Test #2 Flashcards
Surveys and questionnaires are most commonly used in which type of method?
Quantitative
What are surveys/questionnaires used for?
Obtaining information about what people do, and respondent’s attitudes or characteristics.
What are surveys?
Conversations between the researcher and respondent. (One way communication)
Ways to collect survey info:
Paper/pencil Face-to-face Phone Mail Computer assisted
What are the benefits of computer assisted surveys?
Less time
Cost effective
Wide reaching
Reduces human error
Where should demographics (sex, age, income) be on a survey?
At the end
Respondents use their own words to respond.
Open questions
What must the researcher do after all data are collected?
Code responses
Respondents given a question or statement and given a set of responses to select from.
Closed questions
What kind of scale is this?
Please indicate your attitude toward blank
Highly undesirable—————-highly desirable
Graphic rating scale
What is a benefit of the graphic rating scale?
Sensitive system that required measuring actual physical distance on the line for dating coding.
What is a disadvantage of the graphic rating scale?
Significant amount of time and labor.
When is a graphic rating scale practical and useful?
During online surveys
What kind of scale is this?
Please indicate your attitude toward…
(1). (2). (3). (4). (5).
Highly undesirable. Highly desirable
Itemized rating scale
Benefits of itemized rating scale
Easier to respond and code data
More practical than a graphic scale
Disadvantages of an itemized rating scale?
Lack of sensitivity
What kind of scale is this?
Highly undesirable Desirable Highly desirable
Combination of graphic and itemized
Question contains several issues in one question but only provides one set of responses.
Ex. Do you think there is too much sex and violence in today’s media?
Double barreled questions
Questions presenting only one aspect of an issue on which respondent’s reactions are being sought.
Ex. Do you think the mad media are negatively influencing individuals mental health?
(Yes). (No)
One sided question
What is the problem with one sided questions?
People tend to agree with whichever side is presented.
Steer respondents toward a certain answer.
Ex. Don’t you think driving a SUV is harmful to the environment?
Leading questions.
A 7 point scale is always better than a five point scale. True or false?
True
Usually consists of multiple items and 5 choice categories in each question. (Multi dimensional variable)
Likert scale
Also called cross sectional or non experimental studies. Common, quick, and easy form of research.
Descriptive design
What can descriptive design not determine?
Causal relation. (No manipulation of the IV)
There is no what in descriptive design?
Condition
What are surveys useful/not useful for?
Useful for detecting differences/relationships. Not useful for finding cause and effect
Used to determine causation. (Conducted in a lab or controlled setting)
Experimental designs
Why are experimental designs set in a lab?
To control extraneous influences.
What are extraneous influences also known as?
Confounding variables (weakens the relationship)
The independent variable is manipulated by the researcher during…
Experimental designs
3 Necessary conditions of causality
- Temporal ordering
- Evidence of association
- Control for other variables
Cause variable must occur before the effect variable
Temporal ordering
There must be a relationship between cause (IV) and effect (DV). Must be a covariance
Evidence of association
The causal variable must be manipulated while others are being controlled for
Control for other variables
Experiment done in artificial setting. High internal validity, low external validity
Lab experiment
Ability to determine whether the observer result (DV) is due solely to the manipulation (IV)
Internal validity
ability to generalize the research findings to the real world.
External validity
Done in the “real world” instead of a lab setting. High external validity, low internal validity
Field experiment
What can feed experiments be interfered by in Communication research?
Unexpected situations
Field experiments are popular in what?
Marketing
Specific external events that occur during experiment that can affect the dependent variable. (Ex. Temperature)
History effect
Effect of psychological/physiological changes among the participants during the course of the experiment.
Ex. Tiredness
Maturation effect
The result of post-manipulation test is being affected by the pretest conducted previously.
Pretesting effect
Effects of inconsistent or inaccurate instruments used between pre testing and post testing.
Instrument variation effect
Effects of selecting non equivalent samples between control and experimental.
Selection bias effect
What are the 5 threats to internal validity?
- History effect
- Maturation effect
- Pretesting effect
- Instrument variation effect
- Selection bias effect
What are the 3 threats to external validity?
- Reactive bias
- Pretest (manipulation interaction bias)
- Jon representative sampling bias
Effects of participants reacting to the experimental environment, causing them to exhibit abnormal behavior.
Reactive bias
Effects participants becoming more or less sensitive to the manipulation variable due to pretesting measurement.
Pretest (manipulation interaction bias)
Effects of selecting participant samples that do not represent the targeted population.
Nonrepresentative sampling bias
- Little or no control over confounding variables.
- Participants are not randomly assigned to a condition
- Still, often used as a basis of causal inferences because of practical reasons
Quasi-experimental design
- Tighter control over validity issues
- Utilize experimental and control groups
- Utilized random assignments to create equivalent groups.
- Sometimes matching, or matching random assignment can be used to create equivalent groups.
True (classical) experimental design
What is the most complete and true experimental design?
Solomon’s 4 group design
Solomons 4 group design has
No inherent threat to validity at all
What are the pitfalls of solomons 4 group design?
Cost/time inefficient
Almost impossible to implement in the real world.
Treatment of groups based on two or more levels of independent variables. (More than one cause)
Factorial experimental design
When is the dependent variable measured in the factorial experimental design?
Once after treatment is given.
What can factorial experimental design test for?
Main and interaction effects
Multiple measurements of the dependent variable across time. (Long-term effects)
Longitudinal experimental design
What is the threat to longitudinal experimental design?
