Hi I am conducting research with two independent variables and one mediating variable. 1. I am confused on how to per test of linearity on my data.Should I perform separate tests for each Hypothesis? 2. And wat If my linearity and deviation from linearity are both have significance less than 5%? 3. Also, I want to know which test should I use fro actual analysis and How (multiple regression, Sobel, Bootstraping)? Thank You so much in advance and please reply soon. I have a deadline approaching.

0 |

12/23/2016 3:25:14 AM

On Thu, 22 Dec 2016 19:25:14 -0800 (PST), sara javaid <sarajavaid19@gmail.com> wrote: >Hi I am conducting research with two independent variables and one mediating variable. >1. I am confused on how to per test of linearity on my data.Should I perform separate tests for each Hypothesis? >2. And wat If my linearity and deviation from linearity are both have significance less than 5%? >3. Also, I want to know which test should I use fro actual analysis and How (multiple regression, Sobel, Bootstraping)? >Thank You so much in advance and please reply soon. I have a deadline approaching. Why are you doing the tests? What are you interested? It is usually helpful when such questions are answered with a description of the actual variables, design, and purpose. I will leave out further mention of your "mediating variable" since that could mean so very many different things. Main effects: I can say that if you are interested in showing a predicted linear trend of a composite of two variables, you will have far better power for whatever analyses you do, if you compute a /single/ composite variable that represents the two. And a test for a linear trend will have more power (when a linear trend is what exists) than a test with several d.f. across the several within-subject or between-subject means. So: testing for "linear" is often preferable to testing separate means; testing one composite is often preferable to testing two separate variables when they represent, basically, the same hypothesis (such as, "improvement"). What is the N? If you have huge N, you may have little inhibition against doing multiple tests -- unless you want to draw concise conclusions. For instance, it might be helpful to know that the two IVs are largely redundant, and neither one adds very much to the other. Or not. With a huge N, you may see a small, separate effect elevated to "significance" by the sample size. Similarly, you may have a main effect that is 95% linear across groups (or times), but a huge N can elevate that remaining, non-linear component to "significant". -- Rich Ulrich

0 |

12/23/2016 5:56:08 PM

My independent variables are Social media marketing and electronic word of = mouth, mediating variable is brand knowledge and dependent variable is impu= lse buying behaviour. My assumption is that social media marketing and EWOM= leads to brand knowledge which leads to impulse buying. I am using SPSS 20

0 |

12/25/2016 12:25:48 AM

And my n = 210

0 |

12/25/2016 12:26:22 AM