Statistical Analysis Of Sleep Problems, Alcohol Consumption Essay.

Part a

The independent sample t-test is used to test the differences in “how many alcohol drinks per day” for the persons “trouble falling asleep.” A t-test is used to compare the differences in the dependent variable (how many alcohol drinks per day) for the two independent groups (trouble falling asleep).

Part b

In order to test the differences we test the hypothesis that there are no significant differences in trouble falling asleep.

Null Hypothesis: There are no differences in trouble falling asleep between people taking alcoholic drinks.  

 

Alternate Hypothesis: There are differences in trouble falling asleep between people taking alcoholic drinks.   

 

Table 1: Independent Samples Test
			how  many alcoholic drinks per day
			Equal variances assumed	Equal variances not assumed
Levene's Test for Equality of Variances	F		.242
	Sig.		.623
t-test for Equality of Means	t		.088	.085
	df		253	181.632
	Sig. (2-tailed)		.930	.933
	Mean Difference		.01556	.01556
	Std. Error Difference		.17641	.18404
	95% Confidence Interval of the Difference	Lower	-.33186	-.34758
		Upper	.36299	.37871

 
From the t-test we find that t(253) = .088, p-value = .930 more than a = .05, level of significance. Thus we reject the Null Hypothesis. Hence there are statistically no significant differences in “alcoholic drinks per day” and persons having “trouble falling asleep.”

Part a 

Table 2: sex * problem with sleep? Crosstabulation
			problem with sleep?		Total
			yes	no
sex	female	Count	67	81	148
		% within problem with sleep?	57.3%	53.3%	55.0%
		% of Total	24.9%	30.1%	55.0%
	male	Count	50	71	121
		% within problem with sleep?	42.7%	46.7%	45.0%
		% of Total	18.6%	26.4%	45.0%
Total		Count	117	152	269
		% within problem with sleep?	100.0%	100.0%	100.0%
		% of Total	43.5%	56.5%	100.0%

 
24.9% of total persons are females who have sleep problems. Similarly 18.6% of the total people are males who have sleep problems.  

Part b

The total number of people surveyed is 269 out of which 152 persons have no problem with sleep. Out of the 152 people 81 are females and 71 are males who have no problems with sleep.

Part c 

Table 3: Chi-Square Tests
	Value	df	Asymp. Sig. (2-sided)	Exact Sig. (2-sided)	Exact Sig. (1-sided)
Pearson Chi-Square	.422a	1	.516
Continuity Correctionb	.277	1	.599
Likelihood Ratio	.423	1	.516
Fisher's Exact Test				.538	.300
Linear-by-Linear Association	.421	1	.517
N of Valid Cases	269
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 52.63.
b. Computed only for a 2x2 table

 
In order to analyse the difference in problem with sleep across gender the Chi-square test is used.

The test shows that c² (1, N = 269) = .422, sig = .516. Since the p-value is more than .05, level of significance, hence there is no significant differences in problem with sleep across sex.

Part a

In order to recode the syntax used is :

Transform > Recode into different variable

 

In the Old value - range, Lowest through value "6" was written and New value "1" was written. Similarly, in the Old value - range, value through Highest "7" was written and New value "2" was written. 

 

Next in the variable view the value labels were changed as

1 – “with health problems”

2 – “In good health”

Part b

In order to analyse if there are difference in general health of people with problem with sleep the Chi-square test for association was used. 

Table 4: Chi-Square Tests
	Value	df	Asymp. Sig. (2-sided)	Exact Sig. (2-sided)	Exact Sig. (1-sided)
Pearson Chi-Square	4.909a	1	.027
Continuity Correctionb	4.233	1	.040
Likelihood Ratio	4.877	1	.027
Fisher's Exact Test				.039	.020
Linear-by-Linear Association	4.890	1	.027
N of Valid Cases	266
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 21.99.
b. Computed only for a 2x2 table

Table 5: problem with sleep? * Rate General Health Crosstabulation
			Rate General Health		Total
			With Health Problems	In Good Health
problem with sleep?	yes	Count	29	88	117
		% within Rate General Health	58.0%	40.7%	44.0%
	no	Count	21	128	149
		% within Rate General Health	42.0%	59.3%	56.0%
Total		Count	50	216	266
		% within Rate General Health	100.0%	100.0%	100.0%

 
The likelihood ratio tests shows that there are significant differences in problem with sleep and people with health problems and people with good health, c² (1,266) = 4.877, p-value = 0.027.

Results and Interpretation

Part c

The Chi-square tests shows that there are statistically significant differences in problem with sleep between persons “with health problems” and “in good health” c² (1,266) = 4.909, p-value = 0.027, less than .05 level of significance.

Part d 

Table 6: Symmetric Measures
		Value	Approx. Sig.
Nominal by Nominal	Phi	.136	.027
	Cramer's V	.136	.027
N of Valid Cases		266
a. Not assuming the null hypothesis.
b. Using the asymptotic standard error assuming the null hypothesis.

 The effect size for the test is 0.136 (phi). Thus the differences in the groups is small.  

Part e

There are statistically significant differences in problem with sleep between persons “with health problems” and “in good health.” 

