Upgrade to remove ads
Research Methods of Mediated Comm.
Terms in this set (16)
What are the advantages of multiple methods?
- goals are explicit
- the whole > the sum of its parts
- findings appeal to a wider audience/community
What are the disadvantages of multiple methods?
- more costly
- more time
- requires greater expertise
- requires pluralism
What dictates the design of multiple methods?
Requirement, characteristics, and questions dictate the design.
Quantitative --> Qualitative
- theory/prior research spurs the study
- specific focus = "traditional science"
- need generalization
- contrary findings = data pertaining only to hypothesis -> stone-walled
- qualitative phase = way to explain anomalies
Qualitative --> Quantitative
Qualitative: - when we "know" something -> rigorous hypothesis testing (experimental or survey design)
Quantitative: - no existing theory/research
- way to reduce 'noise'
- discovery not testing hypothesis
Expected vs. observed outcomes
- the nothing hypothesis meaning that the things being compared have no difference, they are the same.
3 versions of null hypothesis
1. the observed difference was created by sampling error (note that the term sampling error refers only to random errors, not errors created by bias)
2. There is no true difference between the two groups (the term true difference refers to the difference we would find in a census of the two populations. That is, the true difference is the difference we would find if there were no sampling error.
3. The true difference between the two groups is zero. Probability- tells us how likely that the null hypothesis is true.
there is a real difference. It is what we wish to confirm.
Yoga- null hypothesis: yoga and non-yoga group will have the same average test anxiety.
Alternative hypothesis - yoga group will have lower average test anxiety than will non-yoga group
- a test to evaluate the hypothesis
- measures difference between what is observed in data and what is expected.
- a large p-value means the value of test statistic supports null hypothesis. Observed difference due to chance.
- a small p-value means the value of test statistic suggests that this difference is not due to chance. There is a real difference or a real relationship.
Rule of thumb: if p-value (alpha) is less than .05, we conclude there is a real difference or a real relationship/association.
- In other words: difference = statistically significant
Type 1 error
= False positive (large sample) - we tend to find relationships by chance but they are very weak. Benchmark put at less than 99% instead of 95%. Error of illusion. Rejecting the null hypothesis when it is true (declaring a difference when there is no difference)
Type 2 error
= False positive (small sample) - more difficult to find relationships. Failing to reject the null hypothesis when it is true. (failing to declare a difference when there is one.)
Too much chance?
- we sometimes need to use p<.01 instead of p<.05 based on sample size. Examples: Medical research, drug trials, in this case may want to decrease p<.01 so only 1% chance we've made an error.
- a coincidental statistical correlation between two variables, shown to be caused by some other 'third' variable
- when something looks like it's there. Ex: tea drinking linked to cancer, but tea drinkers tend to be smokers that was the actual link.
- weak relationship or small difference can be statistically significant
Examples of Multi-Methods:
1. Conversion analysis
3. parallel studies
Sets with similar terms
Hypothesis Testing and P Values
PSYC 201 Exam 3
Stats 26: The Tests of Significance
Other sets by this creator
CBS exam - beer styles
Marketing 370 - tyaggi
Span 201 writing 2 vocal
Other Quizlet sets
ACC MODULES 11-16 QUIZZES
Agronomy Test 2
Ischemic Heart Disease and Anti-Anginal Agents