p-values for t-statistics: how to compute them?

[irene eleonora]

Hello.
I have a problem with calculating the p-values for the t-statistics of my path-coefficients.
I know I should refer to the tabled t-distribution, but what about the degrees of freedom? Shouldn't I know how many degrees of freedom I have in order to compute the critical values for the t-distribution in my case?
And if yes, how can I compute the degreees of freedom in my model?

Thank you very much for your precious help!!!

Irene[/b]

[Diogenes]

Hi,

You could compute p-value from t-values in the Excel.

=TDIST(t value; df; 2)

t value = computed using bootstrap algorithm
df = number of resamples in the bootstrap – 1 (if you use 1000 resamples, df = 999)
2 = two tail.

Best regards,

Bido

[irene eleonora]

Thank you so much for your clear and quick reply!

Best regards,

Irene

[Ennajeh Leila]

when I use excel(LOI.STUDENT(t-value;999;2) to calculate p value I obtain the following results :


0,010790415
1,6154E-75
0,004651571
1,28578E-16
1,6854E-134
0,038633957
2,0726E-280
0
6,0476E-246
0,000241838
1,2518E-198
9,00554E-72
0,001299928
0,011430339
0,404866285
0,05456066
8,4771E-169
2,8862E-289
0,551344293
0
0,093322527
3,3592E-131
0,346166107

is this correct?

then I have another question: my sample size is 205 so t must be >1.96 ? so p>0.05 why in some papers or thesis I found that some coefficients are significant at 1% or 10% ?? should I validate hypothesis that have a t value lower than 1.96??

thank you for your help

[Diogenes]

Hi,

Yes, it is correct.

1,28578E-16 = 1,28578 * 10 ^ (-16) = 0,000000000000000128578
That it means: p < 0.001

The threshold of 1.96 is based on the number of resamples (usually 1000), not in the sample size.

The cutoff of 10%; 5%; 1%; 0.1% depends on the field of research, effect size, consequences of the type I and type II error.

For more information about this issue, see:
COHEN, J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Psychology Press, 1988.
MORRISON, D. E.; HENKEL, R. E. (EDS.). The Significance Test Controversy. New Brunswick, NJ: Aldine Transaction, 1970.
MURPHY, K. R.; MYORS, B.; WOLACH, A. Statistical power analysis: a simple and general model for traditional and modern hypothesis tests. 3rd ed. New York: Routledge – Taylor & Francis Group, 2009.
ZILIAK, S. T.; MCCLOSKEY, D. N. The Cult of Statistical Significance: how the standard error costs us jobs, justice, and lives. Ann Arbor, MI: The University of Michigan Press, 2008.

Best regards,

Bido

[Ennajeh Leila]

Thank you very much :)

[gustavosalati]

In Excel 2010, the TDIST function has been replaced by two functions, the T.DIST.RT function (which calculates the right-tailed Student's T-Distribution), and the T.DIST.2T function (calculates the two-tailed Student's T-Distribution).

In my case (portuguese excel), the formulas are:
TDIST=DISTT
T.DIST.RT=DIST.T.CD
T.DIST.2T=DIST.T.BC

[palizadeh]

Hi all,
I have a big problem with bootstrapping
I run the algorithm but faced with the following result for all of the roads:
Sample mean: n/a
STD: n/a
the column for T statistics and P-value is empty
what is wrong with my model??
I had 27 samples and the results for measurement model is OK

[cringle]

Do you use SmartPLS 3? In SmartPLS 3, use bootstrapping and the BCa option, "basic", no sign changes, and 5,000 subsamples. Then, you would not interpret the bias-corrected confidence intervals.

Best regards
Christian