Quantitative Assesment World Medical and Health Games (Research Paper Sample)

World Medical and Health Games
Analytical Report
Software used: Statistical package for social science (SPSS)
1.0 Introduction
World Medical and Health Games (WMHG) is an international sporting event that is held on annual basis in the month of June in France and targets medical and healthcare professionals. The event objectives included a programme of participatory individual and team sports event, an academic symposium, and a range of social events and trips. A survey research was conducted jointly by the local authority and host venue in Stirling (Scotland) targeting 4000 medical- and health-related professionals’ participants from across the world with an aim of establishing city’s attractiveness as a tourist destination. The sample size constituted 204 participants, selected through a systematic random strategy, completed an interviewer-administered survey on the last day of the event. The yielding data was coded and entered into SPSS. This report provides a critical evaluative analysis of this data relating to the experiences of the participants experiences during the event.
2.0 Appropriateness of sample size
Sample size plays a critical role in determining the accuracy and the credibility of the results. According to lenth (2001), sample size forms a crucial aspect of any study design. Indeed, if a study design includes very small samples, then it will not be able to yield results that poorly reflect the population parameters. On the other hand however important a large sample size or entire population is in yielding credible result, its implementation is limited by costs, subject availability, time and the overall complexity of the task. With this regard, Hinkle (2003) notes that, in determining the appropriate sample size that will produce credible results and is feasible three factors should be put into consideration; precision level, confidence level and variability of attributes under evaluation.
The level of precision commonly referred to as sampling error represents the range in which the true value is estimated to be and often expressed as a percentage. The confidence level also known as risk level refers the probability that the value of a parameter obtained from the samples falls within a specified range of the true population parameters. The variability of attributes refers to general distribution of population under investigation. Therefore, a researcher’s choice of these variables in determining the sample size communicate the credibility of the results obtained from such study (Hinkle, 2003). A high sample size has a lower margin of error and this error increases with reducing sample size. It is therefore recommended for researcher to obtain adequate sample size to minimize the error margin for the yielding results to be credible.
As such the adequacy of the sample size used in this research depends on the choice of the three values of these factors. According to Lai (2001) the most widely and accepted method for sample size calculation is Cochran’s (1963) formula;
n0=Z2pqe2
Where n0 is the sample size, e is the precision error, Z is the desired confidence interval, p is the proportion of the factor under evaluation in a population and q=1-p. However if the population number is known finite population factor is applied to slightly reduce the sample size without significantly affecting the quality of the results by the following formula.
n=n01+(n0-1N)
In the light of the current survey, p is not known and therefore, Crawley (2002) recommends use of p as 0.5 in this case. If a 95% confidence level and ±5% error is desired then;
n0=1.962*0.5*0.50.052=385, but with use of fpc =3851+(385-14000)= 351 persons
This sample size of 204 used in this study is less than the desired minimum sample size of 351. It produces a margin of error of 6.78% which is higher than the desired 5%.
Then it is concluded that in light of the accepted ±5% error and 95% confidence level commonly applied the sample size is not adequate. However, according to Lai (2001) a large sample size (n≥30) is capable of producing statistically reliable results.
3.0 Use t-test
T-test lies in inferential statistics category that is used to draw conclusions from the population under study. T-test offers an opportunity for a statistician to compare two groups on differences of study unit scores. Crawly (2002) notes that the t-test analysis goes behold description of values from a data sample and seeks to draw inferences and conclusions among the population under study. In fact, t-test is used to analyse the difference between two means i.e two averages obtained from different group scores. It conveys a message whether the two means under evaluation are significantly significant. According to Crawley (2002) t-test can be used under two scenarios; dependent (paired) and independent samples t-test.
In paired sample t-test the two groups under consideration are somehow meaningfully related, for instance, where relation involves a pre-test (treatment) and post-test (treatment) of a sample unit’s research design. However, in independent samples t-test the two groups are not related. The hypotheses are stated in such a way that they are mutually exclusive.
The null hypothesis is that there exist no difference between means i.e
H0: μ1 = μ2
And the alternative is that there exist significant difference between means of the two different groups’ i.e
H1: μ1 ≠ μ2
In the light of comparing between males and females, independent t-test is evaluated for the comparison of the number of nights stayed to participate in the games and spending during the stay in Stirling. T-test can be used test whether the amount spent during the stay in Stirling and the number of nights stayed to participate in the games are significantly different between males and females category. This is because the data collected is continuous and in scale form. Further, the male and female categories are independent sample prompting use of independent sample t-test. Hinkle (2003) also puts forward that if the sample size is greater than 40 then, it is appropriate for t-test but is should also not have outliers. Outliers are values that lie abnormally far from the mean. By use of 1.5 x IQR Rule, outliers are values defined to lie below Q1-1.5IQR and above Q3 + 1.5 IQR. Where Q represent the respective quartile and IQR is the interquartile range. An exploratory analysis on the estimated total complete cost of stay in Stirling found that case 23, 24, 49, 83, 126 and 145 were outliers (Appendix 1). It is therefore appropriate to exclude them for T-test to be carried out on the average spending on the categories of male and females because apart from having negative effect on the dependent t-test, they can reduce results validity and also affect the statistical significance of the test as noted by Lenth (2001).
3.1 The number of nights stayed to participate in the games between males and females.
In testing the different between means of the number of nights stayed between males and females the following hypothesis are generated for testing
The null hypothesis is that there exist no significant difference between mean number of nights stayed for males and females i.e
H0: μ1 = μ2
And the alternative is that there exist significant difference between mean number of nights stayed for males and females i.e
H1: μ1 ≠ μ2
Group Statistics


Gender of respondents

N

Mean

Std. Deviation

Std. Error Mean

No. of nights staying

Male

119

7.30

1.670

.153


Female

78

6.77

2.221

.251

Independent Samples Test


No. of nights staying


Equal variances assumed

Equal variances not assumed

Levene's Test for Equality of Variances

F

7.519



Sig.

.007


t-test for Equality of Means

t

1.920

1.811


df

195

132.749


Sig. (2-tailed)

.056

.072


Mean Difference

.533

.533


Std. Error Difference

.278

.294


95% Confidence Interval of the Difference

Lower

-.015

-.049



Upper

1.081

1.116

The test results indicate that there is a mean difference of 0.533 days with males having a higher mean of 7.30 indicating that men used longer time to participate in the games. In testing the equality of variances by use of Levene's Test, p-value (“Sig.”) < .05 we reject the null hypothesis of equal variances and thus use the second line of t-test results. In testing the equality of the two means, the, p-value (“Sig.”) = 0. 072>.05, therefore we do not reject the null hypothesis. This leads to the conclusion that the number of nights spent are not significantly different for males and females.
The independent t-test in this evaluation not appropriate because the dependent variable ‘more than 7 nights’ category is of unbounded interval and can lead to biasness in estimation of the mean. Secondly, the t-test may not be appropriate because it does not show any significant difference in the mean number of night spent between males and females.
However, independent T-test can be appropriate because it shows the mean difference between males and females number of nights stayed.
3.2 Spent Amount during the stay in Stirling between males and females
The null and alternative hypotheses are;
H0: μ1 = μ2 i.e there is no significance difference in the mean amount spent between males and females in Stirling
Ha: μ1 ≠ μ2 i.e there exist a significance difference in the m...