PDA

View Full Version : Anisotropy in the Hubble constant?



William
2008-Aug-30, 04:37 AM
The authors of this paper assert that the Hubble flow data shows the expansion of space is anisotropic, by around 13%. The explanation that the space expansion is anisotropic does not make sense to me physically. (There does not appear to be correlation of voids with surrounding high redshift objects which would be the case if the cause was anisotropic space expansion.)

In addition, as noted in the paper the anisotropic expansion is not random and our galaxy is aligned with the anisotropic feature.

I have not looked in detail at the CMB data. Is there any correlation of CMB observations to support this finding?

“Anisotropy in the Hubble constant as observed in the HST Extragalactic Distance Scale Key Project” results by M. McClure & C. Dyer


A statistically significant difference in expansion rate of 9 km s−1 Mpc−1 is found to occur across the sky. Comparing maps of the sky at different distances appears to indicate two distinct sets of extrema with even stronger statistically significant variations. Within our supercluster, variations tend to occur near the supergalactic plane, and beyond our supercluster, variations tend to occur away from the supergalactic plane. Comparison with bulk flow studies shows some concordance, yet also suggests the bulk flow studies may suffer confusion, failing to discern the influence of multiple perturbations


http://arxiv.org/abs/astro-ph/0703556v1

GOURDHEAD
2008-Aug-30, 01:18 PM
Could a reasonable interpretation be that an isotropic expansion is being modulated by the clumpiness of galactic clusters and superclusters giving the appearance on anisotropy? Has anyone been able to postulate the scale at which "local cohesion" from gravitational attraction supercedes cosmological expansion?

William
2008-Aug-31, 03:52 AM
In reply to GOURDHEAD's comment:


Could a reasonable interpretation be that an isotropic expansion is being modulated by the clumpiness of galactic clusters and superclusters giving the appearance on anisotropy? Has anyone been able to postulate the scale at which "local cohesion" from gravitational attraction supercedes cosmological expansion?

The redshift anisotrophic does not appear to be due to cluster formation. Cluster formation would create specific motion of the galaxies (redshift and blue shift) toward the cluster is orders of magnitude less motion than is required to explain this observation. (If the redshift was due to expansion of space at the same relative distance from the earth, but in different regions of space there is a difference in space expansion velocity of 3000 km/sec.

This paper should probably have be entitled redshift anomalies rather than Hubble anisotrophy. See figure 1 and 2 in the paper, where the authors connect the redshift anomalies with contour lines. It looks like galaxies in regions of space are affect the redshift of other galaxies or there appears to be something that is evolving in a large regions of space that affects the redshift of all galaxies in that region of space.

The authors suggest that the difference is due to expansion of space variance from one region to another, however, that explanation would require that there would be strips of different Hubble velocities rather than contour lines.

Does that make sense?

http://arxiv.org/abs/astro-ph/0703556v1

Jerry
2008-Sep-05, 12:49 AM
In reply to GOURDHEAD's comment:



The redshift anisotrophic does not appear to be due to cluster formation. Cluster formation would create specific motion of the galaxies (redshift and blue shift) toward the cluster is orders of magnitude less motion than is required to explain this observation. (If the redshift was due to expansion of space at the same relative distance from the earth, but in different regions of space there is a difference in space expansion velocity of 3000 km/sec.

This paper should probably have be entitled redshift anomalies rather than Hubble anisotrophy. See figure 1 and 2 in the paper, where the authors connect the redshift anomalies with contour lines. It looks like galaxies in regions of space are affect the redshift of other galaxies or there appears to be something that is evolving in a large regions of space that affects the redshift of all galaxies in that region of space.

The authors suggest that the difference is due to expansion of space variance from one region to another, however, that explanation would require that there would be strips of different Hubble velocities rather than contour lines.

Does that make sense?

http://arxiv.org/abs/astro-ph/0703556v1
I can't make 'expanding space' into a rational argument; so I wouldn't have a clue how to tweak the rate or assign variable attributes.

