Multidimensional Scaling - Science topic

I am looking for the Relational Resilience Scale (RRS): The RRS is a self-report, multidimensional scale measuring couple's ability to recover after traumatic life experiences with 27 item which was developed by Aydogan and Ozbay.

Hi,

Kindly, I have two questions related to the multidimensional scale model:

1- Is there a minimum number of objects (product) I need to apply to use the Multidimensional Scale (2 dimensions, not 3)?

2- What is the best Likert scale option to apply for Multidimensional Scale (1-4) (1-5) or (1-7)? "considering that I'm trying to reduce the neutral answer option."

Thank you!

Hello researchers, I have this doubt and it is because as the name itself says "non-metric" (NMDS). I have assumed to use this technique but I don't know its limitations.

Could you get me out of this doubt please?

although I had thought about using the MFA but decided to go for the NMDS

I am investigating how (dis-)similar bachelor students interested in pursuing a Master's degree perceive according offers (i.e. Master programmes from different universities). Altogether, I am comparing 15 Master degree programmes. Hence overall there are (15*14)/2 = 105 pairwise comparisons to be made. For obvious reasons, I cannot ask each respondent to make similarity ratings for each of the 105 pairings. I wonder if it is legitimate to have each repondent rate one (and only one) pair to which they are randomly assigned. As I do have access to a sample of several thousand students, I expect to have at least 10 similarity ratings for each pair of objects compared. I wonder if this procedure is statistically sound or if, in multidimensional scaling analysis, each participant must rate all possible pairings of objects (Master degree programmes in this case).

Hi all,

Do I need to present a stress value for the fitting of my principal component analysis (as you would with an nMDS plot)?

If so, how can I calculate this in R?

For example in the paper De Jong, J. P., & Den Hartog, D. N. (2008). Innovative work behavior: Measurement and validation. EIM Business and Policy Research, 8(1), 1-27,

Innovative output is a dimension of innovative work behavior. Could I use this separately in a survey?

I am constructing a multidimensional scale that measures factors as well as the prevalence of academic dishonesty. the factors include intrinsic motivation, extrinsic motivation, identified regulation ( which are correlated), parental influence, peer influence, social influence (which are correlated), and academic dishonesty. but not all the factors are correlated with each other. which rotation should I use? and the scale is prepared based on a developed model with strong theory. the model is supported when quartimax is used. but varimax is more accurate which is changing the solution. similarly, parallel analysis is giving 9 factors with the first tryout on sample of 329 and when went for second try for EFA-2 on a sample of 400 it is suggesting 8 factors... how will we get a proper model if it changes based on the sample

Instances of the 4th dimension include:

Time in Minkowski’s space-time (Raum und Zeit).

As flow or motion in various 4/3 laws.

But:

In a space-time distance, time squared is preceded by a sign opposite to that of the other lengths squared. Time is different.

Flow, motion and time trace a moving point along a line. The 3 spatial dimensions are static.

In the 4/3 law pertaining to energy the same energy in 4 dimensions has 4/3 as much energy in the corresponding 3 dimensional space. How can energy occupy a 4th dimension that models a moving point? Perhaps the model in the 4/3 law is wrong or incomplete? If it is incomplete, how is it incomplete? Is some aspect of time missing? In this portion of the comment on the question, accounting for the 4th dimensional status of motion affects understanding of the 4/3 laws.

Or is the 4th dimension nothing more than a mathematical construct?

Through multidimensional scaling I have obtained spaces of three different data sets. How could I go about comparing the spaces of these three data sets? Note that I'm not intending to find out which has the best goodness-of-fit, but how well the spaces fit onto each other.

Carroll and Ostlie in An Introduction to Modern Astrophysics, second edition at page 1099 remark: “Cosmological redshifts are caused by the expansion of the space through which the light travels, so for extremely large distances the total elongation of the wavelength depends on how the expansion of the universe has changed with time.” The 4/3 laws are based on dimensional capacity and imply a distance in 3 dim space stretches by 4/3 compared to the same distance in 4 dim space-time. Is there a connection?

Hi~

I was trying to run a multidimensional unfolding with the Prefscal function of SPSS, but it wouldn't execute because:

“Invariant part of the data found. Check, depending on conditionality chosen, your data for constant parts. ”

The variables entered were 10 colums of dissmilarity scores. I don't know if I understand what the warning means, but these scores surely are not constant.

Could you give me some advice on that？

The no. of camera traps and the area of the study sites are not equal.

Multidimensional scaling (MDS) is used by multivariate data to create distances and analyze, similar to analyzing the data using categorical principal components analysis of normalized data. What is the difference between PROXSCAL and ALSCAL in MDS?

