+ Reply to Thread
Results 1 to 3 of 3

Thread: moderation in SPSS vs. PROCESS

  1. #1
    Points: 5, Level: 1
    Level completed: 9%, Points required for next Level: 45

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    moderation in SPSS vs. PROCESS




    Hi. I did mediation and moderation analysis of one model in 2 programs: SPSS and PROCESS (macro). Mediation was the same, but moderation was different in PROCESS. In SPSS both X and M were significant (p=.001), but in PROCESS significance of X lowered to .0077 and M became insignificant, while significance of the moderator stayed the same (.042) and R^2 is also the same as in SPSS. Why is that? Did I do something wrong? What can it be and how to work with that?

    Thank you a lot.

  2. #2
    Points: 12, Level: 1
    Level completed: 23%, Points required for next Level: 38

    Posts
    4
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: moderation in SPSS vs. PROCESS

    I don't know much, but maybe it's because PROCESS uses bootstrap?

  3. #3
    Points: 358, Level: 7
    Level completed: 16%, Points required for next Level: 42

    Posts
    15
    Thanks
    0
    Thanked 3 Times in 3 Posts

    Re: moderation in SPSS vs. PROCESS


    Hi,

    I used PROCESS quite a bit, maybe I can shed some light on this. My responses are in bold.

    I did mediation and moderation analysis of one model in 2 programs: SPSS and PROCESS (macro). Mediation was the same, but moderation was different in PROCESS. Not unexpected. In SPSS both X and M were significant (p=.001), but in PROCESS significance of X lowered to .0077 and M became insignificant, while significance of the moderator stayed the same (.042) and R^2 is also the same as in SPSS. Why is that? Did I do something wrong? What can it be and how to work with that?

    Your R^2 value is the amount of variance accounted for by the variable (predictor, moderator, etc.). Even though the value slightly changes, the amount of variance accounted for is stable. Depending on your analysis, p=.0077 is not non-significant; remember the p=.05 value. Also, PROCESS tends to favor confidence intervals, which is generally more appropriate, than p-values. p-values tend to be the 'all-or-nothing' thinking of NHST, and do not have anything meaningful contained, whereas confidence intervals do. If your significance value lowered, it does not make it 'more significant' or 'less significant'' it is only significant or non-significant.

    PROCESS uses a slightly different method of analyzing the data; it does use bootstrapping, as someone above mentioned, and it also controls for some assumptions that other methods tend to violate. If you did moderation/mediation the old fashioned way (a la Baron and Kenny 1986), this would contribute to your different numbers. Keep in mind that you can have a significant predictor predicting the outcome, and a significant moderator predicting the outcome, and not have significant moderation. Moderation tests the effect of the moderator on the relationship between your predictor and your outcome; I tend to think of moderation as controlling for the 'third variable problem'.

    I would be inclined to say report the results from PROCESS, as it tends to be more accurate. Run the analysis, making sure you have things set up appropriately, and analyze everything.

+ Reply to Thread

           




Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats