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;
56 Cards in this Set
- Front
- Back
|
Independent Variable a.k.a..... |
AKA Explanatory Variable. This is the variable that is manipulated and changeable. |
|
|
Dependent Variable a.k.a... |
AKA Response Variable. This is the variable that is being affected by the independent variable. The values that result from the independent variable. |
|
|
What are the 3 types of variables? How do you define each one? |
1. Measurement/Numerical (in numbers) 2. Nominal/Categorical (in "names") 3. Ranked/Ordinal (in positions, i.e. 3rd) |
|
|
Observational vs. Experimental Studies |
Observational: Where you depend on how the independent variable variates naturally
Experimental: Where you control/manipulate the independent variable. |
|
|
What are confounding variables? |
Variables that might influence the dependent variable. These variables are not the variables you are testing, but are those that might be mixed up and cause confusion when trying to figure out the cause of the dependent variable. |
|
|
Press vs. Pulse vs. Natural Experiments |
Press: Treatment conditions are constant and maintained over time Pulse: Treatment applied once in the beginning of experiment Natural: Taking advantage of the natural variations of the experiment |
|
|
The design of an experiment is directly linked to the details of..... |
Replication, Randomization, Independence |
|
|
Pseudoreplication |
When treatments aren't replicated & When replicates are not independent of each other. |
|
|
What are 3 types of Single Factor Experimental Designs? and What is 1 thing they all have in common? |
They all have 1 CATEGORICAL Independent Variable
1. Completely Randomized Design 2. Randomized Block Design 3. Latin Square Design |
|
|
Completely Randomized Design: Define, What does it require? |
- Design where all replicates are assigned randomly to all treatment levels. - Simplest Design
Requires: Relatively homogeneous (similar) environment; Replicates need to be similar before treatment is applied. |
|
|
Randomized Block Design: Define BLOCK, Define |
Block: Group of individuals that are known to share similar characteristics that will affect the response of the treatment
RBD: Randomly assigning treatments to individuals within blocks of similar conditions |
|
|
Benefits of Randomized Block Design? |
- Control over spatial heterogeneity (differences) - Conditions are more similar within blocks - Pairing the individuals within a treatment is not random within a block |
|
|
Latin Square Design |
Each individual (treatment level) occurring once in every row and column |
|
|
What are 2 types of Multiple Factor Experimental Designs? |
1. Factorial Design 2. Split-Plot Design |
|
|
Factorial Design |
- Where 2 or more independent factors are tested simultaneously tested in one experiment - Can test for interactive effects - Best factorial designs are ones that are fully replicated, crossing each independent variable |
|
|
Split Plot Design: What is it an extension of? And when is it used? |
- Extension of the BLOCK design - Has 2 or more experimental treatments - The experimental unit has already been "split" into sub units, and ANOTHER treatment is applied to these units. |
|
|
Nested Design |
Any Design where there is a sub-sampling within each experimental replicate.
Two layers of analysis
** Tests for variation between treatments and WITHIN the treatment. |
|
|
Repeated Measures Design |
Used when multiple observations on the same replicate are collected at different times. Control and experiment subject is the same. |
|
|
Central Tendency: What is it, 3 types |
Measure of the middle, most typical value. Can be expressed through: Mode, Mean, Median |
|
|
Dispersion: What is it, 6 types |
Measurement of how far stretched out the data is. - Range - Sum of Squares - Variance - Standard Deviation - Standard Error - Confidence Intervals |
|
|
Range |
Difference between the largest and smallest value |
|
|
Sum of Squares |
Sum of the squared deviation of each observation and the mean |
|
|
Variance |
Measures the amount of variation in the data. Can be measured by mean square deviation. Value is always 0 or positive |
|
|
Standard Deviation |
STD describes how close the set of values are to the mean. Measured by the square root of the variance |
|
|
Standard Error |
Measures how close the sample mean is to the population mean. |
|
|
Confidence Interval |
The amount of uncertainty/reliability associated with the sample mean |
|
|
How do we know if a data set displays normality? |
- Frequency Histogram is bell-curved - Mean = Mode = Median |
|
|
Skew: +/- directions, what is helpful in determining skew of a data set? |
Median is useful in determining the skew of the data plot.