I wish to use self-monitoring scale developed by Richard D. Lennox and Raymond N. Wolfe ("Revision of self-monitoring"). It is a 2 factor structure scale with total 13 items-

(1) ability to modify self-presentation-7 items & (2) sensitivity to expressive behaviour of others-6 items .

But for a part of my study I am trying to show the impact of self-monitoring on consumption behaviour , so I feel only the first dimension (ability to modify self-representation) is useful for me . Also I have 6-7 more latent variables so the survey is already touching 90 questions so I want to minimise the items .

Dear all,

does anybody know how many similarity judgments I need for each pair of stimuli in multidimensional scaling. For clarification: I am not talking about the number of comparisons. I know that there is the option for reduced designs, but this is not what I mean. I mean how many subjects have to indicate their perceived similarity among two stimuli. Does anybody know a source that provides an answer to this question?

Best,

Max

Hi everyone,

I performed multiple regression and a nonmetric multidimensional scaling analyses in order to understand which environmental factors have the biggest impact on the abundance of several nitrogen cycling genes. From what I have been reading, the vectors fitted onto the ordination plot are calculated performing some kind of correlation analysis, so I was hoping to obtain similar results; however they seem to contradict each other.

Why is that?

Here are my multiple regression analysis results and the NMDS plot (colors represent the sampled stations and the shapes represent the sampled depth)

I have images (shreds) and a dissimilarity matrix of these images .

How can I find a virtual coordinates that represent these images in 2-D (convert each image to one point with x & y coordinates) .

These coordinates should be order-respecting according to the dissimilarity matrix which mean if the true order of the images is : 5 - 2 - 1 - 7 - 3 - 4 - 6 .

Then when I find these coordinates and calculate new distance matrix between them it should give the same order .

I'm writing my code in Matlab and I used mdscale (a built-in function) and it's not work good

[coordinates,stress] = mdscale(D,2) ;

It find the coordinates, but it's not order-respecting and I tried another code in R but i didn't get a right solution .

This problem called "Non-classical Multidimensional Scaling" .

Any one can help me to find the true solution and I will be thankful for him/her.

Currently, I am working on a study and I want to identify the correlation of my two variables (Workload perception and Stress Appraisal). I don't have SPSS so I'm using excel in my computation. However, I'm a bit confused because the two scales don't have similar number of items. The first is unidimensional with 13 items and the other one is multidimensional which is consisted of 7 subscales with 4 items respectively. Both of them are a 5-point likert scale. I know that I need to correlate the unidimensional scale to each subscale of the multidimensional scale but I'm not sure if it will lead to a reasonable result considering that scores from a 13-item scale will be correlated to a 4-item subscale. Thanks in advance!

I'm working on a study and the first thing I did is to pilot test the research instruments I will be using. After the pilot testing, I used cronbach alpha to calculate its reliability. The research instruments are multidimensional. It is consised of different subscales. When I calculate for cronbach alpha some subscales showed low reliability scores.

One scale is consisted of 6 subscales and 2 from this subscales have low scores when I used cronbach alpha.

The second scale is consisted of 8 subscales and 1 from these subscales showed low reliability score.

What should I do about it? Can I still use the scales for my research? Or should I simply remove the subscales that show low reliability scores?

I am a PhD student currently exploring questionnaires that I could potentially use to identify older adults (50+ in this project) levels of social isolation, as part of a larger project. I have come across several questionnaires (e.g. Duke Social Support Index, Lubben Social Network Scale, Social disconnectedness questionnaire, Medical Outcomes Study Social Support Survey, Multidimensional scale of perceived social support, Social network index, De Jong Gierveld Loneliness scale) and I am currently weighing up the pros and cons of each to make a decision on which to use in this project.

I have found that some cross over with loneliness, but I would like to have separate 'scores' for social isolation and loneliness, which potentially means separate questionnaires. I have found loneliness questionnaires relatively easy to narrow down, but social isolation I've found much harder.

I am working on diversity analysis, NMDS has been used to discriminate ecosystems , its getting a stress value 0.02,0.03 etc how can is explain the stress value in the interpretation, i may wonder if any provides reliable literature also.

Many thanks in advance

Recent research has shown key important aspects with measurability of theoretical postulates verifying hypotheses parameters, processes, phenomena, and models.

Tensor matrices have necessarily roles to bring about complex nature manifesting spatially and temporarily. Explaining everything that is existing in terms of the fundamental entities have lead to realization of geometric topology space tensor manifold time evolving event gridnetwork.

Einstein's General Theory of Relativity, Quantum to Particle Theory of Everything, String Theory among others have measurability in mind a proof of model requirement automatically. For example, Schwartzchild blackhole mathematics helped to identify, observe, and measure singularity blackhole consequently, the recent telescopic photos observing directly, proving validity with General Theory of Relativity tensor predictive capability.