There should at first be an attempt to identify possible observational errors and cosmic baryonics that could explain the appearance of anomalies - we know that the most distant galaxies appear warped and distorted as well as reddened and redshifted. Gas and gravity - we are working with bent rulers. Start with what we know - baryonic matter and such - eliminate and constrain as much as possible, applying dark stuff only when absolutely necessary, and always with a big asterik*

*This is a place holder, until we have a better explanation.

William
2008-Sep-05, 03:00 AM
Originally Posted by Jerry: There should at first be an attempt to identify possible observational errors and cosmic baryonics that could explain the appearance of anomalies - we know that the most distant galaxies appear warped and distorted as well as reddened and redshifted. Gas and gravity - we are working with bent rulers. Start with what we know - baryonic matter and such - eliminate and constrain as much as possible, applying dark stuff only when absolutely necessary, and always with a big asterik*

Your approach to problem solving makes sense to me. Look for an explanation that does not require changing the laws of physics or positing new particles, new dimensions, as a first solution.

I would suggest also there is a benefit to scientifically theorize and compare different theories. Faraday was a master of that approach. He had a list of different hypothesizes, including ones from other scientists which he thought he had disproved that he continued to rework and discuss in his "diary'. He appeared to show no bias and would continue to try to revive a hypothesis. In the case of this problem he would have looked for a solution with and without dark matter.

In terms of the expansion of space anomaly or the redshift anomaly if space is not anisotropically expanding, is there other data that supports the existence of large scale anisotropic in the cosmos?

I was trying to think of a mechanism that would create a cosmic magnetic field.

http://arxiv.org/ftp/astro-ph/papers/0703/0703325.pdf

Does the Universe Have a Handedness? Michael J. Longo

M. Longo's data that shows alignment of spiral galaxies with a cosmic magnetic field is interesting.


We also see that the spin asymmetry is well aligned with the North Galactic Pole of our Galaxy (NGP in Fig. 3) at (193°,27°). This is due to the fact that our Galaxy also has its axis along the spin alignment with a handedness like that of the majority of the spiral galaxies. These two axes are 8.4° apart. The probability that this came about by chance is 0.5%. Since most astronomical surveys, including the SDSS, tend to make observations toward the NGP and SGP to avoid the obscuration due to the Milky Way, the spin alignment axis is accidentally aligned with the SDSS coverage.


The approximate agreement of the spin alignment axis with the WMAP quadrupole/octopole axes reinforces the finding of an asymmetry in spiral galaxy handedness and suggests that this special axis spans the universe. The fact that the spin asymmetry appears to be independent of redshift suggests that it is not connected to local structure. On the other hand, the spiral galaxy handedness represents a unique and completely independent confirmation that the AE is not an artifact in the WMAP data due to foreground contamination.

parejkoj
2008-Sep-05, 06:29 PM
Spiral galaxy spin alignment is isotropic. See the results from Galaxyzoo, which contain many more objects, with more robust analysis of spin direction, than any such previous study. They also find that people seem to preferentially see on spin direction over another, which could result in the biases that the above studies found.

BigDon
2008-Sep-05, 09:21 PM
Can someone define anisotropy (and isotropy for that matter) for folks playing the home game?

parejkoj
2008-Sep-06, 12:32 AM
iso = equal, tropos = direction, thus isotropy means equal in all directions or the same in all directions, etc. an means not.

BigDon
2008-Sep-06, 03:42 AM
Thank you, Parekoj.

I actually knew that, but I needed it as a lead in for my next question: Why is this of special significant? I'm not being snarky a bit.

parejkoj
2008-Sep-06, 04:09 PM
Why is what of special significance? William has brought up a few different things in this thread, and I haven't read them, so can't comment directly.

