Statistics Project Case Study Examples

The case objectives of this research were to indentify the variations in children’s risk of an unintentional or intentional fatal injury arising from allegations of physical abuse, neglect or other maltreatments. Secondly, they were examining whether previously referred children for physical abuse had an increased risk of both unintentional and intentional fatal injury as compared with children referred to other forms of maltreatment.
The population studied was 514232 children. This consisted of children born in California and those who were referred to Child Protective Services (CPS) in the California state between the years 1999 and 2006.
In the research, several variables were measured. These were classified into three folds. First and foremost, were the dependent variables which sums up to the injury deaths among children referred for maltreatment. Secondly, the independent variables were also measured. This category was further sub-divided into physical abuse, neglect and other maltreatments. Maltreatment details were strictly derived from the established administrative systems used by CPS workers in California State to enter emotional abuse, sexual exploitation and abuse of a sibling among the children. Last but not the least, was the other covariates. These were eight birth covariates which generally affect the antenatal and post-antenatal births among the study area or rather the population. These include child gender, child health, health insurance used for the birth, maternal race/ethnicity, maternal age, maternal education, paternity and finally birth order. A controlled experiment was employed in this research. This involves isolating the effects of one variable on a system while holding the rest constant. The research tested for each of the three categories (abuse, neglect or other maltreatments) separately while holding the others constant.
The research employed different types of sampling. These which were used were stratification. This involves splitting the entire population into sub-populations. For instance, the research sub-stratified the population into children exposed to physical abuse, neglect and injury. In addition, random sampling was also employed. This is where a sample is chosen on a standard basis such as mathematical tables and each item stands and equal chance of being selected. Moreover, haphazard sampling criterion was also employed. This is where there is no method of any kind to sample selection.
The research also had different sample sizes. For instance, in the first linkage phase, out of a population of over four million children born in California between the years 1999 and 2006, only 24987 cases of deaths reported in California. In a different linkage, CPS record extracted from 1999 and 2006 on maltreatment before the age of 5 years only 596962 children were successfully matched.
Confidence interval approach was employed. For instance, the use of probabilistic linkage software program version 2.0 to analyses data. This software employs a statistical model for weighing the likelihood that two records represents the same individual with a confidence level of 95%. Later an upper class limit and lower class limit are set. The individual outcomes are tested along these predetermined outcomes. This statistical test meets the required cannons. The assumptions made were the use of normal distribution and categorical data. This was ensured in sample population since the samples were stratified in accordance to their composition and similarities. Secondly, 2-proportion Z-test was also employed. This is used to test the degree of variance of two or more groups in a study. In the study the variance of injuries due to three causes were measured. According to me, the regarded assumptions on the statistical tests were met. For instance, in 2-proportion Z-test, a random population for each of the groups was selected. Secondly, there was also a categorical data from CSP offices in California that was used in the study. Sufficient level of random sampling was met because all the samples had an equal chance of being picked.
Hints of conflict of interest on the authors’ side could be as a result of questioning the credibility of information. For instance, 14% of the CPS records were not properly linked to birth records. This is a big variance in statistics and would raise eye brows. Additionally, the authors coded maltreatment allegations hierarchically based on their own interests. These interests may be biased and do not apply to the real cases on the ground. This too may lead to the questioning of the data. One of the assumptions in the study is that a child under the care of well educated parent or guardian will be less prevalent to injury as compared to one under the care of a less educated caretaker. Secondly, deaths in children due to fatal injury were very rare since the children could be treated on time due to the availability of Child Protective Services in California.