Providing the thumb rules, below certain associative relationships might connect mathematics with physics to measure model......

. typically scalars, scalar matrices are helpful to get statistical measurements that are analyzable observationally experimentally......

. tensors have stochastical vector matrices that aren't amenable to direct measurements. Hence transforming tensors or matrix tensors to scalar matrix systems are key to make measurable operational parametric graphical experimental observational gridnetworks.

My analyses metrix protocol techniques have yielded a rough estimate of overall globally ~80% of objects universally are measurable statistically. This will mean ~20% are uncertainity stochastic probabilities with a few% inherent immeasurable tensor network, aether may be example. I have space time sense 2x2 tensor grid part of a large tensor matrix that if transformed to 5 dimensional like scalar matrix natural manifolds protocol will help eventually in the quantitative grand unified theory of everything. There are more to come after our QFM/EM modeling going on with collaborative platform TEI.

Given below are a few references that are associated, not exhaustive, suggestions welcome. Additions editions expansions!!!!!

(1) https://www.researchgate.net/project/TOWARDS-THE-GRAND-UNIFIED-THEORY-OF-EVERYTHING-LINKING-FUNCTIONALITY-MODEL-CONCRETE-TO-ABSTRACTIONS?_sg=XNNU-DQ2rx6YNmUfxUjYZ6LaLamHZK8jGl_UOQblslDC1A9ygt_x0mY-iylt49cJFCqpUNKip5okOdqjIf43mH-lRm2JzYvjuM4j including working papers with tensor analyses......

(2)

(3)

(4)

(5) , , , & https://www.researchgate.net/project/A-Synthetic-Magnetic-Unipolar-Array-Not-a-Dirac-Monopole?_sg=zOOB5jYpcBuzrnEjYVwI8d7oSR2-UNEsI1Yu3Jf2e2uQQ9DbTVGxz3rQLfVEU7-n8Y6Fo0ZY-NFy8-gDCORKogPQF099GP5bghsc

(6) Zurek, Wojciech H. (2003). "Decoherence, einselection, and the quantum origins of the classical". Reviews of Modern Physics. 75 (3): 715. arXiv:quant-ph/0105127. Bibcode:2003RvMP...75..715Z. doi:10.1103/revmodphys.75.715 & Dan Stahlke. "Quantum Decoherence and the Measurement Problem" (PDF). Retrieved 2011-07-23.

(7)

(8) https://www.researchgate.net/post/Measurable_function_Real_analysis?_iepl%5BgeneralViewId%5D=UrJPT0Tf7xZ5zvi3IaLKUrm7s2pwF1roCYYk&_iepl%5Bcontexts%5D%5B0%5D=searchReact&_iepl%5BviewId%5D=DuVPrSZERB3gc1XS1sy4M7xRN2eix4sWRMm6&_iepl%5BsearchType%5D=question&_iepl%5Bdata%5D%5BcountMoreThan20%5D=1&_iepl%5Bdata%5D%5BinteractedWithPosition1%5D=1&_iepl%5Bposition%5D=1&_iepl%5BrgKey%5D=PT%3A5cadcf48aa1f099c5752a2f4&_iepl%5BtargetEntityId%5D=Q%3A5cadcf48aa1f099c5752a2f4&_iepl%5BinteractionType%5D=questionView

I am proposing a study to test for moderation between social support and quality-of-life (QOL). For social support I am using the three subscales (family, friends, significant other) Multidimensional Scale of Social Support. For QOL I am using the WHOQOL-BREF which has four domains (Physical, psychological, social, environmental). For a moderator I am using a subscale measuring non-affirmation. I've been advised that i have to write out a hypothesis for each outcome variables (i.e. family support will predict higher Physical QOL; family support will predict higher Psychological QOL etc). If I do this, I will be writing 12+ hypotheses which seems to contradict the APA manuals' recommendation to be concise. I understand I have to first establish hypotheses that my predictor variables will have a relationship with my outcome variables but since there are multiple scales it seems repetitive.

I have a large set of molecules and I'd like to assess its diversity in terms of molecular structure, assuming I'm designing a screening library. My question is not what tools to use, but rather how should I interpret the results?

Let's say I choose clustering. How many clusters indicate a diverse library?

"Modern Approaches in Drug Discovery" suggests comparing Morgan fingerprints against Tanimoto distance matrix with multidimensional scaling applied, among others. But what will it get me?

It is a multidimensional scale and there are three factors. how to correct the alpha scores?

I'm trying to set up my data in a way that allows me to use INDSCAL to find individual data. Each participants comes with a 40x40 table representing their judgements of similarity of 40 stimuli, like so:

1 2 3 ... 40

1

2

3

...

