Online calculation for the Sobel Test (see http://www.people.

Calculation for the Sobel test: An interactive calculation tool for mediation tests Kristopher J. Preacher (University of Kansas)

Geoffrey J. Leonardelli (University of Toronto)

The Sobel test works well only in large samples. We recommend using this test only if the user has no access to raw data. I assumptions. Consult Preacher and Hayes (2004, 2008) for details and easy-to-use macros that run the necessary regressio

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and compar Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simp
See also:

SPSS and SAS macros for bootstrapping mediation effects SPSS and SAS macros for bootstrapping multiple mediation effects SPSS macro for bootstrapping moderated mediation effects
Mediation effects

A variable may be considered a mediator to the extent to which it carries the influence of a given independent variable (IV) significantly affects the mediator, (2) the IV significantly affects the DV in the absence of the mediator, (3) the mediator ha the mediator to the model. These criteria can be used to informally judge whether or not mediation is occurring, but MacKin by which mediation may be formally assessed.

An illustration of mediation a, b, and c' are path coefficients. Values in parentheses are standard errors of those path coefficients. Description of numbers needed a = raw (unstandardized) regression coefficient for the association between IV and mediator. sa = standard error of a. b = raw coefficient for the association between the mediator and the DV (when the IV is also a predictor of the DV). sb = standard error of b. To get numbers 1. Run a regression analysis with the IV predicting the mediator. This will give a and sa.

2. Run a regression analysis with the IV and mediator predicting the DV. This will give b and sb. Note that sa and sb sho Sobel test equation z-value = a*b/SQRT(b2*sa2 + a2*sb2) Aroian test equation z-value = a*b/SQRT(b2*sa2 + a2*sb2 + sa2*sb2) Goodman test equation

z-value = a*b/SQRT(b2*sa2 + a2*sb2 - sa2*sb2)

The Sobel test equation omits the third term of the variance estimate in the denominator. We recommend using the Aroian assumption that the product of sa and sb is vanishingly small. The Goodman version of the test subtracts the third term for a effect of yielding a negative variance estimate. References

The Sobel test and the Aroian test seemed to perform best in a Monte Carlo study (MacKinnon, Warsi, & Dwyer, 1995), and

Aroian, L. A. (1944/1947). The probability function of the product of two normally distributed variables. Annals of Mathema

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Concept Goodman, L. A. (1960). On the exact variance of products. Journal of the American Statistical Association, 55, 708-713.

Hoyle, R. H., & Kenny, D. A. (1999). Sample size, reliability, and tests of statistical mediation. In R. Hoyle (Ed.) Statistical S

Krull, J. L., & MacKinnon, D. P. (1999). Multilevel mediation modeling in group-based intervention studies. Evaluation Revie

MacKinnon, D. P., & Dwyer, J. H. (1993). Estimating mediated effects in prevention studies. Evaluation Review, 17, 144-15

MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test med

MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behaviora

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simp

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommen Acknowledgments

Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociologic

Original version posted March, 2001. We wish to thank David MacKinnon and David Kenny for advice which made this intera

www.people.ku.edu/~preacher/sobel/sobel.htm)

s no access to raw data. If you have the raw data, bootstrapping offers a much better alternative that imposes no distributional n the necessary regression analyses for you:

ssessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879-891. direct effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 717-731.

ndependent variable (IV) to a given dependent variable (DV). Generally speaking, mediation can be said to occur when (1) the IV ator, (3) the mediator has a significant unique effect on the DV, and (4) the effect of the IV on the DV shrinks upon the addition of n is occurring, but MacKinnon & Dwyer (1993) and MacKinnon, Warsi, & Dwyer (1995) have popularized statistically based methods

edictor of the DV).

. Note that sa and sb should never be negative.

mmend using the Aroian version of the Sobel test suggested in Baron and Kenny (1986) because it does not make the unnecessary tracts the third term for an unbiased estimate of the variance of the mediated effect, but this can sometimes have the unfortunate

arsi, & Dwyer, 1995), and converge closely with sample sizes greater than 50 or so.

ables. Annals of Mathematical Statistics, 18, 265-271.

ogical research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

ociation, 55, 708-713.

R. Hoyle (Ed.) Statistical Strategies for Small Sample Research. Thousand Oaks, CA: Sage Publications.

studies. Evaluation Review, 23, 418-444.

ation Review, 17, 144-158.

on of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83-104.

es. Multivariate Behavioral Research, 30, 41-62.

direct effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 717-731.

rocedures and recommendations. Psychological Methods, 7, 422-445.

Leinhart (Ed.), Sociological methodology 1982 (pp.290-312). San Francisco: Jossey-Bass.

ce which made this interactive web page possible. Free JavaScripts provided by The JavaScript Source and John C. Pezzullo.

s no distributional

ethods, 40, 879-891. mputers, 36, 717-731.

cur when (1) the IV upon the addition of tically based methods

make the unnecessary have the unfortunate

ogy, 51, 1173-1182.

mputers, 36, 717-731.

hn C. Pezzullo.

On this page, we link to a number of web-based, T-value --> P-value conversion c Please remember that the degrees of freedom (DoF) is simply the number of boot So if you resample your bootstrap 200 times, the degrees of freedom is 199 (rega Also, we do not warranty (or guarantee) that any of the results computed by these #1: This first calculator tells you the one-tailed and two-tailed probability values of Please supply the necessary parameters, and then click the 'calculate' button:

#2: This second calculator is much more versatile than the first. It calculates a num chi-square to P, t to P, F to P, Fisher r-to-z transformation, and a couple of others

value conversion calculators, both one-tailed and two-tailed. he number of bootstrap resamples minus 1. edom is 199 (regardless of your original sample size). computed by these websites are correct (although it is easy to check!)

obability values of a t-test, given the t-value and the degrees of freedom. lculate' button: It calculates a number of statistical conversions, including z to P, a couple of others:

Cohen, J. (1992). A power primer Psychological Bulletin, 112, 155-

1) Select the partial regression m with the largest number of predictive latent constructs ( indicators) in 'Mult R' (3 in this example).

2) Select the desired (alpha) le < 0.05 in this example).

3) Select the desired effect size (medium in this example).

4) Find the N in the correspondin and column (76 in this example)

5) These heuristics indicate that need a minimum sample size to have adequate power for your purposes.

6) NOTE: This minimum sample s = 76) is a reasonable starting to achieve your desired power.

A power primer. etin, 112, 155-159.

al regression model number of t constructs (or R' (3 in this

ed (alpha) level (p mple).

ed effect size xample).

e corresponding row this example).

s indicate that you m sample size of 76 power for your

mum sample size (N able starting point sired power.