Answering your question in a more general sense:

Two assumptions underlying modern big bang theory are homogeneity (any given "chunk" of the universe is essentially the same as any other "chunk," for sufficiently large "chunks") and isotropy (there is no preferred direction or place in the universe, so if we look out from our position, things should be the same in each direction). There are ways to test this, for example if there tended to be more clockwise spirals than anti-clockwise spirals (from our point of view) these assumptions would be wrong (since there is no a-priori reason for a galaxy to twist one way or the other). This is one thing that the Galaxyzoo project tested, and they found that spiral directions are isotropic (at least within the SDSS coverage area), but that people were biased. Some earlier posts at the Galaxyzoo Blog (http://galaxyzooblog.org) have more on this.

Remember, homogeneity and isotropy are assumptions of BBT. But they are assumptions that we can test through observation. So far, homogeneity has passed with flying colors; on large scales, such as that probed by the 2dF and SDSS, structure in the universe is statistically uniform. Isotropy is harder to test, but the WMAP results imply it is correct. There have been some recent papers suggesting that if we were in the middle of a very large underdense region (supervoid, if you will), it might produce effects that would look like dark energy. Again, this can, and will be tested with new observations such as the Planck satellite.

William
2008-Sep-06, 05:15 PM
In reply to parejkoj: Spiral galaxy spin alignment is isotropic. See the results from Galaxyzoo, which contain many more objects, with more robust analysis of spin direction, than any such previous study. They also find that people seem to preferentially see on spin direction over another, which could result in the biases that the above studies found.

Thanks for the link parejkoj:

I am not sure the finding that the general population appears to mis classify spiral galaxies using the Galaxy Zoo site tools necessary rules out Michael Longo's results. (i.e. I was expecting a paper that contradicted Longo's finding.)

Are you saying that all humans mis classify spiral galaxies, for some physiological reason?

William
2008-Sep-06, 07:10 PM
The Longo analysis use both human and a computer program to distinguish between spiral direction (clockwise or anticlockwise). The human tests agreed 100% with the computer test which ruled out human bias for the analysis.

It sounds based on what has been presented in the forum that Longo's finding are still valid.


http://arxiv.org/ftp/astro-ph/papers/0703/0703325.pdf

Does the Universe Have a Handedness? Michael Longo

In an earlier version of this analysis (Longo, 2007) the JPEG file for each spiral galaxy was submitted to an IDL program to determine its handedness. The algorithm used for this analysis was based on a rotating one-armed spiral mask similar in shape to half of the spiral in Fig. 1(c). However, the algorithm was developed using spirals at fairly low redshifts, and it was belatedly realized that its efficiency at larger redshifts was poor. Visual scanning was found to be considerably more efficient and it was easy to take precautions to eliminate any human bias toward left or right-handedness. The results of the two analyses are in complete agreement. Only the visual scan results that include ≈50% more galaxies will be presented here. (Comment, my bolding.)

Cougar
2008-Sep-07, 09:23 PM
...each spiral galaxy was submitted to an IDL program to determine its handedness.

Did Longo define the "handedness" of a galaxy? Seems to me a clockwise spin viewed from one side would be counter-clockwise from the other. So it's not an intrinsic quality of the galaxy, but dependent on the observer?

trinitree88
2008-Sep-08, 12:08 AM
Why is what of special significance? William has brought up a few different things in this thread, and I haven't read them, so can't comment directly.

Answering your question in a more general sense:

Two assumptions underlying modern big bang theory are homogeneity (any given "chunk" of the universe is essentially the same as any other "chunk," for sufficiently large "chunks") and isotropy (there is no preferred direction or place in the universe, so if we look out from our position, things should be the same in each direction). There are ways to test this, for example if there tended to be more clockwise spirals than anti-clockwise spirals (from our point of view) these assumptions would be wrong (since there is no a-priori reason for a galaxy to twist one way or the other). This is one thing that the Galaxyzoo project tested, and they found that spiral directions are isotropic (at least within the SDSS coverage area), but that people were biased. Some earlier posts at the Galaxyzoo Blog (http://galaxyzooblog.org) have more on this.

