APA, Ch. 5, 7, 8, 9 Flashcards
A sample from a population is called?
Sampling
Parts of a research paper
Structure-content-citation rules
The ultimate goal of a sample is to?
Generalize (external validity + represent)
Is the accuracy with which the results of an investigation maybe generalized to a different group from that one study
External validity
When an investigator is interested in studying a group of people with particular characteristics of interest, that group is known as a
Population
We might instead select a subset of the population or universe thought to represent the entire group, a subset known as a
Sample
Is the degree to which the samples parameters DIFFER from the parameters of the population from which it was selected
Sampling error
There are two sampling methods
Probability sampling and nonprobability sampling
Is it generally most preferred by researchers. It involves the selection of elements from a population or universe in accordance with some set of mathematical rules, thereby permitting calculation of the probability of sampling error.
Probability sampling
Is the most elementary form of probability sampling. Each element in the population or universe is afforded an equal opportunity of being selected to the sample.
SRS
Simple random sampling
The second variety of probability sampling, like simple random sampling, requires a complete sampling frame, from which every element is selected following a random start
Systematic sampling
Like the previous two techniques, ______ requires the generation of a complete sampling frame. It’s particular advantage, however, is that it permits the researchers some assurance that elements with particular characteristics are included in the sample.
Organizing the elements in the sampling frame into subsets based on some characteristics of interest, or using one of the previous two techniques to select a proportional representation from each subset to the sample.
Stratified sampling
Is a probability sampling technique that is particularly useful when dealing with a very large target population or universe when it would be inconvenient or impossible to generate a complete sampling frame of elements.
The choices of elements are continuously narrowed until a complete sampling frame becomes possible, then the final elements are chosen from the sampling frame in accordance with one of the previous three sampling techniques
MCS
Multistage cluster sampling
While most researchers prefer probability sampling techniques, there are numerous occasions went non-probability must be used
Nonprobability sampling
How can we improve sampling?
We can replicate (different place, different people, different time)
We can use theory or logic to support the claim
Based on mathematical rules
Probability sampling
Uses some form of random selection-requires a complete frame.
Probability sampling
n = sample size,
Systematic sampling
Uses proportional reduction Tatian’s of a certain valuable(Gender, ethnicity, or age)
Males = 60%, females = 40%
Stratified sampling
Separate the population into mutually exclusive sets (strata)
Example = sex-male •female • draw random samples from each stratum by using one of the previous two techniques
Stratified sampling
Useful for a very large target population-when it seems impossible to generate a complete sampling form
Multistage cluster sampling
MCS
Not based on probability (no mathematical rules, not random
Nonprobability sampling
Availability sampling, relies on a available sample
Convenience sampling
Judgmental sampling, selecting sample based on specific characteristics of interest to the researcher.
Example = topic-combination effectiveness in the successful business.
IBM or Microsoft because of success
Purposive sampling
Selecting sample according to some quotas-but not randomly. Represents major characteristics of population (ethnicity, gender) by sampling proportional amount of each.
Quota sampling
Network sampling. One person recommends another, who recommends another, who recommends another. We use when = hard-to-reach populations.
Snowball sampling
Less effort, less time, less resources.
Nonprobability sampling
Limitations? Possible to misrepresent population. Cannot estimate the sampling error, which may cause potential problems in generalizing.
Nonprobability sampling limitations
The gap or difference between the nature of the population
Big circle = population small circle = with in population-sample
Really big population, really small sample = big gap!
If people in population are similar to each other-possible to select any element that represents the population.
Homogeneity
If people are dissimilar - samples must increase in size to reduce the likelyhood of error
Heterogenous
The variables expected to influence a change in another variable
Independent variable
Those expected to change as a result of the actions of the independent variables
Dependent variables
All other variables that might somehow influence the relationship between the independent and Dependant variables, those extraneous to the relationship, are called
intervening variables
While another group receives imposed treatment, and is referred to as the
Control group
In many cases one or more of the groups receives some level of the independent variable, or some treatment, and is referred to as the
Treatment group
We have two groups, each receiving some form of treatment, whether it be lecture or discussion, and those groups are compared with one another, and are therefore known as
Comparison groups
Prior to the imposition of the independent variable, all the groups were equivalent with regard of the dependent variable. This assumption is referred to as
Group equivalence
Participants selected for an investigation are assigned to a treatment, control, or comparison group based on some randomizing technique.
This randomizing technique can be used for the lottery, use of a set of random numbers, or a systematic sampling technique.
Random assignment
Established treatment, control, or comparison groups are evaluated on the dependent variable prior to the introduction of any treatment.
Pretesting
Participants in the treatment, control, or comparison groups are matched on characteristics thought to be important to the D pendant variable.
Matching
All participants in all groups are kept uniform with regard to significant characteristics thought to influence the dependent variable
Constancy matching
Each participant in the treatment, control, or comparison group is matched with participants in the other groups, based on characteristics thought to influence the dependent variable
Pairing
If the research setting has been created by the investigator, who maintains complete control over all that occurs in that setting, the study is known as a
Laboratory experiment
The other setting is the naturally occurring research setting, where the investigator has no opportunity to shape the setting to his or her preference
Field experiment
They examine only one independent variable at a time
Single-factor studies
Many times, however, researchers choose to examine the influence of two or more independent variables, or factors, as they simultaneously influence a dependent variable.
