Summary 

The authors split up the galaxies into redshift bins. They do this in
two different ways. For GOODSS galaxies, they associate a galaxy with
a bin if its redshift exceeds a certain threshold. For GOODSN
galaxies, they associate a galaxy with a bin if the point estimate is
within a redshift range.

The reader might think that using the two different methods would lead
to different results. But, the authors reassure us that the binning
results match for the two methods because the confusion matrices are
diagonally dominant.

The authors show that the binning results are robust to the threshold
value (at least for 0.6 vs. 0.8).

There are some expectations from cosmologists that were fulfilled by
the authors. Cosmologists believe that the younger universe is more
chaotic than the current universe since the current universe has more
galaxy mergers. Also, M and I statistics increase with redshift, and
the proportion of zeroes decrease in both statistics from band V to
band H.

The authors perform bandwidth selection and significance tests and
then show their results in tables. The authors have evidence for their
hypothesis that there is evolution in galaxy morphology over cosmic
time.

The authors want to check whether the two estimation approaches (which
are the two? oh, I got this later, but make sure you say it early in
the text) give similar results. The discrete bin approach is commonly
used in cosmology, so the authors hope to get similar results with
their method.

The two approaches yield similar results in the GOODSS
galaxies. However, their estimates in the high redshift region are not
reliable because the confidence intervals get large. The two
approaches yield different results in the GOODSN galaxies, but only
for some. This indicates that the continuous redshift approach gives
largely consistent results compared to the discrete bin approach.

Comments 

It might be useful to start the results section with an introductory
sentence saying something like, "From the procedures that we used, our
results indicate that..."

The results section is missing what we discussed in class as narrative
tension. This is partially due to the fact that I'm only reading
section 4, but I think it could still help. I am not sure why you
would "associate a galaxy with a bin if the probability of its
redshift being within that bin exceeds a certain threshold (either 80%
or 60%)". What is the problem you are trying to solve? Is it that you
are trying to classify the galaxies into two separate

Could you add captions to the tables in addition to the table title?
That way when the reader looks at the table she knows what the
punchline is.