# Trouble using Dref() in gnm()

variables
#outcome: R_SDNN (continuous). origin: m107_f. destination:y19_f
#m107: quartiles of socioeconomic status (SES) at 107 months. 1-4: low to high.
#m107_f is the categorical version of m107. values: Low, MidLow, MidHigh, High
#y19 and y19_f are quartiles of SES at 19 years, with the same meaning of the values as m107 and m107_f

output:

drm_SDNN = gnm(R_SDNN ~ Dref(m107_f, y19_f), family=gaussian, data = Sey518)
Initialising
Running main iterations......
Done
DrefWeights(drm_SDNN)
Refitting with parameters of first Dref weight constrained to zero
\$m107_f
weight se
0.4338944 0.2086603

\$y19_f
weight se
0.5661056 0.2086603

drm_SDNN

Call:
gnm(formula = R_SDNN ~ Dref(m107_f, y19_f), family = gaussian,
data = Sey518)

Coefficients:
(Intercept) Dref(m107_f, y19_f)delta1
161.84178 -0.23969
Dref(m107_f, y19_f)delta2 Dref(., .).m107_f|y19_fLow
0.02629 -1.13972
Dref(., .).m107_f|y19_fMidLow Dref(., .).m107_f|y19_fMidHigh
3.60514 9.65879
Dref(., .).m107_f|y19_fHigh
-10.22298

Someone please tell me why I got warned "Refitting with parameters of first Dref weight constrained to zero"
and what are Dref(m107_f, y19_f)delta1 Dref(m107_f, y19_f)delta2

later I added upward and downward

Sey518\$upward = ifelse (Sey518\$m107 < Sey518\$y19, 1, 0)
Sey518\$downward = ifelse (Sey518\$m107 > Sey518\$y19, 1, 0)
drm_SDNN_mob = gnm(R_SDNN ~ Dref(m107_f, y19_f,
delta = downward + upward),
family=gaussian, data = Sey518)
DrefWeights(drm_SDNN_mob)

#> DrefWeights(drm_SDNN_mob)#error
#Refitting with parameters of first Dref weight constrained to zero