27 terms

Healthcare Statistics Ch 11


Terms in this set (...)

descriptive statistics
describe populations, which can refer to patients, medical services, nursing units, or hospital depts. Provide an overview of the general features of a set of data. can assume a # of diff forms, the most common being tables and graphs.
scales of measurement of categorical data
nominal, ordinal, ratio, interval - ratio and interval data considered metric variables
metric variables
numeric variables that answer questions of how much or how many
nominal data
lowest level of measurement, pertaining to a name - observations organized into categories in which there is no recognition of order. examples are true/false, male/female, types of insurance carriers, football jerseys or patient occupations. often #s are used to represent categories & simply serve as labels for some piece of info, used for convenience only. just labels. avgs cannot be computed on this level. the proportion (or how many) that falls into each category is reported.
ordinal data
types of data where the values are in ordered categories. only the order of the #s is meaningful, not the #s themselves. this is because the intervals/distances between categories are not necessarily equal. example is level of severity where 4 is fatal, 3 is severe, 2 is moderate and 1 is minor. a natural order exists among the groupings w/largest # representing the most serious or could be reversed. another good example is Likert scale for surveys. grades, levels of satisfaction.
interval data
include units of equal size, such as IQ results. there is NO zero point. most important characteristic is that the intervals between values are equal. an example of interval scale is time, temps.
ratio data
the highest level of measurement. there is a defined unit of measure, a real zero point, and the intervals between successive values are equal. ratio data may be displayed by units of equal size placed on a scale starting with zero and thus can be manipulated mathematically such as 0, 5, 10, 15, 20. an example is age
question like this on quiz
A physician asked you to help her collect info on effects of drinking alcohol on pregnancy and birth weight of babies. you were asked to collect the following info.
-did mothers drink alcohol during pregnancy, yes/no-nominal
-birth weight of baby-ratio
-apgar score at 1 min-ordinal
-apgar score at 5 min-ordinal
numerical statistical data
two types: discrete data and continuous data
discrete data
finite numbers that have only specified values. examples are # of children in family. #s represent actual measurable quantities rather than labels. # motor vehicle accidents in particular community, # of times woman has given birth, # of new cases of cancer in state within past 5 yrs, # beds available in your hospital. a natural order exists among possible data values. measurements on nominal and ordinal scales are discrete.
continuous data
represent measurable quantities but aren't restricted to certain specified values. can take on fractional value. examples are fever temps, height, age. only limiting factor is degree of accuracy with which it can be measured. for analysis, often are converted to a range that acts as a category like 0-20, 21-30, 31-40. measurements on ratio and interval scales can be grouped and are continuous.
data display
critical to data analysis as it reveals patterns and behaviors.
when prepping a stat report
user must define objectives and scope by: what info is needed? what info is avail? are data collected routinely by facility or must addl data be collected?
data must be in table or graph
if purpose requires frequencies, percentages or relationships among variables.
statistical tables
used for summarizing data, they simply list values into rows and columns and do not easily capture audience attn.
graphs and charts
can present data for quick visualization of relationships
an orderly arrangement of values that groups data into rows and columns. almost any type of quantitative info can be grouped into tables. columns allow you to read data up and down and rows allow you to read data across.
frequency distribution tables
shows values that a variable can take and the # of observations associated with each value. a variable is a characteristic or property that may take on different values. for example, 3rd party payers, discharge services and admission day are examples of data. may show the proportion of patients admitted on any of the days. value is divided by the total.
best means for presenting data for quick visualization of relationships. supply a lesser degree of detail than tables. data in graph can be helpful in displaying statistics in concise manner. they grab audience attn and easy to understand. show trends or comparisons. should be easy to read, simple in content and correctly labeled.
bar graphs/bar charts
appropriate for displaying categorical data. simplest is one variable bar graph.
pie chart/pie graph
method of displaying data as component parts of a whole. use when want to show percentage of total.
line graphs
used to show data over time. consists of line connecting series of points. allow for several variables to be plotted. aka run charts in quality mgmt field.
graph used to display frequency distribution for continuous numerical data (interval or ratio). created from frequency distribution tables. like a bar graph but they are touching to show continuous nature. bars should be equal width.
frequency polygon
similar to histogram as it depicts frequency of continuous data but is in line form instead of bar form. advantage is several can be placed on same graph to make comparisons.
attractive alternative type bar graph that uses pics to show frequency of data.
scatter diagram
aka scattergram or scatter plot. used to graphically show relationship between two numerical variables. used to determine if there is a correlation/relationship between 2 characteristics. if two are somehow related, the pattern of points will show tight clustering in certain direction. the closer the points look like a line in appearance, the more the two are likely to be correlated.
implies that as one variable changes, the other also changes. doesn't always mean there is a cause-and-effect relationship between two variables because there may be other variables that could cause the change.