Table 6: Ranks
	agegp3	N	Mean Rank
sleepy & assoc sensations scale	<= 37	79	123.52
	38 - 50	79	114.62
	51+	72	107.67
	Total	230

Table 7: Test Statisticsa,b
	sleepy & assoc sensations scale
Chi-Square	2.162
df	2
Asymp. Sig.	.339
a. Kruskal Wallis Test
b. Grouping Variable: agegp3

 
In order to investigate the differences in the sleep and association scale in the three age groups the non-parametric test Kruskal –Wallis test was used.

The Kruskal Wallis test shows that there are statistically no significant differences between the groups, c² (2) = 2.162, p = .339.   

To investigate the differences between the groups one-way ANOVA (between groups with planned comparisons) was used. Further Post Hoc tests were done using the method of Tukey HSD. The groups which were tested for differences were Hospice, Hospital with LCP and Hospital without LCP.   

The Null Hypothesis: There is no difference in mean level of communication between the three groups. 

The Alternate Hypothesis: There is difference in mean level of communication between the three groups. 

The Null Hypothesis should be rejected since p-value < .0001, less than .05, level of significance. Thus, there was statistically significant differences between the three groups at p = .05 level, F(2,255) = 16.6, p < .0001.

The results of one-way ANOVA showed that there were statistically significant differences in the average scores of the three groups for the four composite scales. In order to check for differences in the groups and compare two groups individually the post-hoc analysis was done.

The Post-Hoc analysis shows that there are statistically significant differences between Hospice group and Hospice and LCP group, p-value = .002, less than .05, level of significance. Similarly the Pos-Hoc analysis shows that there are statistically significant differences between Hospice and Hospital without LCP, p-value <.0001, less than .05, level of significance.

On the other hand there are no statistical significant differences between Hospital with LCP and without LCP group, p-value = .86, more than .05, level of significance.

The results show that there are statistically significant differences in communication to dying patients and his families between hospitals with Hospice and Hospitals with LCP. Similarly there are statistically significant differences in communication to dying patients and his families between hospitals with Hospice and Hospitals without LCP. On the other hand there are statistically no significant differences in communication between Hospitals with and without LCP. The communication behaviour of Hospitals were assessed using ECHO-D questionnaire.     

Conclusion

The one-way ANOVA for the spiritual need – next of kin has F(2,255) = 22.6. The results show that there are statistically significant differences between the three hospital groups since p-value for the test < .0001, less than a = .05, level of  significance.

The Post-Hoc analysis shows that there are statistically significant differences between Hospice group and Hospice and LCP group, p-value = .006, less than .05, level of significance.

Similarly the Pos-Hoc analysis shows that there are statistically significant differences between Hospice and Hospital without LCP, p-value = .0001, less than .05, level of significance.

Similarly the Pos-Hoc analysis shows that there are statistically significant differences between Hospital with and without LCP, p-value = .0002, less than .05, level of significance.

The results show that there are statistically significant differences in the spiritual need – next of kin between all the three groups. The spiritual need of the bereaved family was assessed using ECHO-D questionnaire.   

Palliative or end-of-life care is provided to patients who are in their final hours of death. Such a patient would be unable to provide information that has been accessed by Maryland et al. (2014). The evaluation of the support that has been provided to a patient on end-of life can best be provided by the family of the patients. Even though it is a difficult time for the next-of-kin for such a family but the information for research that Maryland et al. (2014) did can only be provided by the patient family. 

One-way ANOVA was used to assess the differences in groups of composite scales and variables. Since Maryland et al. (2014) considered three groups of hospitals – Hospice, hospitals with and without LCP hence one-way ANOVA is a very appropriate test for testing all the three groups at the same time. A t-test on the other hand is used to analyse the differences between two groups only.

A chi-square test was done between RAAPS users and non-users and race. The chi-square for race was, c² (2,N = 210) =  1.2865, p = .53.

Since, the p-value (.53) for Chi-square is more than a = .05, level of significance, hence there is no significant association between race and RAAPS users and non-users.

A chi-square test was done and statistically significant differences were found between % of adolescent patients and RAAPS users and non-users, c² (1, N = 154) = 7.3780, p = .01.  

The Null hypothesis for the question is: There is no association between RAAPS users and non-users for years in practice.

The Alternate hypothesis: There is an association between RAAPS users and non-users for years in practice.

From the Chi-square test it can be said that there is no association between RAAPS users and non-users for the years in practice, c² (1, N = 157) = 6.2587, p = .01. Thus, we accept the Null Hypothesis, since p-value .01 is less a = .05, level of significance.

Table 2 of the paper by Darling-Fisher et al. (2014) presents the comparisons of demography of the patients between RAAPS users and non-users.   7 demographic characteristics are tested for differences in RAAPS assessment. For 5 of the demographic profiles the p-value is less than a = .05, level of significance. Thus for 5 of the demographic profiles – provider type, practice setting, % adolescent patients, years in practice and US practice region there are statistically significant differences between RAAPS users and non-users. For the demographic profile race and provider age in years there is statistically no significant differences between RAAPS users and non-users.