1

STEP 1: Taking Prevention focus as Independent variable and Perceived behavioural control as Mediating variable The relationship between IV (Prevention Focus) & MV (Perceived behavioural control) is significant (0.010) Beta -0.173

STEP 2: Taking Prevention focus as Independent variable and Entrepreneurial intention as dependent variableThe relationship between IV (Prevention focus) & DV (Entrepreneurial intention) is significant (0.03) but with negative Beta value. -0.146

STEP 3: Introducing Mediating variable in direct effect model When MV ( Perceived behavioural control is introduced between IV (Prevention Focus) & DV (Entrepreneurial intention) the relationship between IV & DV remains negative and becomes insignificant.However, now the beta value of Perceived behavioural control is positive and significant which I interpreted as full mediation enter image description here

STEP 4: Relation between Mediating variable BC and Dependent Variable Entrepreneurial intention. enter image description here I am following Baron & Kenny approach to find mediation.

How do I interpret this ?

Rahul Yadav
  • 11
  • 4
  • 2
    What do you mean *"When MV is introduced between IV & DV"* - a regression model knows nothing about "in between". What exactly are the models you fitted in each step ? – Robert Long Jan 30 '19 at 13:51
  • Hi Robert I have posted screen shots of the results so as to make my questions clear. – Rahul Yadav Jan 30 '19 at 15:01
  • Rather than just post SPSS output, it would be better if you describe what the variables are, in each model. I don't like the Baron/Kenny approach - I would much prefer a path analysis approach. [Here](https://stats.stackexchange.com/questions/112575/is-baron-and-kenny-method-for-mediation-now-outdated) is some info about B/K and it's limitations. – Robert Long Jan 30 '19 at 17:21
  • IV is Prevention Focus , DV is Entrepreneurial intention , Mediator is Perceived behavioural control. – Rahul Yadav Jan 31 '19 at 06:32
  • The Baron-Kenny approach requires a regression of the dependent variable on the presumed mediator. You don't seem to have done that. It us unclear what you are asking. – Robert Long Jan 31 '19 at 10:03
  • I have edited the post.Read: Alt in the output as Entrepreneurial intention. – Rahul Yadav Jan 31 '19 at 10:41
  • Note that, these days, there are better approaches to mediation analysis than B and K. – Peter Flom Jan 31 '19 at 10:54
  • @RobertLong I've done the regression of dependent variable on presumed mediator. Please,relook . – Rahul Yadav Jan 31 '19 at 11:05
  • @PeterFlom I've used Hayes and Preacher, Process 3 plugin in SPSS. However,I am confused with the interpretation. Please, share if you have a document on how mediation is analysed in H & P if one of the interval is negative. – Rahul Yadav Jan 31 '19 at 11:09
  • Look at the link that @RobertLong provided – Peter Flom Jan 31 '19 at 11:41
  • So what is the problem ? I don't see any change in sign of any coefficient between these regressions (and even if it did, it should not be a problem). It appears that you have found evidence that the effect is indeed mediated. – Robert Long Jan 31 '19 at 13:11
  • Problem is how can a negative direct relationship become positive just because of a mediator ? In terms of moderation I can understand that under XYZ context the relationship is positive. The theory says the direct relationship is positive I be got it to be negative but positive when mediated. – Rahul Yadav Jan 31 '19 at 15:03
  • The coefficient for PVF is negative throughout, and the coefficient for BC is positive throughout. Nothing has changed sign as far as I can see (but it woundn't surprise me if it did, because that is exactly what *can* happen when you add a mediator into a regression) – Robert Long Jan 31 '19 at 16:34

0 Answers0