Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
81 Cards in this Set
- Front
- Back
Name the different levels of measurement |
Nominal ordinal Interval Ratio |
|
Nominal data |
Dividing objects into groups that don't overlap (ex: male/female) *Can assign numbers to groups - 1-f/2-M *one category is not better than the other *Chi square & Cochran Q |
|
Ordinal Data |
Distinct ordered groups (scale): ex: mild, moderate, severe *scale does not have to be linear or consistent with reality *Mann-Whitney U test & one/two-way ANOVA |
|
Interval Data |
Distances between points on a scale are known and constant (ex: IQs and scores on standardized testing) *Equally appearing intervals *No absolute zero *T-test & ANOVAs |
|
Ratio data |
Basically interval data with scores that represent: *Distinct groups *Ordered levels *Equal intervals *Same tests as interval: T-test & ANOVAs AND have an absolute zero: zero point |
|
Interaction effects |
The combined effects of two or more independent variables on a dependent variable in a factorial design |
|
Main effects |
Differences between groups or conditions on one independent variable that are NOT influenced by other independent variables. |
|
Descriptive statistics |
measures of central tendency and measures of variability |
|
Mode |
*Most common number in a set of numerical data *used for NOMINAL data because the scores are not meaningful and have no scale |
|
Median |
*Middle score in a set of numerical data *Used for ORDINAL data, sees which category encompasses the middle score *Distance between scores in unknown so mean cannot be used |
|
Mean |
*Average *Used for INTERVAL data & RATIO data because there are ordered categories and equal intervals *outliers can affect mean |
|
Name the three measures of central tendency (mentioned in class) |
Mode Mean Median |
|
Name the Measures of Variability |
Range Standard deviation |
|
What is a measurement of variability? |
How much data varies around central tendency |
|
Range |
The difference of the highest and lowest values *Not good for NOMINAL data *Outliers can affect range |
|
Standard Deviation |
*Most used for measuring variability *Based on average deviation from the mean |
|
1 SD = how much of the population? 2 SD = " " 3 SD = " " |
1. 68% of the population 2. 95% of the population 3. 99% of the population |
|
Which is better a higher or lower standard deviation? |
LOWER |
|
Parametric is used for what type of data |
Interval & ratio |
|
Non-parametric is used for what type of data |
Nominal & ordinal |
|
Distribution becomes more normal as sample size gets |
Larger |
|
What does a small sample size threaten? |
Besides generalizability, statistical power |
|
If a test is not set up for normal distribution, what type of data is used? |
Non-parametric data |
|
If there are two different variances between groups, than what type of data is used? |
Non-parametric |
|
Which type of tests are more powerful? Paremetic or nom-parametric? |
Parametric |
|
If a t-value is more than the critical value, it is |
Significant |
|
If a p-value is small than set alpha level, it is |
Significant |
|
Type 1 error in Hypothesis testing |
Researcher concludes there is a significant difference when there is none |
|
Type 2 error |
Researcher concludes that there is no significant difference when there really is one |
|
When can a type 1 error happen? |
P-value is not strict enough |
|
When can a type 2 error occur? |
The alpha level was set too high and needs to be reduced Also, can occur with a sample size that is too small |
|
Mean difference: explain |
This refers to how different the mean scores are, the larger the difference, the less likely that difference is due to chance |
|
Variability: explain what this means if a large variability is found |
*The larger the variance, the more this suggests variability in population. *If you replicated the study, you would likely get a different result |
|
Which sample size is more likely to reject the null hypothesis? |
A larger sample size |
|
When determining a main effect, what are you doing with the independent variables? |
You collapse across variables, and compare one IV while ignoring the other |
|
When a graph is parallel, is there an interaction? |
No, this indicates no interaction |
|
When the lines of a graph are parallel, what does this mean for the main effect? |
If the lines are going upward it means there are two main effects |
|
What's involved when determining an interaction? |
You are looking to see how tow IV interact |
|
What statistical test do you use to see an interaction? |
2-way ANOVA -- not multiple t-tests because the ANOVA takes into account how many analysis' have been done |
|
What do you need to remember when explaining these interactions/main effects concepts? |
TO specifically state the main effect/interaction AND to differentiate between analyzing for the main effect (IV only) versus interactions (IVs against each other) |
|
Type of test used for measuring Effect size |
Cohens-D |
|
Differences of the mean/SD What number is a small effect size? What number is a medium effect size? What number is a large effect size? |
1) 0.2 Small
2) 0.5 medium 3) 0.8 large
|
|
ANOVA |
The ANOVA asks the question, is there a significant difference among more than two groups? ***It is an extension of a T-test but you do not want to do a bunch of t-tests |
|
What counteracts the problem of multiple t-tests? |
The Boneferroni |
|
Using the Boneferroni, you use what ratio instead? |
The F-ratio which is like a t-test to test significance |
|
Which ANOVA is a multivalent study? |
A one-way ANOVA is a multivalent study but only one IV |
|
Which ANOVA is a parametric study? |
A two-way ANOVA |
|
Three-Way ANOVA? |
Three IVs |
|
