 3 & 4 will be described for
Medical 1 and Medical 2 separately in the coming sections.

Inferential Statistics on Medical 1


Question 2: Use analysis of variance on both data sets. State the hypotheses being tested in
each case. What are your conclusions?

ANOVA for Medical 1


Hypothesis for Testing

Null Hypothesis – H0 The population means of depression scores of healthy people above 65
years across the three states do not differ significantly.
Alternate Hypothesis – The population means of depression scores of healthy people above 65
H1 years across the three states differ significantly.

Decision Rule: Reject Null hypothesis, H0, if calculated F value is higher than F Critical.

SUMMARY
Groups Count Sum Average Variance
Florida 20 111 5.55 4.576315789
New York 20 160 8 4.842105263
North Carolina 20 141 7.05 8.05

ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 61.03333333 2 30.51666667 5.240885809 0.008139895 3.158842719
Within Groups 331.9 57 5.822807018

Total 392.9333333 59

Conclusion: Since Computed F (2, 57) is 5.24 is greater than F Critical (3.16) at given level of
significance (0.05), null hypothesis can be rejected. Which means, “The population means of
depression scores of healthy people above 65 years across the three states differ
significantly”

Question 3 - Use inferences about individual treatment means where appropriate. What are
your conclusions?

Based on the results, it is evident that mean depression scores of individuals above 65 years and
healthy are affected by location. As noticed, people from New York have higher depression scores. I.e.
higher mean value.

This is also confirmed by the p-Value of 0.0081 from ANOVA, which means that the risk of rejecting
null hypothesis is just 0.81%. Hence H0 can be comfortably rejected. Which also implies that the
depression scores means differs significantly by location and location does influence the depression
levels on reasonably healthy people above 65 years.
Question 4 – T-Test for two samples in the order of Florida vs New York, Florida vs North
Carolina, New York vs North Carolina.

Equality of two population means are tested using this t-test. Since the population sample is < 30, we
can apply t-test. T Test Two sample assuming equal variances is used here. The level of significance
(α) is set at 0.05.

Florida vs New York

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean value of depression scores among healthy people above 65 years in Florida

µ2 – Mean value of depression scores among healthy people above 65 years in New York

Florida New York


Mean 5.55 8
Variance 4.576315789 4.842105263
Observations 20 20
Pooled Variance 4.709210526
Hypothesized Mean Difference 0
df 38
t Stat -3.570195772
P(T<=t) one-tail 0.000493602
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.000987204
t Critical two-tail 2.024394164

Conclusion: Since Absolute computed t value is above t critical for a two tail test, null hypothesis is
rejected. The mean values of depression scores among healthy people above 65 years in Florida and
New York differ significantly.

Florida vs North Carolina

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean value of depression scores among healthy people above 65 years in Florida

µ2 – Mean value of depression scores among healthy people above 65 years in North Carolina
Florida North Carolina
Mean 5.55 7.05
Variance 4.576315789 8.05
Observations 20 20
Pooled Variance 6.313157895
Hypothesized Mean Difference 0
df 38
t Stat -1.887851952
P(T<=t) one-tail 0.033349806
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.066699613
t Critical two-tail 2.024394164

Conclusion: Since Absolute computed t value is less than t critical for a two tail test, null hypothesis
cannot be rejected. The mean values of depression scores among healthy people above 65 years in
Florida is the same as that of in North Carolina.

New York vs North Carolina

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean value of depression scores among healthy people above 65 years in New York

µ2 – Mean value of depression scores among healthy people above 65 years in North Carolina

New York North Carolina


Mean 8 7.05
Variance 4.842105263 8.05
Observations 20 20
Pooled Variance 6.446052632
Hypothesized Mean Difference 0
df 38
t Stat 1.183250459
P(T<=t) one-tail 0.122029919
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.244059837
t Critical two-tail 2.024394164

Conclusion: Since Absolute computed t value is less than t critical for a two tail test, null hypothesis
cannot be rejected. The mean values of depression scores among healthy people above 65 years in
New York is the same as that of people in North Carolina .
Inferential Statistics on Medical 2
Question 2: Use analysis of variance on both data sets. State the hypotheses being tested in
each case. What are your conclusions?

