Research and Program Development Flashcards
Dependent Variable
The variable that is (HOPEFULLY) changed based on your experiment, the outcome variable
Independent Variable
What is controlled, utilized, given in the experiement
Parsimony and Occam’s Razor
Interpreting results in the simplest way
Confounding/Flawed Research
Multiple testing (seeing multiple counselors, receiving multiple treatments) that is not controlled for in the experiment)
Basic Research
Advances our theory
Applied Research
Advances our practices
Control group
Does not receive the experimental treatment, no exposure to the independent variable
Hypothesis Testing
Developing an experiment in order to explore a hunch or idea developed by R. A. Fisher
Null Hypothesis
States that the treatment or IV will not have an affect
Alternative Hypothesis
States that the treatment or IV does affect the outcome of the experiement
Between Subjects Design
Different subjects get exposure or lack of exposure to different things
In Subjects Design
One pool of subjects receive or don’t receive the treatment (pre-test post-test design)
Parameter
Property that defines a sample (age, sex, etc.)
Probability
The likelihood that something will happen, also known as significance level, P for our field is generally .05 but can range from .000 to .01 and still be considered significant
findings or not due to chance, P can be translated into a percentage that describes the portion of the sample whose results were achieved by chance (i.e. .05 = 5% of the sample’s scores were obtained by chance – not your experimental design)
Type I Error or Alpha Error
Rejecting the null when it is true (saying there is significance in your treatment when there isn’t), increasing P levels will reduce this error, increasing sample size will reduce this error
Type II Error or Beta Error
Accepting the null when it is not true (saying there isn’t significance in your treatment when there is), increasing sample size will reduce this error, increasing P will increase the chance of this error
T Test
Used for two samples to compare means, you obtain a single t score and compare it to the critical t value based on the sample size and your significance level and if the t value you found is greater than the critical t you have significance
F Statistic
Used for more than two groups, represents and ANOVA test, same process used as with the t test
Two Way ANOVA or MANOVA
Used for more than two groups and more than one IV
Correlation
Represents a relationship between two variables, ranges from -1.00 to 1.00, the closer to -1 or 1 the stronger the relationship, can have negative correlation or positive, a score close to 0 represents no or low correlation, strong correlation does not imply causality
Baseline Measure
Testing before any IV has been performed
Single-Blind Study
Either the researchers or the participants (but not both) are unaware of what group each represents
Double-Blind Study
Neither the researchers nor subjects know what category or group they belong to
Normal Curve
Bell shaped, mean, median, and mode all fall on the same line, 68% of scores fall in -1 to +1 standard deviation, 95% fall within -2 to +2 standard deviations, 99.7% fall into -3 to +3 standard deviations
Negatively Skewed or Left Skewed
Distribution with outliers towards the negative side of the x axis
Positively Skewed or Right Skewed
Distribution with outliers to the postive side of the x axis
Mean
Average of scores, most commonly used statistic, represented with X with a bar over it, strongly effected when dealing with a skewed distribution
Median
Middle score
Mode
Most common number/score, score obtained most frequently
Bimodal Distribution
Has 2 modes or peaks
Factorial Design
Has more than one IV
Raw Score
Simplest view of a score, need more information to compare or evaluate the score
X Axis or Abscissa
Horizontal axis where IV scores are recorded
Y Axis or Ordinate
Vertical axis where you plot the frequency of the dv (deviation)
Replication
Equates to increased reliability
Range
Measures a spread of scores by subtracting lowest score from highest
Scatterplot or Scattergram
Each score represents a point on the graph, can give a visual representation of correlation
Variance
Measure of how scores are arranged around a measure of central tendency (mean, median, or mode), this is standard deviation squared (if SD for a sample is 4 then the variance is 16)
Z Scores
The same thing as a standard deviation, also known as standard scores
Platykurtic Distribution
Low, long, flat curve
Leptokurtic Distribution
High, spiked, narrow curve
Nominal Scale
Simplest type, catagorical (i.e. male, female, democrat, republican, etc)
Ordinal Scale
Ordered scale (i.e. 1st, 2nd, 3rd most important, etc.)
Interval Scale
No true zero, numbers represent true, distinct, equal distances (i.e. IQ score, temperature in Celsius or Fahrenheit , etc.)
Ratio Scale
Has a true zero, numbers are true, distinct, and equal distances (i.e. Kelvin, height, weight, etc.)
Naturalistic Observation
Researcher does not manipulate or control variables, just watches/observes/records
Survey
Simplest form of research, need a 50-75% return rate to establish accuracy
Placebo Effect
Showing an effect or reaction to a treatment that you believe you are getting but are not really beign exposed to
Hawthorne Effect
If subjects know they are being observed, they tend to perform better
Rosenthal Effect or Experimenter Expectancy Effect
If the experimenter provides other observers with information (they will excel or they will do worse, etc.) then the observers notice changes
Halo Effect
When a trait which is not being evalutated impacts the observer’s rating
Statistical Regression
Implies that the more a test is administered, the more scores will move to the central mean
Standardized Test
Are normed and have specific proceedures for scoring and administering
Counterbalancing
Changing the order that iv are administered
Random Sampling
Made by change, every member of the population has an equal opportunity
Stratified Sampling
Allows for specific characteristics to be represented in random sampling to mimic the overall population
Cluster Sampling
Used when the population whole is not known, not as accurate as a random sample
Horizontal Sampling
Subjects are selected from single socioeconomic group
Vertical Sampling
Subjects are selected from two or more socioeconomic groups
Systematic Sampling
Pulling every nth person from the sample (2000 in your population, you pull every 5th, 5, 10, 15, 20, 25, etc)
Parametric Test
Scores are normally distributed
Nonparametric Tes
Scores are skewed
Inductive Logic
From specific example to generalized
Deductive Logic
From generalized knowledge to specific
Standard Error of Measurement
Allows you to predict a person’s score if they were to retake a test
Likert Scale
Numerical range that represents how someone feels or their opinion