Introduction to Biostatistics

Introduction

• Biostatistics involves applying statistical principles to biological questions.
• This includes medicine and public health.

Agenda

• The topics covered are the definition of Biostatistics, the necessity of statistics, populations and samples, and scales of measurement.

Why Statistics are Needed

• Statistics are needed as a tool for biologists to evaluate publications critically.
• Statistics are neded to know how to undertake studies using correct methods, to generate proper results.

Statistics

• Statistics summarizes and collects numbers.
• Statistics is a method of handling and analyzing quantitative data.

Lies and Statistics

• Statistics should reveal truth, but can be misused to mislead.
• Valid data may be cherry-picked, presented deceptively, or misinterpreted.

Recognizing Statistical Manipulation:

• Watch out for misleading graphs, which can exaggerate or downplay trends by changing axis scales or skipping axis values.
• Be wary of small sample sizes that can lead to misleading results.
• Be aware of data cherry-picking, which supports a claim while ignoring contradictory evidence.

Selective Sampling and Causation

• Sampling from only certain groups leads to biased data collection and results.
• Correlation does not equal causation; two things happening together does not mean one causes the other.

Parameters

• A parameter is a characteristic or number that is related to a population in statistics
• A parameter is fixed, unknown, and summarizes the entire population.
• Parameters describe population properties; they are estimated using sample statistics.

Samples

• The principle of simple Statistics is to evaluate a parameter from an estimated sample.
• A sample is a subset of the population.
• Each unit of the sample provides an observation.
• A sample has a limited set of observations from a population.

Population

• A population includes all subjects with a quality in common for study.

Parameter vs. Statistic

• Parameters describe a population-related characteristic.
• Statistics describe a sample-related characteristic.
• Parameters use Greek letters, and statistics use Roman letters
• Parameter values are fixed, while statistics vary
• An example of parameter is the average height of all students in a university.
• An example of statistics is the average height of 50 randomly selected students

Why Sample?

• Populations are usually to large to get data from all objects in the population
• Gathering data has a cost as more data leads to higher costs

Sampling and Inference

• Techniques extract samples from a population.
• Parameters are used to describe the population.
• Inferences, conclusions, and estimates are used from those samples to the population.

Randomness and Independence

• Ideal samples are random, free from bias, and representative of population.
• Samples are random and individuals in the population have an equal chance of being drawn
• Each sampling unit is assigned with a unique number
• The observation in a sample will be independent

Variable vs. Parameter

• Variable: Attribute that can assume different values.
• Parameter: Characteristic or measure obtained from a population

Statistical Types

• Observation may be a count, and data may be descriptive.
• Variable studied can be a number of ants, and the parameter may be ants by type.
• Sampling unit is the space for the ants, and samples are all the units examined across.
• Measurements lead to statistical populations.

Descriptive vs. Inferential Statistics:

• Descriptive statistics is collection, organization, summarization, and presentation of sample data.
• Inferential statistics is generalizing from sample probability and relationships that can be found between variables, to make future predictions.

Example Data

• A sample is the group of households in the city.
• A variable is the household's monthly electricity consumption measured in kWh.
• A sampling unit is an individual household.
• The population parameter of interest is the average monthly electricity Consumption.
• The population studied is all the households int the city.

Variable Types

• Two types of variables are Categorical (Qualitative) and Quantitative
• Qualitative answer what kind and are categories such a adult and juveniles.
• Quantitative anser How much, how many and take number form values.

Scales of Measurement

• Qualitative can be Nominal and Ordinal.
• Quantitative data will be Discrete and Continuous.
• Examples of Qualitative are Binary, performance satisfaction.
• Examples of Quantitative data are length and size.

Nominal Scale

• Nominal Scale is has application or designation of numbers, and has integers.
• Assigned numbers lack ranking or ordering
• Mathematical operation is meaningless with a Nominal Scale
• Used for label like race, genders, religions, marital status etc

Ordinal Scale

• Ordinal scales is a more advanced, rank based approach.
• This involves classifying data into categories that can be ranked
• Burn categories, 1st 2nd 3rd etc

Interval Scale

• Interval Scales measure at meaningful differences between values, but there is no meaningful zero
• Ex, temperature and time are elements that can be used

Ratio Scale

• Ratio scales are interval data that has order, equal intervals and natural zero point
• Comparative to interval data; make assumptions based on this.

SPSS Software and Measurement Scales:

• In SPSS, the Variable View window can determine which column of data is Nominal and which is Scale.
• Nominal encompasses Nominal/Ordinal scales.
• Scale signifies Interval/Ratio scales.

Random Versus Systematic Error

• Random error implies that the effects are random.
• Systematic implies some pattern of error.
• Random error does not relate to the true score.
• Systematic error adds a constant.

Reliability and Validity

• Consistently and repeatability leads to reliability.
• Measuring similar properties of the same item on the same properties
• Validity is the test and ratings scales associated to measurements

Variance, Standard Deviation, and Standard Error

• Variable is σ² or s²
• SE deals with sampling variation, not individual data points.
• Standard Error measures uncertainty in estimating the population mean value
• A larger sample size gives a better estimate of the true population mean