Dr Mac's email: List of topics to be familiar with for mid-term Flashcards
What are inferential statistics?
Inferential statistics allows a researcher to generalize the results from a sample to a population through hypothesis testing.
What are descriptive statistics?
Descriptive statistics are used to describe data or depict data to be organized summarized and described in a format that is more easily understood We can present the data with the use of graphs, charts, tables, an for numerical measures.
Give an example of inferential statistics.
Suppose you are hired by the US dept of health to examine how Americans feel about healthcare reform. You could survey every American about their opinion of the reform but this would be lengthy. Instead you take a sample of Americans by following a strategy to ensure that this sample accurately represents all Americans. The collected sample asks the participants their opinions. These opinions/data are then used to draw conclusions. These conclusions can then be inferred into the rest of the population.
Give an example of descriptive statistics.
Collecting data such as age, GPA, family income, and standardized test scores from applications. Through descriptive stats the administrators at the college are able to understand the characteristics of the student population admitted to the college in a given year.
Know the difference between an independent and dependent variable.
Independent variable is the variable to be manipulated by the researcher. Independent variable affects the dependent variable.
Dependent variables are affected by and the end result of the independent variable.
example: Examining whether hep B antigen affects liver function test results. The presence or absence of the hep b antigen is the independent variable. The results of the liver function test is the dependent variable.
Nominal measurement
Means to name or to categorize. A discrete variable
Ordinal Measurement
Mutually exclusive categories. Ranking or ordering is imposed on categories. A discrete variable
Interval measurement
Classified into categories with ranking and are mutually exclusive. Specific meanings are applied to distances between categories. There is no absolute value of zero. A continuous variable
Ratio measurement
There is a meaningful zero, and an equal proportion is present. A continuous variable
What level of measurement is this: Gender, ethnicity, zip code, religion, medical diagnosis, names of medicines, hair color
Nominal; One group cannot be deemed better than another group-no rankings
What level of measurement is this: Likert scale, histological rating, pain scale, age groups, grades, performances
Ordinal; there is a ranking or ordering of the values
What level of measurement is this: IQ, education achievement tests, temperature, SAT score, depression score, time of day, date (years)
Interval; no absolute value of zero
What level of measurement is this: Income, age, height, weight, BP, years of work experience, time to complete a task
Ratio; there is a meaningful zero
Why sample a population?
Because it is too costly to do a whole population; take a sample and try to infer what we find from sample to population
Random Sampling
o Selecting subjects based on chance alone; the strongest approach to sampling; based on equal chance of selection
o Challenging if population includes unknown elements; requires that the entire population is known
o Ex: if a researcher wishes to draw a random sample of 10 from a population of 100, each subject has 10 chances in 100 of being selected
Simple random sampling
o Taken so that all subjects in a population have an equal opportunity of being selected.
o Identify all subjects in the population and then randomly select
o Uses a random number table or generator
Systematic Random Sampling
o Begins with assigning a number to each subject in the population and then selecting every kth person
o Order of subjects needs to be random
o Ex: Select every 20th subject
Stratified Random Sampling
o To ensure that certain groups are represented equally in the sample
o Stratified by gender/sex, race/ethnicity, income
o Divide the population into groups (strata) based on elements and then randomly select subjects from each group
o More precise than simple random sampling; homogenous
When is a stratified random sample proportionate? When is it not proportionate?
o Proportionate if the subject size of each stratum in the sample is in proportion to that in the population
o Nonproportionate if the subject size of each stratum is not proportionate to that of the population
Cluster Sampling
o Similar to stratified, used when a population is large or when a study covers a large geographic area
o Heterogenous; different subjects
o Dividing a population into different groups (clusters)
o Less precise than simple random sampling or stratified
o More cost effective and feasible
What are the four types of random sampling?
Simple random sampling, systematic random sampling, stratified random sampling, cluster sampling
Nonrandom sampling
o Do not rely on chance, no equal chance of selection
o Prioritize feasibility or access to the population of interest
o Ex: Testing a drug for pancreatic cancer so you find patients with the disease
Convenience Sampling
o Based on accessibility of subjects
o Less representative of the population
o Used a lot in clinical research
o Ex: every student in a course at school
Volunteer Sampling
o Only those who offer themselves as participants in the study are included as the sample
o Inexpensive way to ensure a sufficient number of subjects for a sample
o Obtains skewed opinions on the characteristics of interests as this group will have a narrower range of opinions than a randomly selected group
o Most health related research, even when samples are randomly drawn
Quota Sampling
o Dividing the population into mutually exclusive groups and selecting subjects from each group
o Similar to stratified but not randomly selected
o May use convenience sampling within each quota
o Limits generalizability, may be proportionate
Snowball Sampling
o Uses word of mouth, nomination, or referral to accrue subjects
o Researcher must make at least one contact with a subject and then that subject nominates others
o Useful for finding hidden subjects
o Can be biased or misleading results
o Ex: find teens doing drugs
What are the four types of nonrandom sampling?
Snowball, quota, volunteer, and convenience
Numeric values may be?
Discrete or continuous
Discrete or Categorical variable
values that are countable but do not include the fractions between countable categories
Continuous variable
Have every possible value on a continuum
Confounding variable
any uncontrolled variable that may influence the outcome of a study
Open coding
the most basic coding, the first step taken to group data into logical categories
Axial coding
Advanced form of coding, takes the analysis process further by generating categories, themes, and patterns; the second step in coding after open
Focused coding
identify patterns and multiple layers of meaning, and to delineate variations and interconnections among sub-themes within the general topic; in grounded theory, a more finalized set of coding
Why do coding?
To stay close to the data, to give quantifiable answers to qualitative data
For content analysis, to quantify or provide frequency, to categorize the text
What is variability?
The degree to which scores in a distribution are spread out or dispersed; in quantitative research
Can either be homogeneity or heterogeneity
Data for a variable can be completely described in terms of the shape of the distribution, central tendency, and variability