2 by 3 by 2 ANOVA |
1 IV with two levels 1 IV with three levels etc., |
|
Scatterplot is used to... |
see predictor versus un-predictor variable |
|
Post-Hoc |
An ANOVA tests with more than two groups to see where the main effect is |
|
A priori power Analysis |
completed before testing |
|
why might you do a power analysis after testing? |
To test error variance, you want to see if the effect size and alpha level in the study had adequate power to avoid type II error |
|
On a scale of 0-1, what is enough power? |
0.8 |
|
What else can a power analysis be used for? |
The power analysis may be used to see how many subjects researchers need to avoid a type II error |
|
Single case versus group studies |
*Both experimental (cause & effect statements can be made) *compare two or more conditions to see if they have statistical differences *use measures of variability to see if any difference of conditions is above other variables affecting participants |
|
What do single case designs use? |
They use multiple measures on the same person |
|
What do group designs measure? |
They measure variance outside of IV due to multiple particpants |
|
Generalizability in single versus group design |
Groups use average parametric tests, but who does the average pertain to? Single cases use individual performance, but you can't generalize to anyone but that individual |
|
Advantages of smaller sample sizes? |
*Good if limited availability of subjects *More clinically significant (but not statistical |
|
Levels of evidence |
Hierarchcy's of categorizing rigor of scientific research |
|
Systematic reviews (level 1) |
*Comprehensive *Clear research question *Avoid publication bias by hand searching *Includes grade about level of evidence supporting research *logically defined inclusion and exclusion criteria *Includes grade about level of evidence supporting resarch *recommendations for future research |
|
Meta-Analysis (level 1) |
*Slighter stronger than systematic review *Meta-Analysis are results that provide confidence intervals *Provide summary effect estimates (average effect included in meta-analysis weighted by size of each contributing study) |
|
Deciding which studies to include in systematic review/include in meta-analysis |
*Look @ research question *Do an unsystematic review (there might already be research, but can expand as well) *Check the database of systematic reviews (Cochrane database of systematic reviews) -scoping search *Formulate Critical Research question *PICO *Evaluate the articles |
|
PICO |
P = Population I = Intervention C = Comparative treatment - what are you comparing the two, other levels of IV O = Outcome - DV |
|
Level 2 |
*Double blinded (fixes Hawthorne and Rosenthal effects *prospective (you choose aspects before data collection, randomized, controlled clinical trials *Prospective Randomized Clinical trial (PRCT) ***Subjects were enrolled to participate before study ***Randomly assign groups before study ***compare for statistical significance due to intervention |
|
Level 3 |
Non-randomized (convenience sample) intervention studies (quasi-experimental *Quasi-experimental *Convenience sampling |
|
Level IV |
Non-intervention studies (descriptive studies) *prospective - cohort: studies that follow a group of participants over time; I.e., who gets the disease at a later date (e.g. meningitis following CI) *retrospective - case control: retrospective study to identify factors associated with a disorder of condition, starts with two groups with or w/o disease and compare the groups on predictor variables (e.g. personal listening devices and HL: compares two groups with and w/o HL and measures PLD use |
|
Level 5 |
Case Reports: study of an indiviudal in detail |
|
Level VI |
Expert opinion of respected authorities |
|
Strong experiments.... |
*Ask research question first *Researcher in control of selection and exclusion criteria *measurement of IV and DV |
|
Boneferroni Procedure |
An adjustment procedure in which the alpha level is made more stringent when a statistical analysis is used multiple times on data gathered from the same participants, done to reduce a type I error |
|
Methods of analyzing relationships |
Nominal Data *Contingency coefficient (c) *CHi Square (x2) Ordinal Data Spearman Rank *Order correlation Coefficient (RHO) Interval/Ratio Data *Product moment Correlation Coefficient (r) *Multiple regression analysis |
|
Methods of analyzing differences (within Subjects) |
Within Subjects (related samples) Cochran Q WIlocoxon Matched pairs signed ranks test (t) Friedman Two-Way Anova T-test for correlated groups z-ratio ANOVA (f) ANCoVA (F) |
|
Methods of analyzing differences (between subjects) |
Between Subjects (independent samples) Chi square test Mann-Whitney U test Kruskal -Wallis One-Way Anova t-test for independent gruops z-ratio ANOVA (F) ANCOVA (F) |
|
ANCOVA - Analysis of Covariance |
A parametric statistic procedure related as a counterpart to ANOVA that evaluates group differences while taking into account pretest differences or statistically controlling for other characteristics (covariates). |
|
Mann-Whitney U test |
A non-parametric statistic used to find differences between two independent groups |
|
Cohen's d |
A measure of effect size or practical significance |
|
Post hoc means comparison tests |
When an F-ratio for a main effect including more than two groups or an interaction effect is found to be significant in the analysis of variance, post hoc tests determine which groups or cells in the design are significantly different from one another |
|
One-tail tjest (directional hypothesis) |
One technique needs to be more or less |
|
Two-tail test (bi-directional hypothesis) |
either A or B technique needs to be more or less,k |