Reliability and Validity
This word is associated with the concept of "validity."
"Truthfulness"
"Accuracy"
"Consistency"
Truthfulness
In a hypothesis, this is the variable that is thought to produce an effect or difference.
The independent variable
This is a word that means "average." It is a measure of central tendency.
Mean
Cohen's d is a measure of effect size when looking at the ________ between two groups (means).
* Association
* Reliability
* Difference
Difference
If a p value < .05, then we ______ the null hypothesis.
Reject. We would retain the null hypothesis if the p value is > .05.
The following is a not a type of reliability:
"Inter-rater"
"Infra-rater"
"Intra-rater"
"Test-retest"
Infra-rater
Identify the independent and dependent variables in the following hypothesis:
"Attention improves more quickly in patients with cognitive communication deficits following TBI when speech therapy begins 2 weeks post-injury compared to 6 weeks post-injury."
Independent: Speech therapy timing (i.e. 2 weeks vs. 6 weeks post TBI)
Dependent: Improvement in attention
This is a measure that captures the average distance of each data point from the mean.
Standard deviation
True or false: a correlation (r) of 0.7 is considered weak.
False.
Correlation of < 0.2 is considered weak, 0.3 – 0.5 is considered moderate, and anything > 0.5 is considered strong.
A p-value tells us the ________ of getting our results if the ______ hypothesis is true.
Probability/likelihood/chance
Null
Which concept can be measured as a percentage: validity or reliability?
Reliability
This is a variable where the difference between values is meaningful, such as temperature. It does not have a true "0" measurement.
Interval
If a data point has a z-score of +3, what does that tell you about the data point?
* It falls within 3 percentage points of the mean.
* It falls 3 standard deviations above the mean.
* It is considered reliable.
It falls 3 standard deviations above the mean.
Effect sizes help you to determine...
* The likelihood of getting your results, or those more extreme, if the null hypothesis is true.
* The magnitude of difference or association between variables.
* How far from the mean a single data point falls
The magnitude of difference or association between variables.
True or false:
Each statistical test is associated with its own distribution. For example, the z-test is associated with a normal distribution.
True.
Fix the error in this statement:
An invalid measure can never be reliable, but an unreliable measure can still be valid.
An unreliable measure can never be valid, but an invalid measure can still be reliable.
In a non-experimental study, what are the two ways of dealing with nuisance/confounding variables?
Matching or controlling
Trick question (this is actually an inferential stats question!):
What is the name of the hypothesis that states there IS a difference or association between two study variables?
The alternative hypothesis (H1)
Two-part question: What does the effect size r2 tell you? Is it used to measure difference or association between variables?
r2 tells you how much of the variability in one variable can be explained by another variable. It is a measure of association between variables (i.e. imagine you are associating temperature of a state with its latitude. If r2 = .68, that means that 68% of the variability in temperature in a state can be explained by the latitude of that state)
As Rhiannon asked in her lecture, which two factors related to the study's variables are important to consider when selecting a statistical test?
Whether the variables are categorical (i.e. nominal/ordinal) or quantitative (i.e. ratio/interval) and whether the researchers are looking for a difference or association.
How does a randomized experimental design (i.e. RCT) improve the validity of a study?
Randomization accounts for nuisance/confounding variables. This allows the authors to conclude that the independent variable is truly what caused a change in the dependent variable. In other words, they are measuring what they think they are measuring, which is the basis of validity.
A Likert scale is a commonly used measure in research with questionnaires (i.e. asking a subject to circle a number from 1 to 7 to indicate their agreement with a statement). What kind of variable is a Likert scale?
An ordinal variable
What percent of data in a normal distribution falls within 3 standard deviations +/- the mean?
99.7%
Trick question (it has nothing to do with effect sizes!):
What is it called when the effect of one independent variable differs based on another independent variable?
An interaction
To achieve statistical significance using an alpha level of .01, would your test statistic have to be higher or lower than when using an alpha level of .05?
It would have to be higher.