Study your flashcards anywhere!
Download the official Cram app for free >
 Shuffle
Toggle OnToggle Off
 Alphabetize
Toggle OnToggle Off
 Front First
Toggle OnToggle Off
 Both Sides
Toggle OnToggle Off
 Read
Toggle OnToggle Off
How to study your flashcards.
Right/Left arrow keys: Navigate between flashcards.right arrow keyleft arrow key
Up/Down arrow keys: Flip the card between the front and back.down keyup key
H key: Show hint (3rd side).h key
A key: Read text to speech.a key
67 Cards in this Set
 Front
 Back
What is a Concept?

A concept is something that can be quantifiable in some manner


What is a construct?

A construct is a special concept. However, it is not readily quantifiable. One must use a theoretical framework to decide for operational variables to test for the construct. Ex: Chair is a concept. Comfort of the chair is a construct


What scales are used?

Nominal
ordinal interval ratio 

What scales are used?

Nominal
ordinal interval ratio 

What is a nominal scale

Nominal scales are names that can be associated with numbers. It codes a variable into a number. This is the lowest form of a scale, and some places do not recognize this as such.


Characteristics of an ordinal scale

First real scale
magnitude varies along a continuum no math manipulations ex. class rank a track record a simple list of 1st, 2nd, 3rd pitfall: doesn't address distances between magnitudes 

Interval scale characteristics

equal distances on a number line explains equal maginitudes
can have linear transformations ex: temp scale  Celcius relative: pull to the right or left, distances remain the same shapes of distribution should remain the same If you see two shapes and want to see if they share the same interval scale, plot out and if in a straight line, they share the scale Each interval scale is also ordinal, and nominal pitfall: doesn't give info about proportions 

ratio scale characteristics

zero is fixed. can't be moved, isn't relative
proportion of magnitude is also equal to proportion on number line hierarchal in nature  upgraded interval scale Slope may vary depending upon the equation for transformation. However, it always goes through zero. Kelvin scale is example of temperature as a ratio. 

What is metaanalysis

summary of results over a huge body of studies.
You increase the validity of findings IF the study has been conducted properly. 

Qualitative study

Uses Inductive reasoning
generalize over separate studies 

Quantitative Study

Hard sciences like Psychology
Deductive Reasoning is used Uses more causal relationships (If Then) 

Major types of Validity

Content Validity
Criterion Related Validity Construct Validity 

Three types of Validity

Validity in Qualitative Research
Validity in Quantitative Research Validity in Measurement 

Explain Construct Validity

this is the degree to which a test measures the theoretical construct or trait it was designed to measure.
Based on current theory regardng the trait, the researcher makes predictions about how test scores will behave. The predictions are tested. If data supports prediction, construct validity is enhanced. If data are not supported: 1.the exp. was flawed 2.theory was wrong and should be revised 3. test doesn't measure trait 

How do you make a prediction support consruct validity?

Allow for group differences
allow for changes look for correlations Look at processes 

What factors threaten internal validity?

history, maturation, pretest effects, instruments, statistical regression towards the mean, differential selection of participants, mortality, and interactions of factors


What are the threats to external validity?

Interaction effects of selection biases and treatment, reactive interaction effect of pretesting, reactive effect of experimental procedure, and multiple treatment interference


Types of validity in Qualitative research

Description
Interpretation Theory Also, validity threats: Researcher Bias (researcher's existing theory or preconceptions and the selection of data stand out to researcher.) Reactivity (influence of researcher on the settings or individuals) 

Validity threats in qual studies

Researcher Bias
(researcher's existing theory or preconceptions) Reactivity (influence of researcher on the settings or individuals) 

Checklist of options to reduce validity threats

MO approach
(hold validity threats in contrast  treat as events) Search for discrepant events and negative cases analyzing these data is key! Triangulation collecting info from a diverse range of individuals and settings Feedback Member checks feedback from the people you are studying Rich data  "Thich" in that it is detailed and complete enough to reveal the picture of what is going on Quasistatistics running the simple numeric results that can readily be derived from the data. Comparison 

Generalization in Qualitative Research

Internal generalizability  gen of a conclusion w/in the setting or group
External generalizability  not as crucial  face value? 