Remember, homogeneity and isotropy are assumptions of BBT. But they are assumptions that we can test through observation. So far, homogeneity has passed with flying colors; on large scales, such as that probed by the 2dF and SDSS, structure in the universe is statistically uniform. Isotropy is harder to test, but the WMAP results imply it is correct. There have been some recent papers suggesting that if we were in the middle of a very large underdense region (supervoid, if you will), it might produce effects that would look like dark energy. Again, this can, and will be tested with new observations such as the Planck satellite.


parejkoj. The universe appears not to be homogeneous at scales as large as one billion light-years according to this study of quasar polarizations. see:http://arxiv.org/abs/astro-ph/0507274

pete

note bien: That magnetic fields are fairly ubiquitous in the universe, and neutrinos interacting with baryons via weak currents are equally ubiquitous, showing gross scale anisotropy due to parity effects was not first suggested by Mr. Longo, ...that I can assure you. I put that in writing in April 1982 and have certainly been both visible and audible at the National AAPT Meetings, sometimes held jointly with the American Physical Society since November of 92. Like Teddy Roosevelt..."walk softly, but carry a big stick"...or an extended paper trail, in the case of science. pete

Jens
2008-Sep-08, 02:08 AM
Did Longo define the "handedness" of a galaxy? Seems to me a clockwise spin viewed from one side would be counter-clockwise from the other. So it's not an intrinsic quality of the galaxy, but dependent on the observer?

Of course, that's true, but I think the point is that the galaxies seem to be aligned in the same position as seen from our perspective. For example, if you had a wind generator farm, you would see all the fans to the north spinning clockwise, for example, and all the fans to your south spinning counterclockwise. Although it depends on the observer, it would still indicate that they are aligned in some way.

Nereid
2008-Sep-08, 03:03 PM
The Longo analysis use both human and a computer program to distinguish between spiral direction (clockwise or anticlockwise). The human tests agreed 100% with the computer test which ruled out human bias for the analysis.

It sounds based on what has been presented in the forum that Longo's finding are still valid.


http://arxiv.org/ftp/astro-ph/papers/0703/0703325.pdf

Does the Universe Have a Handedness? Michael Longo(my bold)

There are many ways you can proceed here.

On the one hand, there are two published papers with conclusions that seem to be conflicting.

Yet there are big differences between the two studies, as reported in those papers.

Can you devise a hypothesis that tests the (joint) conclusions of the two papers? That would, could, show both to be 'right', in some sense?

Can you repeat either of studies, independently, from the same sets of input data?

There's an interesting implication that follows from your post (that I am quoting):

It sounds based on what has been presented in the forum that the Galaxy Zoo team's finding are still valid.

William
2008-Sep-10, 02:19 AM
Originally Posted by Nereid: Yet there are big differences between the two studies, as reported in those papers.

Can you devise a hypothesis that tests the (joint) conclusions of the two papers? That would, could, show both to be 'right', in some sense?

Can you repeat either of studies, independently, from the same sets of input data?

There's an interesting implication that follows from your post (that I am quoting):

It sounds based on what has been presented in the forum that the Galaxy Zoo team's finding are still valid.

I did not see a paper presented for the Galaxy Zoo findings, only a web link. The Galaxy Zoo team's ran a test which they said showed humans for a physiological reason non-randomly misclassified spiral galaxies. Why would that be so? Is there a selection problem with the people they chose for the task? There is no explanation for the finding. Is it possible there was human purposeful miss-classification?

The Longo findings do not support the Galaxy Zoo's findings. Longo used a computer program to classify 50% of their spirals galaxies and then used humans for an additional 50% of the study data. A portion of the human tested data was re-run as a check with the computer selection program and there was 100% agreement. I think it is perfectly reasonable if people are trained and their work checked for humans to classify spiral spin direction. The Longo conclusion is humans can accurately classify spiral direction and for the near vicinity universe, spiral galaxy alignment is non-random.

I believe the Longo findings are still valid based on what has been presented in the forum. The Longo assertion of alignment of spiral galaxies is not the first study that has shown non-random spiral galactic alignment.

