Quantitative research (analyzing bivariate) Flashcards
What is bivariate analysis?
test the relationship between variables. Is there a link that exists or not between the variables?
What are the types of differences between responses?
- mathematical: numbers are different
- statistical: statistically significant = difference is large enough to be unlikely to have occurred because of chance or sampling error
- managerial: difference is important if results or numbers are sufficiently different
What are the steps of the bivariate analysis process?
- start form the research problem
- identify the questions you cant to analyze and identify the variables related to each question
- decide to conduct bivariate analysis
- identify the measurement scales for each variable related to the question
- choose the analysis technique (cross-tabulations, comparisons of means, correlation)
- conduct the statistical test (chi-square, T-tes, F-test)
What are the 4 analysis techniques for bivariate analysis?
- cross tabulations -> comparing frequencies between variables
- comparison of means & ANOVA -> comparing averages
- correlation -> strength of a linear relationship
- regression -> dependency between variables
When should you use a chi-square test?
cross tabulation
When should you use a T-tes
comparison of means
when should you use an F-test
correlation or regression
what should be the value of p to reject H0
<=0.05 or 0.025 (T-tes)
If p = 0.07, can you reject H0?
No
What are the steps to perform a statistical test?
- determine the appropriate analysis and test to perform
- identify H0 et Ha
- calculate the statistic and p-value to test Ho
- decision (reject or no)
- calculate the strenght and direction of the relationship (if applicable)
- interpret the result
What should be the null hypothesis for cross tabulations?
there is no difference to X variable when Y is this or that
What should be the null hypothesis for a comparision of means?
the means are the same for both variables
When should you use anova instead of T-test?
when you have a metric varibale + a non metric one with >2 categories
What should be the null hypothesis for correlation testing?
there is no relationship between the 2 variables
What should be the null hypothesis of linear regression?
X has no impact on Y