3 steps that make a random design

1. Randomly select participants
2. random assignments of participants to groups 3. random assignemtn of treatments to groups 

Quasiexperimental design

deals with intact groups like classrooms


major goal of randomness

To provide control in equal conditions that may affect outcome of treatment


How to begin randomizing

1. Decide and define population
2. Select a random representative sample. Remember to look at clusters or strata for population. This shrinks image of population. This is called the stratification approach 

Validity with experimental design

Internal validity:
Does treatment produce any difference w/in a certain study? External validity: Does treatment make a generalization to a population? 

Sample group size

The more factors you introduce, the smaller the sample group. (think of this as refining a search on the internet!)


best random design?

Randomized Solomon 4group design


Reliability

The degree to whichthe measurement is accurate and precise


Types of reliability

Internal consistancy indicates to what degree the test scores reflect the same construct


Cronbach's alpha

varies between 01
0=low reliability 1= high reliability 

testretest reliability

reliability for stability across time points
ex. changes in vision 

Pearson correlation, r

r shows the correlation between x & y
value between 1 and +1 want a + score for test/retest 

content validity

1. face validity (arm chair)
2. logical validity  exam the degree to which items reflect content domain measured by test. all cells proportionally represented at all levels 

Criterion Related Validity

ex: entrance exam test


2 aspects to criterion related validity

Predictibility (predictive validity)
concurrent 

construct validity

to what degree does measurement measure what it is supposed to measure
theoretical frameworks operationalize measurable variables 

CFA

confirmatory factor analysis
will allow to test for hypothesized structure of test 

pvalues

when alpha is set before testing, the pvalue is the actual figure of error found. Thus, use the highest alpha you can afford, but typically is 5%.
So: if p=0.023 when an alpha has been set to 0.05, the error is less than 5%. Then we reject the null hypothesis. (Thus saying that the hypothesis is not equal to zero.) Accept the alternate hypothesis. The smaller the alpha, the easier it is to reject or accept the null. 

Variance problems

not all numbers are in the same unit, so must find standard deviation


Standard Deviation

drawbacks to finding SD:
when you square the variance, you're artificially inflating the numbers variance has different units of measure 

target variable

dependent variable
outcome 

Predictor variable

Independent variable


Variance proportion

square the correlation. This answers: What proportion of variance in Y is explained by the variance in x?


Variance

a magnitude that indicates the spread of scores about the mean.


finding variance

Ssquared=(x1meanx)squared+(X2meanx)squared....divided by n1


measurement validity

how well does the test measure the construct?
labeling construct will depend upon your background knowledge. 

measure of linear relationships

correlation between two variables


scatterplot design

the closer the dots are to a median line, the higher/stronger the relationship


finding standard deviation

take Ssquared and find the square root


covariance

if neg, neg correlation
if pos, pos correlation is found by taking 

covariance

if neg, neg correlation
if pos, pos correlation is found by taking 

sig

pvalue


semipartial, or squared part correlation

gives proportion of unique contribution


ANOVA

Analysis of Variance
a statistical procedure 

F

A statistic
higher F= lower p value=higher chance to reject null 

zero correlation

if line on graph is horizontal


Lack of relationship

big circle of plots on graph with no linear line


Curve linear or quadratic

when the line on a graph is curved


how to study a curve linear graph

Analyze the intervals to form a linear relationship
examine meaningful sub groups 

population coefficient

large P with x and y beneath


Null hypothesis

Ho=zero
This means no correlation 

Alternate hypothesis

Ha is not equal to zero


Type 1 error

to falsely reject the null hypothesis


Level of acceptable error

5%


Alpha value of acceptable error

0.05 level of significance


More on alpha

Alpha is the probablility of a type 1 error.
