Final Deck 3 Flashcards
Contrast descriptive/inferential statistics
- Descriptive statistics describe a single variable/distribution
- Inferential statistics make inferences about a population based on a sample (uses 2 variables and compare them to each other)
Mean, median and mode use which type of statistics (descriptive or inferential?)
Descriptive
Standard Deviation uses which type of statistics? (descriptive/inferential)
Descriptive
Correlation uses which type of statistics? (descriptive/inferential)?
Inferential
T-Test uses which type of statistics? (descriptive/inferential)
Inferential
Anova uses which type of statistics? (descriptive/inferential)
Inferential
How to use the mean and standard deviation of standard (deviation IQ) scores to determine normal limits.
On standardize tests, the SD is 15. So if the mean score is 100, the range of normal limits is between 85 and 115 (1 SD below/above the mean)
Define correlation
The way in which two variables are related to each other. Correlation considers strength and direction of the relationship
Strength (as part of correlation)
- The consistency of the pattern (how close the dots are to the line)
- *the closer the dots are to the line, the stronger the correlation.
- *if the dots are farther away from the line, the weaker the correlation
- r value closer to positive or negative 1 means a stronger correlation
- r value closer to 0.3 means a moderate correlation
- r value close to 0 means a weak or non-existent correlation
Direction (as part of correlation)
- Is r positive or negative?
- Positive r- both values are either increasing/decreasing
- negative r- one value is increasing, the other is decreasing (inverse relationship)
What are two disadvantages of a correlation test?
- Does not identify confounding variables
- Does not explain why the variables are related
- Affected by range of scores/outliers
- Not effective for curvilinear relationships
Describe a t-test
- a t test can compare 2 groups only on another variable. Ex: pre-test, give intervention, post test → you are comparing group of scores 1 vs group of scores 2 on another variable (ex: intervention)
Independent Sample T-Test
- A type of t-test where unrelated samples comparing the results
- Two different samples of participants test under different conditions → Ex: Comparing control/treatment groups
Related Sample T-test
- One sample tested under two separate conditions
- Comparing related groups → Ex: comparing pre-test and post test after the same dependent variable (intervention)
Describe ANOVA
- testing to see if there is a difference between more than 2 variables.
- Ex: pre test, intervention, post test, 6 months go by, follow up test.
What is the difference between a t-test and an anova?
T-Test looks at two variables only, Anova looks at 3 or more
What are non-parametric tests?
- A non-parametric test is capable of analyzing quantitative data that is ordinal or categorical, or does not conform to assumptions of normality, linearity, and homogeneity
(every parametric (ie t-test/anova) has a non-parametric equivalent)
Name 4 non-parametric tests
- Mann-Whitney U test
- Wilcoxon signed-ranks test
- Kruskal-Wallis ANOVA by ranks
- Friedman two-way ANOVA by ranks
Define standardization
- the process that specifies standards for a method of measure to be used across places.
- Standardization limits variability and increases the accuracy of our test scores.
- same way of administering a test across everywhere to keep it all the same and increase accuracy of scores
Define norm-referenced
- any assessment where scores are interpreted in comparison to a normed group.
- Most tests are age-normed.
What is the difference between standardized and norm-referenced
- Standardization focuses on consistency in the administration/scoring of an assessment (consistency/comparability of scores)
- Norm referenced means comparing the test scores to a normative group (ie. neurotypical children)
- *a lot of tests are standardized and norm referenced
Define reliability
- Consistency or repeatability of results obtained using a specific method of measurement
Define inter-rater reliability
- consistency between observers who are scoring the same test.
- Ex: if two administrators give the same test to the same child- there should be similar scores.
- There can be subjectivity if you are both observing the same thing.
- *high interrater reliability scores mean the test was given similarly
Define test-retest reliability
- consistency over time.
- Ex: testing each participant twice and make sure that they got the same score each time.
- *they should get around the same score each time you take it
Parallel forms reliability
- Consistency of scores between the same test but in different forms.
- (ex: professor makes test a and test b, the questions are the same and in a different order- the form a and form b scores should be around the same for each participant).
- Not common because there aren’t form a and form b between standardized tests.
Describe how correlations show reliability
- Because if there is a relationship between them, that means they are similar in some way, and the test was given similarly?
- *you want consistent positive correlations to show reliability
Define validity
- the extent to which a test measures or predicts what it is supposed to
- the degree to which the evidence and theory support the use of test results for a specific purpose (ex: the SAT is not intended to be a predictor of future success, it is intended to measure math/literacy skills)
content and response processes (validity)
- Do the wording/format/types of questions match the construct you are assessing?
- Ex: having multiple choice vs. essay questions
- does the way the test is given match the construct you want to measure?
Internal Structure (validity)
- Do the test components agree with the construct structure (weighting)? →
- if we have 10 units to cover in class, and half the final exam is only on the first two chapters, the internal structure doesn’t match the assessment
Convergent Validity
- Test results compared to measures of the same construct → how do the results on this assessment instrument compare to other results from other assessments that measure the same thing?
- Ex: how do results from the GFTA compare to results from the DEAP?
If they get low scores on one test, they should get low scores on the other test. If it is consistent, then the test is thought to have high convergent validity
Discriminant Validity
- The opposite of convergent validity
- Test results compared to tests that measure different constructs.
- Ex: comparing a receptive language result to an articulation test result.
- *This is done because you want to rule out other constructs that you don’t want the assessment to measure.
- *You are looking for no relationship (a correlation coefficient closer to 0).
- You want to demonstrate that your test is consistent with tests that measure the same construct, but it is not similar to tests that assess constructs that are irrelevant to your test.
External Validity- describe
- *does not assess an assessment but rather a study
- Can the results of a research study be generalized from the sample to the general population?
→ Obviously, the whole population can not be tested. - But, if we do a good job sampling, it should look like the population.
- This allows us to generalize the results from the sample to the intended population.
External Validity and random sampling
- *You want to do random sampling to reduce bias
- You also want a large group because the higher the sample size, the more likely you will end up with a sample that looks like the population because the extremes start to cancel each other out
Define Internal Validity
- Are the changes in the dependent variable due to the independent variable or something we didn’t intend to measure (confounding variable)?
Define confound
- A variable that is unintended and causes confusion about the result or difference
Define placebo
- A person in the control group reports positive benefits because they think they’re getting the treatment, but they’re really not
Blind
- The participant doesn’t know what you are testing so it doesn’t impact the way they respond
Double blind
- The participants and the experimenters don’t know what’s being tested so it doesn’t impact the way they respond
Triple blind
the participants, experimenters, and data analysts do not know what’s being tested so it doesn’t impact the way they respond
Define anonymous
You do not collect identifying information about the participants
Define confidential
You might collect identifying information, it is tightly stored
Anonymous vs. confidential
Confidential is more secure, it is tightly stored - anonymous does not collect identifying information
Define consent
Voluntary agreement of an adult with abilities to legally and functionally make decisions
Assent
Agreement from an individual that has some ability to understand/function, but can’t consent on their own
(children, people with low cognition)
What is the difference between consent and assent
Consent is given by adults with full capacity to understand, assent is from other populations that may partially understand