Losing participants over time which can make the conclusion weak
What do numbers collect?
Data, phenomenon, quality, intensity, value, or degree
Specifies how data are collected and become numerical.
Operationalizations
A theoretical distribution of scores. Also know as the bell curve.
Normal curve
Normal curves are what?
Symmetrical
Mean median and mode are identical
Normal curve
Majority of cases distributed around the peak in the middle in what curve?
Normal curve
Curve is asymmetrical
Skewed distributions
Very few high scores
Positively skewed
Very few low scores
Negatively skewed
When mean is the lowest…
Negatively skewed pattern. Caused by low outliers
When mean score is the largest…
positively skewed pattern. Caused by high outliers
Central tendency is what?
Mean, median, more
How much subjects differ from group mean
Dispersion or standard deviation.
Number of cases are represented by what?
N
What are the 3 types of descriptive stats?
- Number of cases
- Central tendency
- Dispersion
Dispersion describes what?
The variability or spread of scores.
If sd equals zero…
All scores are the same
The larger the sd…
The more scores differ from the mean
Standard deviation estimates what?
Sampling error
Sd determines the?
Set range
For the data to be considered a normal distribution, it must be?
- Identical mean, median, and life
2. Data must be within set range
Theoretical normal curve should be divided into?
Equal standards
If the distribution is perfectly normal, what percent of the data should fall within the +1 to -1 range?
68.26%
Standard deviation must be?
Positive
What does it mean if the distribution is not perfectly normal?
There is sampling error
Often used to describe characteristics or attributes of participants
Percentages
Reveal whether the observed differences that might occur by chance
Statistical tests of difference
What are the statistical tests of difference?
Chi square
T test
ANOVA
Statistical test used to evaluate hypothesis and research questions
Inferential statistics
If results are statistically significant then?
Results are assumed to hold true for the population.
What do inferential stats test?
The likelihood that the alternative hypothesis is true and the null isn’t.
What is generally considered statistically significant ?
.05
If p< or equal to 0.5…
Alternative hypothesis is accepted
If p> .05…
Null hypothesis remains
Specifies how many values vary within a statistical test.
Degrees of freedom (df)
Compares the observed frequency with the expected (hypothetical value when everything is equal) frequency
Chi square
Degree must be what for both IV and DV in chi square?
Nominal or categorical
Total sample size divided by number of categories.
Expected frequency
What is the df in chi square?
Total number of categories minus 1
What must the measured value be, to be significant?
Greater than or equal to the critical value
Determines if differences in how cases are distributed across categories of one nominal variable are significant.
One dimensional chi square
What does significant chi square indicate?
Variation of frequency across categories did not occur by chance
What are the limitations of chi square?
- Only use nominal data variables
- Tests may not be accurate if observed frequency is zero or less than 5.
- Cannot directly determine causal relationships.
Represented by t
T test
Determines if differences between two groups of the independent variable on the dependent variable are significant.
T test
In the t test, the IV must be
Nominal data of two categories
In the t test, the DV must be
continuous level data at interval or ratio level
Compares mean scores from IV for two groups of people.
Ex. Difference between males and females on aggression after playing violent video games
Independent sample t test
Compares mean scores of paired or matched scores from IV from the same participants.
Ex. Difference in aggression between before and after playing video games from males
Paired comparison t test
Hypothesis or research question indicates that a difference in either direction is acceptable
Two tailed or non directional t test
Hypothesis or research question that specifies the difference to be found.
One tailed or directional t test
What are the limitations of t test?
- Limited to differences of two groupings of one IV on a DV
2. Cannot examine complex communication phenomenon
Compares the influence of more than two groupings of IV on the DV. Represented by f
Analysis of Variance (ANOVA)
In ANOVA, the IV must be
Nominal
In ANOVA, the DV must be
Continuous level data
What types of variances does ANOVA look at?
Between group and within group
Differences between groupings of IV are large enough to distinguish themselves from one another.
Between group variances
Variation among individuals within any category or grouping. (About average)
Within group variance
The better the categories of the IV explain variation in the DV, the larger the…
F
Tests for significant differences in the DV based on categorical differences of one IV
One way ANOVA
Difference between groups is larger than difference within groups.
Significant f
Determines relative contributions of each IV to the distribution of the DV
Two way ANOVA
What can two way ANOVA determine?
The main effect of each IV and the interaction effect
If there is a simultaneous influence on both IVs
Interaction effect
If interaction effect exists, then the main effect…
Is ignored
What are the limitations of ANOVA?
Restricted to testing IV of nominal or categorical data, and can be difficult to interpret when 3 or more IVS are used.
2 continuous level variables
Correlation
Identified statistically significant linear patterns in the association of variables.
Tests of linear relationships
In inferential stats, both variables must be what for correlation?
Interval or ratio
If r value is closer to zero, the relationship is…
Weaker
What can r value determine?
The strength of the relationship
R = 0 means that…
There is no relationship
Correlation reveals one of the following:
- Scores on both variables increase or decrease together
- Scores on one variable increase (or decrease) scores on the other variable decrease (or increase)
- No pattern or relationship.
What does r reveal?
The degree to which two continuous level variables are related.
It r is found with p that is less than or equal to o.5, then the relationship is
Significant
Both variables increase or both decrease
Positive correlation
One variable increases while the other decreases
Negative correlation
Limitations of correlation
- Can only examine the relationship between two variables at a time.
- Any relationship is presumed to be linear.
- Limited in the degree to which inferences can be made.