Left + Right - |
|
|
How do you describe the shape of distributions? |
- Using statistics of tendency - Skew - Kurtosis |
|
|
Kurtosis: Define, Three types |
Degree of which the peakness is distributed
Mesokurtic: 0 Leptokurtic: + (narrow peak) Platykurtic: - (flat) |
|
|
Categorical vs. Continuous Variables |
Categorical: Variables that are non-changing Continuous: Variables that change |
|
|
p: Define, what does it indicate? |
Probability of outcomes
- It indicates the mode of a frequency distribution. Ex: If there is a 75% chance of getting heads over tails (25%), then p = .75 |
|
|
n |
Number of samples |
|
|
p x n |
Expected number of outcomes |
|
|
If p = .75 If p = .5 If p = .01 |
.75: Skewed to the right 5: Normal distribution .01: Skewed to the left |
|
|
What tests can you use on Normal Distributions with Categorical Data? |
1. Binomial Test 2. Poisson Test 3. Chi-Square Test |
|
|
Binomial Distribution: Define, What does it require, Categorical or Continuous Variables, Outcomes, # of Trials |
Probability Distribution
- Uses categorical variables - Only 2 possible outcomes that are independent of each other (throw 1 has no effect on throw 2) - Fixed number of trials - Probability of Success (p) is constant
|
|
|
Poisson Distribution: What is it good for, What does it need to define distribution, # of Trials, Outcomes, |
- Infinite number of Trials - Unlimited of Outcomes possible - Uses ONLY the MEAN to define the distribution - Variance = Mean - Number of Occurrences independent of the other - Used when variables of time and space are randomly distributed |
|
|
What is one thing that the Binomial and Poisson Distribution have in common? |
They can both be used to test for Independence! |
|
|
Binomial Distribution Disadvantages |
- Difficult if the probability of success/outcomes (p value) is very low or high - Cannot calculate if you do not know the n (number of outcomes) |
|
|
Poisson Distribution Advantages |
Good for:
- Random outcomes - Small predicted Values |
|
|
Chi- Square Test: When do you use it, Distribution Shape, What does it help? |
- Use when testing for multiple categories - Use when you have two nominal variables, each with two or more possible values - Distribution shape depends on Degrees of Freedom - Helps approximate the normal curve in large sample sizes |
|
|
Normal Distribution: What kind of variables, Mean and STD represents... |
- Mean represents location of peak/curve - STD represents dispersion - Continuous variables
|
|
|
Null Hypothesis vs. Alternative Hypothesis |
Null Hypothesis (H0): default, that there is not relationship between the two variables Alternative Hypothesis (H1) is the hypothesis that you're trying to test |
|
|
Alpha Values and P Values are used to test... |
SIGNIFICANCE |
|
|
P-value |
A number between 0 and 1 that tells you the significance of your results. |
|
|
Alpha Value: What does it test and tell us, what is it usually set at |
Tests the LEVEL OF SIGNIFICANCE
It tells us how extreme our observed results have to be in order to reject the Null Hypothesis. Also tells us how often a null hypothesis can be rejected (percentage chance)
Usually set at 0.05 (1/20 chance that H0 will be rejected) |
|
|
If we reject/fail to reject the Null Hypothesis, the p value is ________ to the Alpha Value |
Reject H0: p value is less than or equal to Alpha 0.05 (other factors than chance). Results are statistically significant Fail to Reject: p value is greater than Alpha 0.05 (observed data can be explained by chance alone). Results are statistically nonsignificant |
|
|
Type I Error |
False Positive. When you reject the Null Hypothesis and its actually true |
|
|
Type II Error |
False Negative. When you accept the Null Hypothesis and its actually false |
|
|
T-test: Determines what? Three assumptions? When do you use it? |
- Use when you have one measurement variable Determines if the difference between the sample mean and population mean are different. Check if two sets of data differ significantly.
1. All data are randomly collected + independent 2. Data is normally distributed 3. Each group must have similar variance |
|
|
One-Tailed t Test vs Two-Tailed t Test: Diff. btwn Observed and Expected Value Significance
How do you choose to use one or two-tailed tests? |
One Tailed: Difference is significant - Sample mean different from Population mean in 1 direction Two Tailed: Difference is not significant - Know that the two values on either ends are different, but don't know which group mean is larger. |
|
|
Chi-square Test Problems? |
- Inaccurate if expected values are too small - Increases "False Positive" Errors |
|
|
What tests can you use on Normal Distributions with Continuous Data? |
- Z-scores - Kolmogorov-Smirnov Test - Shapiro- Wilk Test |
|
|
Kolmogorov-Smirnov Test: Define, When do you use it? |
- Use when data is continuous - Tests against normal distribution - Compares difference btwn hSObserved vs. Expected - Used to decide if a sample comes from a population with a specific distribution. |
|
|
Shapiro-Wilk Test: What does it test |
- Tests H0 that sample came from a normally distributed population |