Accepting Longo's result with an asterisk (Confirm later or wait for more data.), the next question is the mechanism to create the large scale magnetic field and proof for the hypothesized mechanism.

There was that paper that showed high redshift galaxies are 10 times more dense than current galaxies. Some mechanism is required to unpack the spiral galaxy. And there was Spaceman Spiff's comment that spiral galaxies reach a maximum size. There was also that comment concerning the spiral galaxy winding problem which has never been satisfactorily addressed.

As a methodology sometimes looking at a set of anomalies provides better clues than a single anomaly. i.e. There is always uncertainty in the data and the conclusions.

Nereid
2008-Sep-10, 02:28 AM
William: I did not see a paper presented for the Galaxy Zoo findings, only a web link.

May I ask how hard you tried to get hold of the paper?

How familiar are you with ADS, and similar databases?

In the case of the Galaxy Zoo (GZ) paper, you may consider reading this thread (http://www.galaxyzooforum.org/index.php?topic=271980.0), and clicking on the relevant links, to get a PDF copy of the paper that was accepted for publication (or you could use your subscription to MNRAS and get your own, paper, copy).

I think that once you've had a chance to read the published GZ paper you'll find most, if not all, your questions answered. If not, I'm sure several members here would be more than happy to help you further.

In any case, when you've finished with your reading, let's go over the points in my post again, especially this one (there's a minor edit): "It sounds based on what has been presented in [the published paper] that the Galaxy Zoo team's finding are still valid."

ETA: I'm leaving aside, for now, all the comments in your post other than those relating to the material in the Longo and GZ papers (and those cited therein).

William
2008-Sep-11, 03:10 AM
This is a link to a preprint to the paper.

http://arxiv.org/abs/0803.3247

Nereid
2008-Sep-11, 06:50 AM
This is a link to a preprint to the paper.

http://arxiv.org/abs/0803.3247
It's more than that ... "Accepted for publication in MNRAS" tells you that it's also the same as what will be (or has already been) published in MNRAS (modulo some formatting perhaps, and bar a gross dropping-of-the-ball).

William
2008-Sep-13, 04:40 AM
Longo's paper examines the spiral galaxy rotation in a sphere of redshift 0.04 about our galaxy. As he notes the spiral galaxies in the sphere have a preference rotation.

The Zoo finding shows that the human observers that participated in the Zoo study were biased to finding anti-clockwise galaxies. Interesting the Zoo finding does show the spiral galaxy spin direction is clumpy rather than random is preferred in either the anticlockwise or the clockwise as one move out with redshift similar to the clumpy change in the hubble "constant" as one moves out with redshift.

It should be noted that Longo's finding required the observation of both clockwise and anticlockwise galaxies, as the galaxies in our local space have a preferred rotational direction of clockwise then when viewed in one hemisphere of the sky and anticlockwise when viewed from the other hemisphere of the sky. (See Longo's paper. Longo uses a picture of screws in the sky to illustrate how that geometric principle works.)

Therefore for Longo's findings to be valid the Longo observers must be able to distinguish both clockwise and anti-clockwise galaxies.

It appears the Zoo observers have some sort of physiological problem with identifying the clockwise galaxies and for some unexplained reason misclassified clockwise galaxies as anticlockwise. The Longo observers appear to not have that problem.

It sounds based on what has been presented in the forum that Longo's finding are still valid.

Longo's paper has other interesting findings. This is a link to Longo's paper.

http://arxiv.org/ftp/astro-ph/papers/0703/0703325.pdf


The SDSS DR5 database [2] contains ~40000 galaxies with spectra for redshifts <0.04 with a wide coverage in right ascensions (RA) and a more limited range of declinations (δ). A few percent of these are spiral galaxies that can be used in a search for a preferred handedness. A universe with a well-defined handedness is illustrated very schematically in Fig. 1(a), which shows a polar plot of spiral galaxies in RA and redshift (z) or distance [3]. Note that to us, galaxies in one hemisphere would be right-handed while they would be left-handed in the opposite hemisphere. (If a predominance of left- or right-handedness were seen in all directions it would be an indication of a bias or systematic error preferring that handedness.) Figures 1(b,c) show typical spiral galaxies from the SDSS and gives our convention for “spin” directions. Note that only the component of the spin along our line of sight can be observed.