Factorial studies
3×2×3×2
36 cells
Would most likely be used by our instructor in the sample scenario. In this design, a separate group of participants, or subjects, would be used in each of the cells so that comparison could be made between cells.
Between-subject design
In some situations, the liability of the between-subject design can be alleviated by using the same subjects in each of the cells in the design diagram.
With in-subject design
In those situations in which that within-subject design is impractical, but the number of subjects required for a between-subject design proves unrealistic, the researcher may combine the two approaches into a mixed factorial design. In this design, the same subjects are used across the levels of one or more independent variables, while a difference that is used across the levels of the remaining independent variables
Mixed factorial design
Intervening variable
Bad, you control
2×2×3 =
12
2 -independent variable, 2 -independent variable, 3 -independent variable.
2 - levels, 2 - levels, 3 - levels.
Method that can even evaluate the casual relationship between the independent variable and the dependent variable while controlling for other intervening variables
Experiments
A simple relationship between independent variable and Dependant variable
Correlation
Goes one way
Casual relationship
Temporal ordering-causes independent proceeds in effect Dependant in time.
Meaningful correlation-theory.
No alternative causes-no other explanation for the causes.
Three requirements of causality
Alternative cause, other valuables that might influence or interfere the relationship between the independent and the Dependant variable
Intervening variables
Most of the independent variables in experiments are manipulated by researchers
Manipulation
Receive some level of treatment on independent variable
Treatment group
Receives no treatment on independent variable
Control group
Must ensure that groups are equivalent with regard to the Dependant variables before the treatment.
Group equivalence
How can we check group equivalence?
Random assignment-it usually works. Can use statistical testing to doublecheck.
Pretesting-pretest (8), posttest (9), Treatment (post-pre-) versus control (post-pre)
Participants in groups are matched on characteristics that are important to dependent variables
Matching
Cannot manipulate independent variable
Comparison group
Each participant in a group is mashed with another participant in another group. Gender gets balanced
Paring
Two or more independent variables in the same argument
Factorial studies
Different participants in each cell. Comparison between the cells
Between subject design
Everyone goes through all, repeated measures
Within subject design
Combination of the between and with in design
Mixed factorial design
Surveys use self-report technique
Survey methods
Advantages of survey methods
Access to subjective info. Access to broadly distributed population (email)
Disadvantages of survey methods
Requires respondents to recall (people have to think). What participants report and actually do, could be different
Survey designs
Cross-sectional study Longitudinal study Trend study Panel study. CLTP
One time data collection. (Once and done) response from a single point in time. Example = teacher semester evaluation’s.
Cross-sectional study
Asking some questions across a period of time
Longitudinal study
DIFFERENT samples from a population at different time points.
Trend study
Circle = population, little circle with in population (May, June, April)
The SAME sample at different time points
Panel study
Circle = population, little circle within population (May, April, June, July)
Panel study is highly vulnerable to several threats
Attrition (drop out)
Test sensitization
History
Maturation (tired, changing mind)
Major sections of designing survey instrument
Introduction
Instructions
Questionnaires
At the end…
Brief introduction of the study. (Researcher, purpose, etc.) participants rights-voluntary, right of withdrawal, ambiguous, confidentiality. Time required (10 to 15 minutes of your time)
Introduction
Complete and concise set of instructions how to select items-only one? Multiple? Rank order?
Instructions
Basic guideline = be clear, simple, understandable language, ninth grade level, be concise (simple to the point) lengthy, participants may skip them, be realistic
Questionnaires
Add filter questions if necessary (do you use Twitter?)
Questionnaires
Demographic information, write a thank you!
At the end
How many hours of TV did you watch last year? = Bad, avoid bias wording-‘‘do you read newspapers or just watch TV?’’ (Take out word JUST), avoid leading questions-don’t you like our product? = Bad, avoid double barreled questions = asking a single question that ask for more than one response.
Things to avoid when designing surveys
Accurate findings about the phenomena under investigation for the particular groups of people studied.
Internal validity
Events which occurred during the study, influence participants behavior with in the study. Changes in the environment.
History
And initial measurement in a research study influences the subsequent measurement. Pretest affects posttest.
Test sensitization
Instruments wearout (out of date)
Instrument Decay
Subjects change their behavior because they know that they are being observed.
Hawthorne effect
Participants are being self-selected because people are self-selected, the study may not be good to represent the whole population. Could occur during the recruitment process.
Self-selection Bias
Natural changes that occur within participants over the course of the study. Tired, sick, bad day, sad, happy. Physiological/psychology.
Maturation
Dropping out from the study. Lost interest, forgot, do not care.
Attrition/mortality
Nonhuman elements. YouTube, blogs, websites.
Data
Participants influence each other. Do not speak about what is going on in the study until it’s over.
Inter-participant bias
Problems with researchers methodology
Personal attribute effect, research bias. PA, RB
Personal attribute effect, research bias. PA, RB
Problems with researchers methodology
Researchers characteristics influence participants behaviors. People may not be honest with you, your personality, outfit, ethnicity, gender.
Personal attribute effect
Accidentally informs the participants of what he/she expects. Do not say anything about your data.
Researcher bias
Type your pre-a validated questions and instructions.
Script
Hire a researcher (or assistant) who can conduct the study. The hired person doesn’t need to know the hypothesis/research questions.
Double-blind study
Relationship between sample and population. Population = college students, sample = group of people that represent population.
External validity
The results of the study can be “generalizable” to population.
Good external validity
Representation
Generalizable
Generalizable
Representation
