Research designs Flashcards
Review for midterm
Research design come form 2 different types.These are:
1- Quantitative =
—-data = numbers
—-people= subjects
2- Qualitative =
—-data= words
—-people= participants
What is experimental research design?
- Design that determines if there is. Relationship btw 2 or more variables
- Used hugely in design for hypothesis
State 3 characteristic of an experimental research design?
1- independent-
2- dependant-
3- observing the effects of independent on dependent
why is control in experimental design important?
it better allows for a stronger conclusion/interoperation of the results
State advantages of Experimental research design?
CARE
- Convergence
- Replication
- Adjustment of variables
- Establishing cause and effect relationship
What does “establishing cause and effect” mean and?
- Having good framework
- use of correct statistical model and analysis
- Proper selection of independent, dependent and control variables.
- Correct interpretation of results
State ,3 criteria’s that shows that cause and effect relationship attained;
- The cause for establishing the effect in time
- Correlated with each other
- The correlation between then cannot be explained by another variable.
What are some disadvantages of experimental research design?
- Cost
- Inability to generalize
- Securing cooperation
- Depending on the type you choose, it can be quite complicated to design and implement.
Define Terms:
Independent variable
(MANIPULATED) –>
influence the other variable and is manipulated
Define Terms:
Dependent variable
(OBSERVED)–>
that is measured observed and cannot be manipulated
Define Terms:
Control or situational variable
(HELD CONSTANT) –>
researcher may not be able to
Define Terms:
Participants
name for the individuals in experimental researchers.
Define Terms:
experimental group
the group that receives a treatment.
Define Terms:
between subject designs
each subject tests only in 1 level for each independent variable (IV)
Define Terms:
between group designs
group of subjects are tested under only one level of each IV
Define Terms:
Within the subjects design
Each subject gets tested more than 1 independent variable, under more than 1 level of IV
Define Terms:
Within groups design
Group of subjects are tested under more than one level of each IV.
What are some advantages for:
“between group designs”?
–> No changes that one treatment can contaminate another since the subject/ group only receive one treatment
What are some disadvantages for:
“between group designs”?
–> Concern that is a possibility that the subjects or groups are different enough to influence the effects of the treatment.
What are some advantages for:
“within group designs”?
Each subject compared to themselves not influenced by the other
What are some disadvantages for:
“within group designs”?
–> Carry over effect: the effects on the prior test might disturb the effects of the other.
To over come this:
- Randomization of treatment levels
- Counterbalancing the treatment levels (make a systemic level)
Define term:
Random sampling:
A sample that is formed by the population that has been selected in unbiased way.
Each person has an equal chance of being selected
Define term:
Random sampling:
Subjects are assigned to condition in an unbiased way
Define term:
matching
Treatment/ experimental subjects are matched with a control subject.
Common matching items are:
o Agde
o Gender
o Weight
o Height
o IQ
o Years of schooling
Define term:
Matched group designs
- his approach helps to control for confounding variables that might otherwise affect the outcome, making it easier to isolate the effect of the treatment or intervention.
QN: Can you truly match items and eliminate everything?
- IT IS ALMOST IMPOSSIBLE because there is factors influences that a researcher simply cannot control.
- personal
-genetic
-environmental
Define terms:
Blind assignment
Subjects does not know if they are being testes.
Define terms:
Double Blind assignment
neither the researcher not the subject know that they are being tested.
type of experimental designs:
Define Pre-experimental:
Experiment done before the actual experimental group:
Lack random assignment
Uses short-cuts that are usually weaker the the OG
It is substituted when a researcher cannot use all the part of the OG experimental design.
Has weaker internal validity
type of experimental designs:
True / Classic Experimental
One of the most powerful designs:
- Generally, a good choice bcs:
— Random assignment
— Control group
— Experimental group
— Pretest and post-test for each group (both control and experimental) - Sometime post-tests only yon the control group
type of experimental designs:
Quasi-Experimental
Sometime a better choice bcs:
— Used when the researcher has limited control over the independent variable.
— these are stronger than pre-experimental
—- variations of classical experimental designs.
—–>Some will have:
- Randomization but lack a pretest
- More than two groups
type of experimental designs:
Randomized Control Trials
o (An experimental design specifically for clinical trials).
RCT is a very effective design with 2 types of trials. These are:
- Parallel group trial:
—- Similar to pretest- post-test control group design
—- Only one set of patients receives the new drug
—- Comparison between the 2 groups
- Crossover trial:
—- Initially, they are split into an intervention group (taking the drug) and a control group (taking a placebo), then after a set period, the groups switch treatments.
Why do we use sampling?