ANOVA for Medical 2


Hypothesis for Testing

Null Hypothesis – H0 The population means of depression scores of people above 65 years with
chronic illness across the three states do not differ significantly.
Alternate Hypothesis – The population means of depression scores of people above 65 years with
H1 chronic illness across the three states differ significantly.

Decision Rule: Reject Null hypothesis, H0, if calculated F value is higher than F Critical.

SUMMARY
Groups Count Sum Average Variance
Florida 20 290 14.5 10.05263158
New York 20 305 15.25 17.03947368
North Carolina 20 279 13.95 8.681578947

ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 17.03333333 2 8.516666667 0.714212152 0.493906129 3.158842719
Within Groups 679.7 57 11.9245614

Total 696.7333333 59

Conclusion: Since Computed F (2, 57) = 0.7 is lower than F Critical (3.16) at given level of
significance (0.05), null hypothesis cannot be rejected. Which means, “The population means of
depression scores of people above 65 years with chronic illness across the three states do
not differ significantly.” The mean values are almost the same across three states.

Question 3 - Use inferences about individual treatment means where appropriate. What are
your conclusions?

Based on the results, it is evident that mean depression scores of individuals above 65 years of age
having chronic illness are not location dependent. If you notice from the value of means, the values are
very close to each other across the states in Medical 2

This can be further confirmed with p-Value from ANOVA and the assumed level of significance are
almost equal. Which in turn confirms that null hypothesis cannot be rejected and the mean depression
scores of people above 65 with chronic illness across three states are not affected by location.
Question 4 – T-Test for two samples in the order of Florida vs New York, Florida vs North
Carolina, New York vs North Carolina.

Equality of two population means are tested using this t-test. Since the population sample is < 30, we
can apply t-test. T Test Two sample assuming equal variances is used here. The level of significance(α)
is set at 0.05.

Florida vs New York

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean values of depression scores among people above 65 years with chronic illness in Florida

µ2 – Mean values of depression scores among people above 65 years with chronic illness in New York

Florida New York


Mean 14.5 15.25
Variance 10.05263158 17.03947368
Observations 20 20
Pooled Variance 13.54605263
Hypothesized Mean Difference 0
df 38
t Stat -0.644399039
P(T<=t) one-tail 0.261593908
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.523187817
t Critical two-tail 2.024394164

Conclusion: Since Absolute computed t value is lower than t critical for a two tail test, null hypothesis
cannot be rejected. The mean values of depression scores among people above 65 years with chronic
illness in Florida and New York are the same.

Florida vs North Carolina

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean value of depression scores among people above 65 years with chronic illness in Florida

µ2 – Mean value of depression scores among people above 65 years with chronic illness in North
Carolina

Florida North Carolina


Mean 14.5 13.95
Variance 10.05263158 8.681578947
Observations 20 20
Pooled Variance 9.367105263
Hypothesized Mean Difference 0
df 38
t Stat 0.568276882
P(T<=t) one-tail 0.286595016
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.573190033
t Critical two-tail 2.024394164
Conclusion: Since Absolute computed t value is less than t critical for a two tail test, null hypothesis
cannot be rejected. The mean values of depression scores among people above 65 years with chronic
illness in Florida is the same as that of in North Carolina.

New York vs North Carolina

H0: µ1 = µ2

H1: µ1 ≠ µ2

µ1 – Mean value of depression scores among people above 65 years with chronic illness in New York

µ2 – Mean value of depression scores among people above 65 years with chronic illness in North
Carolina

New York North Carolina


Mean 15.25 13.95
Variance 17.03947368 8.681578947
Observations 20 20
Pooled Variance 12.86052632
Hypothesized Mean Difference 0
df 38
t Stat 1.14634141
P(T<=t) one-tail 0.129410684
t Critical one-tail 1.68595446
P(T<=t) two-tail 0.258821369
t Critical two-tail 2.024394164

Conclusion: Since Absolute computed t value is less than t critical for a two tail test, null hypothesis
cannot be rejected. The mean values of depression scores among people above 65 years with chronic
illness in New York is the same as that of people in North Carolina.