Many spiral galaxies have undergone collisions with other galaxies since their formation. Such collisions tend to mix “orbital” angular momentum of the galaxies with any primordial spin angular momentum they may have formed with. Collisions between galaxies trigger prolific star formation. Galaxies with recent star formation tend to be bluish. In order to enhance any signal for a preferred handedness, the bluest galaxies were removed from the sample. This selection was imposed by requiring that the ultraviolet
and far infrared magnitudes satisfy the condition |U-V| > 1.6. Approximately 1660 galaxies were in the final sample.


The Wilkinson Microwave Anisotropy Probe (WMAP) studied the cosmic background radiation [6]. Their results for the angular power spectra have been analyzed by Schwarz et al. [7] and many others. Schwarz et al. show that: (1) the quadrupole plane and the three octopole planes are aligned, (2) three of these are orthogonal to the ecliptic, (3) the normals to these planes are aligned with the direction of the cosmological dipole and with the equinoxes. The respective probabilities that these alignments could happen by chance are 0.1%, 0.9%, and 0.4%. This alignment is considered to be so bizarre that it has been referred to as "the axis of evil". [8] The alignment with the ecliptic and equinoxes is especially problematic because this would suggest a serious bias in the WMAP data that is related to the direction of the Earth's spin axis, which is highly unlikely. Resolving this quandary requires data from another source with different systematics than
WMAP.


A large-scale cosmic magnetic field along this axis would naturally lead to an alignment of spiral galaxies like that seen here. Campanelli, Cea, and Tedesco [10] have suggested that an ellipsoidal universe, possibly the result of a uniform cosmic magnetic field ~10-9 G, could explain an anomaly observed in the quadrupole amplitude of the WMAP data. Non-uniform intergalactic magnetic fields have been observed [11]. Their origin is puzzling and has been the subject of much discussion [12]. If a cosmic magnetic field exists, galaxies would tend to form with the axis of the galactic disk along the magnetic field. A large-scale alignment of spiral galaxies would be the first direct evidence for a uniform [13] primordial magnetic field and would have major implications in the development of large-scale structure in the universe and in cosmology as a whole.

Nereid
2008-Sep-13, 10:54 PM
As the Galaxy Zoo (GZ) paper came out sufficiently long after the Longo one, and as the two appear to have been able to cooperate well*, the GZ team was able to undertake a detailed comparison of their results with those of Longo.

Further, as the Longo sample (drawn from SDSS DR5) is a subset of the GZ one (drawn SDSS DR6), a direct comparison between the Longo classifications and the GZ ones is possible.


Of this sample, L07a found 1,266 and 1,365 galaxies to be Z and S-wise respectively, with the remaining ‘Unknown’ (U) 6.We find cleanbc classifications for 2,501 of the sample, and the level of agreement is tabulated in Table 3. We see very good agreement between our classifications, and the 10 cases where the handedness classifications contradict each other we have double checked the SDSS images and find the disagreement can be put down to misclassification in L07a - a demonstration of the usefulness of multiple classifications per galaxy.

Interestingly, the 238 "non-U Longo galaxies" that are not in the GZ cleanbc set have a greater "S" bias than the total Longo sample, though I doubt that the difference is (statistically) significant - see Table 3 of the GZ paper.

The GZ paper makes the point that as there is only partial coverage of the sky, there is a monopole-dipole degeneracy in any apparent asymmetry, and that when "when the monopole term is included and marginalised over"^, "[w]e find the evidence for a dipole and thus for a preferred direction diminishes, and the constraints on D become consistent with zero within ∼ 1σ"


It sounds based on what has been presented in the forum that Longo's finding are still valid.
Er, no ... if you read section 3.3 of the GZ paper, carefully, you'll see that Longo's conclusions are not significantly different from a more appropriate null hypothesis.