This is because we cannot look into everything but we can determine it though choosing a portion of a whole.
Define terms:
Representative sampling
–> It is about over and under representation
–> If you have a sample that takes everything account like the population that is alright, if not that is being considered as biased.
There are 2 types of variability:
- If bigger variability -> large sample to ensure reasonable representation
- If smaller variability -> small sample size will be sufficient
There are 2 different types of sampling techniques?
- Non-probability
—– Non random sample, use qualitative research, have 7 types - Probability
—– Use quantitative data, based on theories of probability from mathematics, Random sampling.
Why is randomization very important?
higher chance that yields to true representation of the population
what is sample error?
underrepresenting a population
Define term:
Sampling element?
unit of analysis/ case in population
Define term:
Target population
the specific pool of cases that are be studied
Define term:
Sampling ratio
ratio of the size of the sample to the size of target population
Define term:
sampling frame
list of cases in population researcher is going to choose.
Define term:
Parameter
characteristic of the entire population that is estimated form the sample
What is measurement in data collection?
– A repeatable, objective procedure for generating a measure.
– Set of possible numbers that may be obtained by the measurement.
Measurement scales have certain characteristics What are they?
MEA
- magnitude = An attribute can be judged greater than less or equal to another aspect of the attribute.
- Equally intervals = the measurements is same regardless the scale unit falls.
- Absolute zero = value that indicated a value at all I being measured.
Factors effecting measurements include:
- Continues variable
—- Interval can always be turned into ordinal or nominal BUT NOT VICE VERSA
—- Ratio level can be turned to interval, ordinal or nominal BUT NOT VICE VERSA - Discrete variable
—- ifinite number of values
What are the levels of measurement? most specific to not specific.
RION
1- Ratio ( Absolute) –> Continues –> most specific –> finite
2- interval ( Interval) –> Continues –> more specific –> finite
3- ordinal (rank order) l–> Discrete —> less specific –> infinite
4- nominal (types of categories) –> Discrete–> least specific –> infinite
What is validty?
The factuality of the results
There are 3 types of validity
1. Internal
2. External
3. Measurement
There are 11 threats to internal validity
silly hippos make tasty ice cream every saturday during cold evening strorms
- Selection bias
- History
- Maturation
- testing
- instrumentation
- experimental mortality
- statistical regression
- Diffusion of treatment
- compensatory behaviour
10.experimenter expectancy - sequence effect
What can you do to control the threats for the internal validity?
- Randomization
- Placebo’s
- Blind setups
- Reactive effects of testing eliminate pre-test
- Instrumentations
- Experimental mortality
Threats to 9 threats External validity?
- location of experiment
- Field experiment
- Population Validity
- Ecological validity
- Experimental realism
- Reactivity
- Howthrown effect
- Demad effect
9.Placebo
Controlling treats for External Validity
*Selecting from larger population
o participants
o treatments
o situations
*Ecological validity:
oDoes the setting capture the essence of the real world?
Measurement Validity:
The degree to which that data accurately represents the idea or concept you’re trying to study.
Define the Term:
Can you do that again?
What is the consistency across time.
test-retest relibility
consistency across items
Internal consistency
consistency across researchers
Interrater reliability
What is Representative reliability?
Measure that yield consistent results for various social groups.
What is Stability Realiabity?
Yields consistent results at a different point, while what is measured did not change
What is equivalence reliability?
reliability across indictor
Ways to improve reliability:
1-Clearly conceptualize all concepts
2- Increase the the level of measurement
3- Use pre-tests, pilot studies and replication
4- Use multiple indicators of a variablle
Define the word:
Probability
Odds and the likelihood of a certain event will occur
Why does researchers use probability?
- Indicating how confident they are that the results are not by chance.
- it is expressed as p
- The numerical value that is been used is Alpha
The alpha components are classified as:
p<.05 because of chance and not the intervention
p<.01
p<.001 because of chance and not the intervention VERY SIGNIFICANT
QN: If researchers say their statistical analyses reveled significant results with a p<.01, they are saying:
there is less than 1 chance in 100 the result is because of a chance
This alpha component they are very confident about the statistical, findings
What is Type I error:
- Ho - there is no effect nor difference
- Rejecting Ho when you should have accepted it.
- Keep the Null hypothesis as nothing changed
What type of Type II error:
- Ha - there is an effect and a difference
- If you accept Ho when you should have rejected it.
- Reject null hypothesis as findings did show chnage
What is the process to reject and accept the null hypothesis?
Step #1-
- Determine if null hypothesis true or false
Step #2-
- if null is tru (there is no effect)
—– accept null
Step $3-
- if the null is false ( there is effect)
—— reject null