Professional Documents
Culture Documents
Lecture No. 12 Community Dentistry
Lecture No. 12 Community Dentistry
Descriptive inferential
statistics statistics
• Purpose of Statistics
– Assemble, Organize and Analyze Data
– Draw Conclusions about Data
– Form Predications
• Usually a subset of the population called a sample is studied in
observational or experimental settings.
• Statistical analysis is descriptive or inferential
• Inferences drawn from the sample are extrapolated to the population
Data
• The health sector generates a lot of data. These can be routine data
such as patients’ records or from special studies. Patients’ data may
be clinical information, images or lab reports. Valuable information
can be generated from data thus collected.
• For research studies, data collection is standardized and systematic.
Studies are broadly classified as observational where the assignment
of subjects to exposure groups is not determined by the researcher,
but rather by the study subjects themselves, or by other factors, or
experimental, where the researcher controls the exposure variable.
A variable is a factor that can take different values for individuals in
a study. A variable may be quantitative, that is, it takes a numerical value;
or it may be qualitative or categorical, in which case it describes an
attribute.
For example, changes in body weight after specific periods on a
prescribed diet are quantitative because, for each subject, the
observation takes a numerical value.
In contrast, a child may or may not show evidence of dental caries at a
particular moment in time. In this case the observation describes the
presence or absence of a characteristic, and it is therefore qualitative
rather than quantitative.
DATA DISPLAY AND SUMMARY
• Types of Data
The first step, before summarizing or plotting of data, is to decide what
type of data one is dealing with as that decides how it is to be
summarized or depicted.
Data can be classified in a number of ways. One is as described before
i.e. qualitative and quantitative. Another way is to classify them as
follows.
Nominal Ordinal
Data is in the form of names, labels, or categories. There can be some sort of ordering but the
The data cannot be ranked or grouped in any order differences are meaningless,
at all. e.g.: Large – Medium – Small;
Examples: Gender, Race, Type of teeth etc. Good – Bad, Malnourished;
Normal – Overweight – Obese,
or Decayed– Missing – Filled
Interval Ratio
There are meaningful differences between data This is similar to interval, but ratio has a true zero
points. or starting point, e.g.: height, weight, length,
However there is no true zero, distance traveled, etc. Each point value can be
e.g. Knoop Hardness for composites. expressed as meaningful ratio to another e.g. –
However one data point cannot be divided by twice the weight
another to give a ratio.
One cannot say 100 C is twice as hot as 50 C.
Data is classified according to
the highest level which it fits
Each additional level adds something the previous level didn’t have.
• Nominal is the lowest level. Only names are meaningful here.
• Ordinal adds an order to the names.
• Interval adds meaningful differences
• Ratio adds a zero so that ratios are meaningful.
• Nominal and ordinal variables are summarized by counts and
proportions whereas interval and ratio variables by measures of
central tendency such as mean and measures of variation such as
standard deviation and range.
• Generally it is easier to summarize nominal and ordinal variables.
Hence, ratio variables are often converted to ordinal variables,
e.g.: height to Tall – Medium – Short, Blood Pressure to Hypertensive –
Non-hypertensive etc. However nominal data cannot be converted to
interval. Hence at data collection it is advisable to retain data type
instead of precoding it.
To make a clinical decision we usually do not need the exact value
e.g.: serum potassium levels. It would suffice if it was within the
normal range. Hence we often convert ratio scales to ordinal.
However by converting a ratio value to an ordinal one, information
is lost and the sensitivity of statistical interpretation is reduced.
Further, the conversion is based on statistical principles rather than
being done arbitrarily.