* "These numbers correspond to the dataset gratefully provided to the authors by M. Longo", for example (footnote, p7 of the GZ paper)
^ yep, they're Brits!

William
2008-Sep-15, 03:23 AM
As the Galaxy Zoo (GZ) paper came out sufficiently long after the Longo one, and as the two appear to have been able to cooperate well*, the GZ team was able to undertake a detailed comparison of their results with those of Longo.

When the GZ paper came out seems in this case to irrelevant. That they used Longo's data maybe true, that the Zoo paper's conclusion are valid is not. The Zoo's analysis is invalid because their observers, as stated in the Zoo paper, are biased.

As I said the Longo's conclusion required his human spiral galaxy classifiers to select an above normal average of clockwise and anti-clockwise galaxies depending with the spiral galaxies are in the northern or southern hemisphere for his conclusion to be valid.

To me it seems rather obvious that the Zoo methodology of selecting spiral galaxy classifiers introduced a human bias towards selection preferentially anticlockwise spirals.

The Zoo findings require some unexplained reason why their observers selected preferentially anticlockwise spirals. (Your comment and the Zoo paper provides no explanation as to why their observers have an anticlockwise spiral galaxy bias.)

If the Zoo group was interested in the science, they would have repeated the spiral galaxy classification with paid human unbiased spiral galaxy classifiers. That they chose to publish, after finding an unexplained bias, which they try to pass as universal bias to classifying all galaxies as anticlockwise by human classifiers or something that is mysteriously incorrect with their web site, indicates that the Zoo group in the case of this subject are not interested in the science, in my opinion.

The objective is to solve a scientific problem not to win a debate.

parejkoj
2008-Sep-15, 04:23 AM
When the GZ paper came out seems in this case to irrelevant. That they used Longo's data maybe true, that the Zoo paper's conclusion are valid is not. The Zoo's analysis is invalid because their observers, as stated in the Zoo paper, are biased.

huh... guess you didn't bother to read the whole paper, since the Galaxyzoo team go to great lengths to describe how they first identified the bias and then corrected for it. And "their observers" are something of the order of 100,000 people.

Maybe you should actually read Nereid's comments again, after spending some more time with that Galaxyzoo paper (and maybe their other papers as well, to get a handle on their systematics and controls).


If the Zoo group was interested in the science, they would have repeated the spiral galaxy classification with paid human unbiased spiral galaxy classifiers.

Odd that you say this. How would one go about finding "paid human unbiased spiral galaxy classifiers?" Do they have them at the local job recruitment center? How could one identify a biased vs. unbiased observers?

Oh, I know! Maybe one could take observations from a very large group of people, check that those observations are consistent, look for systematic trends, and then create tests that determine whether there are any biases in the observation, and use those tests to correct the result. Huh. That might be a good idea. If only the Galaxyzoo team had thought of that... :rolleyes:

Nereid
2008-Sep-16, 12:14 AM
When the GZ paper came out seems in this case to irrelevant. That they used Longo's data maybe true, that the Zoo paper's conclusion are valid is not. The Zoo's analysis is invalid because their observers, as stated in the Zoo paper, are biased.

As I said the Longo's conclusion required his human spiral galaxy classifiers to select an above normal average of clockwise and anti-clockwise galaxies depending with the spiral galaxies are in the northern or southern hemisphere for his conclusion to be valid.

To me it seems rather obvious that the Zoo methodology of selecting spiral galaxy classifiers introduced a human bias towards selection preferentially anticlockwise spirals.

The Zoo findings require some unexplained reason why their observers selected preferentially anticlockwise spirals. (Your comment and the Zoo paper provides no explanation as to why their observers have an anticlockwise spiral galaxy bias.)

