Transform Data: Dataset And Variables

Transform Data Student’s Name Institutional Affiliation Transform Data Identify the dataset and variables and provide context to the research used to collect the data (in an introduction section). In recent years, there have been attempts to elevate the quality of healthcare services provided by hospitals. Consequently, in May 2005, the US Department of Health via the National Quality Forum (NQF) embarked on a survey to gather information on patients’ satisfaction for the experience they had in various hospitals. The forum represents a consensus of several health care providers, consumers (in this case patients), professional associations, buyers, federal agencies, and certified research and quality organizations. Patients in the in-patient unit were the primary target of this data collection methodology. To avoid bias, patients wrote down their perspectives of care provided by the hospitals they had visited. Information collected was then compiled and sorted out to come up with a public report of the whole survey. Comparisons were made across all hospitals and meaningful conclusions drawn as per the responses of the public or rather patients. For standardization, a uniform measurement approach was incorporated for fairness. The spade to spade approach gave birth to HCAHPS Survey i.e. Hospital Consumer Assessment of Healthcare Providers and Systems Survey. Basically, the HCAHPS Survey can be defined as a compiled set of questions with a wider personalized set of hospital-specific items that are a supplement to data collection. As a compliment, these items aid in improving internal customer services and the overall quality of the hospital related stuff. The main objective of the HCAHPS Survey is to come up with measurable comparisons among these healthcare facilities that can be used to: * Improve the quality of health care * Help consumers make a choice on which facility to visit based on the rating * Enhance public accountability via transparency of hospital’s quality To ascertain the public of its authenticity, the survey ensures that it remains credible, useful and above all realistic. All information in this survey is accessible to the public for its convenience. A detailed approach of the survey includes twenty-one patient perceptions on health care and rating items that consolidate to a nine key topic approach (Drukker, 2011). The topics include communication with doctors, with nurses and about medicines, the responsiveness of entire hospital staff, pain management techniques, overall cleanliness of the hospital and its immediate environs, in/out-patient discharge information, discreetness of the hospital surroundings and care transitions. Four screener queries and seven population-related items focus on tuning a patient’s mix in the hospital beside the analytical purpose. The governance modes approved in the survey are mail, telephone, and active interactive voice responses. Entirely, the survey has a total length of 32 questions. In this paper, the dataset on scrutiny is one from the HCAHPS Survey. The variables in the dataset include; * Hospital’s provider ID * Name of the facility * Address * City * State * ZIP code * County name * Phone number * HCAHPS measure ID * HCAHPS question * HCAHPS answer description * Patient survey star rating * HCAHPS answer percent * HCAHPS answer percent footnote * HCAHPS linear mean value * Number of completed surveys * Survey response Start date End date Location. The period when the data was collected starts from 1st July 2017 to 30th June 2018. Identify the analysis performed, pros and cons of calculation, and why they are used (in a method section). Data transformation in statistics is significantly different from transformations in mathematics (Crawley, 2014). By definition, data transformation entails the use of a function, for example, f(y), thus replacing variable x with z. The f(y) function works best on invertible and continuous data. The importance of transforming data is so that the data can be interpreted or graphically represented better than before (Hand, 2008). Additionally, results of data after transformation is fit to meet assumptions of statistical inference in the application. Recall model fit assumptions:- * Normality assumption. * Linearity assumption. * Equal spread condition. * Independence Assumption. These assumptions are necessary for parametric tests where data is assumed to be normally distributed and has homogenous variance. Transforming residuals ensures that the data upholds the normality assumption. Usually, independent variables are subject to transformation (Crawley, 2014). However, in some cases, both independent and dependent variables are transformed in case they greatly deviate from the normality assumption. Generally, most measurements follow a normal distribution. Besides, some are log-normally distributed. As a norm, dependent variables follow a continuous distribution and may take up values from 1 to infinity. In our case, the selected variable ‘HCAHPS Answer Percent’ is a dependent variable even though it has missing values noted as ‘Not Applicable’ (NA). The distribution of this variable is however continuous. To comprehend distributions of variables and data transformation in statistics, one must have first-hand knowledge in measures of variables. Briefly list wise, measures used in inferential and descriptive statistics include: * Measures of dispersion i.e. range, quartile, interquartile range, five number summary, variance, and standard deviation. * Measures of central tendency e.g. mean, median, and mode. * Measures to describe the shape of the distribution, such as skewness and kurtosis. Our interest is on the five-number summary and measures to describe the shape of the distribution. The five number summary includes minimum, first quartile, median, third quartile, and maximum. The minimum is the last entry and the maximum is the largest entry in the dataset. Q1 and Q3 are entries at 25% and 75% in the dataset, while the median is the central entry in the dataset. The difference between the Q1 and Q3 is IQR or interquartile range. Half of all values in the dataset are within this range. A large IQR translates to a large spread of value and vice versa. Skewed distributions could be positive or negative. Positively skewed distributions-tail is on the right have positive values meaning the majority of frequent values are few. Corresponding transformations comprise of square root, cube root, and logarithm. Negative skew has a negative value and the tail is on the left. For such, square root, cube root, and log have constant-x. Kurtosis is also termed as negat...