Statistics Flashcards
What are inferential statistics?
In inferential statistics we test the probability that the test hypothesis is true. The stats that we run give us the probability that the results from the data analysis is due to chance alone (i.e., due to a fluke).
What is quantitative research?
Quantifying relationships between variables
Measure variables with a sample of participants
Express the relationship between the variables
What are variables?
A concept or construct
can have more than one value (i.e., it can vary).
The value can be different for each research participant
What is a constant?
A concept or construct
always has the same value for each participant
What is an independent variable?
Explains or causes a change in another variable
Not dependent on another variable
Manipulated by the researcher
What is a dependent variable?
Explained or caused by IV.
Level is dependent on the other variable
Of interest in explaining and understanding
What is conceptualization in regards to measurement?
Accepting a nominal definition
What is operationalization in regards to measurement?
Accepting a way of measuring
What is precision?
How repeatable are measurements
What is accuracy?
How close is the measurement to the true value of what is being measured?
Two main stems of levels of measurement?
Categorical and numerical
Two main stems of categorical measurement?
Nominal and ordinal
Two main stems of numerical measurement?
interval and ratio
What does nominal data do?
Classifies
What does ordinal data do?
Classifies
Orders
What does interval data do?
Classifies
Orders
Equal intervals
What does ratio data do?
Classifies
Orders
Equal intervals
Has ab absolute zero
How are variables assessed in nominal data?
Variables are
Discrete – data falls into categories
Non-metric – no order in categories
How are variables assessed in ordinal data?
Variables are
Discrete – data falls into categories
Order in the categories; ranking
Tells whether participants are different
(one can be more or less than other group)
How are variables assessed in interval data?
Variables are legitimate numbers and have equal intervals between them.
No zero point
How are variables assessed in ratio data?
Have a true zero, equal intervals and is very precise
Main concerns for measurement?
Constraints (time, funds, equipment)
Validity (are we measuring the construct?)
Reliability (are the results repeatable?)
How do we know we a measuring a concept?
Face value
Content validity
Criterion validity
Construct validity
What should content validity be?
It should cover the entire range of meanings included in the concept.
Based on judgement
What is criterion validity?
Checking against an external criterion believed to be another indicator of the same construct
Main stems of criterion validity?
Predictive validity
Concurrent validity
Known groups validity
What is predictive validity?
Predict some future criterion that expected to predict?
➢Does attendance at BIOL1900 lectures accurately predict student performance on exams?
What is concurrent validity?
Relate to some known criterion concurrently?
➢Do scores on a measure of health-related quality of life correspond to the ratings based on clinician interviews?
What is known groups validity?
Differentiate people in the way you would expect?
➢Does grip strength differentiate between those of low and high risk of cardiovascular mortality?
What is construct validity?
Relates to other constructs in a way that is expected based on theoretical relationships
Two main stems of construct validity?
Convergent and divergent validity
What is convergent validity?
Associated with other measures that it should be related to.
➢Do scores on a vertical jump test correspond to Wall Sit Test times (leg strength)?
➢Do results on BIOL1900 Practice Quizzes relate to Mid-Semester Exam results?
What is divergent validity?
Does not associate with measures of other constructs as closely as it does with other measures of same construct
➢Do BSS scores relate more to measures of Sit & Reach than they do to Standing Long Jump?
Most critical components of reliability?
Degree of consistency
Degree of repeatability
Degree of agreement
Three ways to describe data?
Frequency, central tendency and variability
Three measures of central tendency?
Mean, median and mode
Three measures of variability?
Range, IQR and standard vardeviation
What does univariate refer to?
Frequency: involving one variate or variable quantity.
What does range tell us?
The amount of variation (dispersion) in a data set
How is range calculated?
By subtracting the minimum score from the maximum score
What graph is typically used for inter-quartile range?
Box and whisker plot
How is the inter quartile range calculated?
By subtracting the 25th percentile score from the 75th
Equation for probability?
Possible successful outcomes / # possible outcomes
What is a type 1 error?
You find a significant relationship but one does not exist in the real world
What is a type 2 error?
You find no significant relationship when one does exist in the real world
Three factors of the correlation coefficient?
Sign (+/-)
Magnitude (0 to 1)
Significance (p<0.05)
Why is it called p test?
Pearson’s r Test for significance
What are the two types of t-tests?
Independent-samples t-test and paired-samples t-test
What is an independent samples t-test?
Comparison of two different groups (eg control vs intervention)
What is a paired-samples t-test?
The comparison of repeated measures; one group on two occassions
Assumptions of a t-test
Independent variable only from two groups
Dependent variable normal distributed
Data measured on interval scales
The variance has homogeneity
How is a t-test calculated?
It is the difference between means, divided by pooled standard deviation
What is a chi-square test?
Test for a relationship between two nominal (categorical) variables
Test for independence of the two variables (ie that one is not dependent on the other)
Compares the obtained frequencies within each category to the frequencies expected in those categories by chance
Assumptions of chi-squared test?
Each person/item/entity contributes to only one cell
Expected cell frequencies should be >5
Calculation for degrees of freedom? (in chi-squared test)
n-1
How is a t-test calculated?
It is the difference between means, divided by pooled standard deviation