If the Zoo group was interested in the science, they would have repeated the spiral galaxy classification with paid human unbiased spiral galaxy classifiers. That they chose to publish, after finding an unexplained bias, which they try to pass as universal bias to classifying all galaxies as anticlockwise by human classifiers or something that is mysteriously incorrect with their web site, indicates that the Zoo group in the case of this subject are not interested in the science, in my opinion.

The objective is to solve a scientific problem not to win a debate.
Yeah, what parejkoj said, and more.

William, I don't think you read the GZ paper, esp section 3.3; or, if you did, you seem to have quite misunderstood what you read.

First, the GZ team reproduced the Longo findings, galaxy by galaxy, with the following two exceptions:

* several hundred galaxies classified as S or Z by Longo did not receive weighted GZ classifications clean enough for the team to report; these have a bias (towards S) that is greater than in the total Longo sample (but that may not be significant)

* 10 galaxies got discordant classifications; the GZ team checked each themselves and found Longo's classification to be in error; 10 is way too few to be called significant.

Second, the GZ team did a statistical analysis that Longo did not do: they tried to account for the bias introduced into Longo's (and the GZ team's) analysis due to the fact that the sample was not from the whole sky. When they (the GZ team) did this, based on the Longo dataset, they found no statistically significant anisotropy.

Third, the full GZ sample, cleanbc is some 37k galaxies, compared with ~2.8k in the Longo sample.

Fourth, the GZ team did not find a statistically significant asymmetry in their ~37k cleanbc dataset.

Perhaps the most important point you seem to have missed is that the GZ team's cleanbc dataset is, statistically, identical to Longo's dataset, as they showed in Section 3.3 of their paper.

The second-most important point you seem to have missed is that the GZ team was able, via their methodology, to correct for the classification bias they discovered.

The objective is to solve a scientific problem not to win a debate.
Indeed.

And that's just what the GZ team's paper did ... or at least made a big step towards, standing on Longo's shoulders.

Oh, and in case you missed it, the GZ team's paper also reports on a comparison with an earlier anisotropy study ... noting that Longo's results are inconsistent with this 1995 research.

Of course, the universe may very well contain an anisotropy in S/Z galaxies ... but neither the Longo nor the GZ team found a statistically significant signal of such an anisotropy.

William
2008-Sep-16, 02:15 AM
Nereid,

Cluster structure and alignment is interesting. Let's leave the spiral galaxy alignment question until there is a repeat of this study. For me the Zoo's finding that their observers were biased to finding anticlockwise spirals invalidated their data and results. I did not understand how the Longo study was limited to a sphere of z=0.04 due to limitation in the clarity of the spiral images for classification how the Zoo study could be extended to z=1.5.

Comment:
I have more observational data that seems to be relevant to how the galaxies in a cluster appear to interacting. (X-ray emission of intergalactic gas and the calculations of the expected time for the hot intergalactic gas to cool and collapse. I will see if I can get enough data to start a thread.)

Nereid
2008-Sep-16, 12:14 PM
William, I going to repeat what parejkoj said, and what I said, because it seems you have not yet grasped it.

The GZ team addressed the bias question, and produced a sample free from bias (cleanbc).

When they compared the classifications of Longo (some 2.8k of them) with the classifications of the same galaxies in cleanbc, they found them exactly the same* ... except for 10, all of which were misclassified by Longo.

So one challenge for your stance is if the GZ results on the Longo sub-set ~2.8k (of ~37k) are essentially the same, and you consider the GZ data to be invalid, then you must also consider the Longo data to be invalid (and hence that study's results invalid too).

A similar, but somewhat different, challenge is how to reconcile Longo's results with the earlier 1995 paper (the two are inconsistent), when the GZ team's results enables you to see how all three are consistent.

Now of course you are free to regard the results of any research any way you like, but the room to move in terms of standard science is small.

Oh, and what's the reference to "z=1.5"? I don't know how that is related to the GZ team's paper, or their approach, can you clarify please?

* there are some 238 galaxies classified by Longo that are not in cleanbc, because they did not receive a sufficiently high proportion of single classifications.