40

Whenever I run ALSCAL to calculate the differences, it says I have 1 matrix and not the number of participants, so my data doesn't process right. How can I set my data up so that SPSS knows which portion of data belongs to one participant, to then go through ALSCAL?

I sampled algae from three wetland sites on 13 sampling events. At each wetland site on each sampling event, I collected one sample of algae from the same location at the centre of the wetland and 4 random samples of algae around the littoral zone (wetland margin). So for each event, there is 1 centre sample vs. 4 random littoral samples in each wetland site.

For each wetland site, I ran a non-metric multidimensional scaling (NMDS; PC-ORD) of taxon abundance × [site by event] matrix to visualise differences in community composition among zone types (centre vs. littoral) throughout the study.

I would like to explore the amount of temporal variability between the centre and the littoral zone algal communities throughout the study for each wetland site. I would like to calculate the total Euclidean distance between successive event samples for the centre and for the littoral zone in ordination space using the two NMDS axes in the final solution. This method would be fine for the central zone, since I sampled the same centre location for each event. But for the 4 randomly selected sites in the littoral zone on each event, how do I calculate the Euclidean distance when the 4 littoral sites are not the same for each sampling event. Would it be acceptable to calculate 4 Euclidean distances (even though the sites are random) and calculate the mean Euclidean distances with SE for the 4 littoral sites? Would calculating 16 possibe pairs of Eucli. distances suffice?

is it possible to conduct multidimensional scaling (MDS) in Minitab

I have word association data and would like to work out potential relationships between categories. I'm envisaging output that shows which categories might sit together in space and which are far apart. However, I am not particularly handy with SPSS so have no natural instinct for the kind of analysis I should be doing. Someone suggested multidimensional scaling but I'm not sure this would work for me. Each of my participants have a number of categories assigned to them (because of the words they stated), they have not specified how similar or different they think a number of categories are. I hope my question makes sense. I would be very grateful for any input. Thanks, Katharine

I need it for my dissertation. Thank you.

You can use quantitative variables in an NMS (non-metric multidimensional scaling) ordination & then see the r & r² values of each variable on each axis. But what about categorical variables? What statistical/numeric output can you use to describe trends of the categorical variables by which objects in such an ordination are grouped?

Also, the program's text read-out for an NMS run says "p = proportion of randomized runs with stress < or = observed stress". Is this the p-value (probability-value) of statistics? What can you say about these p values? Each one is associated with an axis, so how would you describe the relationship between the p value, the axis, & the objects within the ordination?

I am working on assessment of water quality of different 04 estuaries.

I have done statistical analysis in SPSS. I need to know how to interprets the PCA during the low tide and high tide condition!!!

our study location around many of industries and domestic wastage dumping in these estuaries.

  i am writing a research paper so i want to add on the paper..

if any have idea than please guide me.

Dear friends,

                Presently, i am involved in a project to assess the diversity of northeaten indian ocean deep-waters. I as used PRIMER v6 for plotting pca of diversity data of prawns in that region. i am uploading the plot, how can i interpret the plot interns of diversity of prawns in the particular region.

Many thanks in advance

Sileesh Mullasseri

I would like to learn more about multidimensional scaling. How to use multidimensional scaling? Where can we use it? What is the procedure to perform multidimensional scaling? Is there any literature support for multidimensional scaling? Kindly update some details.

I'm exploring a species database to do a cluster analysis among samples (Q mode) and I would want to know if possible the use of one dummy variable in this case.

I sampled a community of fungi in different treatments and I would like to assess the effect of treatment in species composition.

I ran a NMDS and a global test with adonis (vegan) and I have found significant differences  among the treatments (p-vaue < 0.001). I have not been able to find any way to run pairwise contrast to know which of the treatments are significantly different. 

I am looking for books focusing on Multivariate Statistics for ecological studies. I get really confused grabbing some commonly used techniques such as cluster analysis, factor analysis, multiple regression, multidimensional scaling, principal component analysis, canonical correspondence analysis, analysis of similarity, similarity percentage analysis.

Please suggest me book/s where above statistical concepts are clearly discussed with relevant example/s (not too much verbose). 

Thanks in advance. 

Hi Everyone,

I'd appreciate your advise on a validated survey instrument used to measure cognitive dissonance (preferably in Management research). Any other survey instrument besides Menasco and Hawkins' (1978), Sweeney et al.'s (2000), Mishra and Kumar's (2016) cognitive dissonance measures.

For instance, while Menasco and Hawkins (1978) measured cognitive dissonance arousal, Sweeney et al (2000) measured cognitive dissonance purchase/marketing, Mishra and Kumar measured emotional dissonance in lieu of cognitive dissonance.

Sweeney, J. C., Hausknecht, D., & Soutar, G. N. (2000). Cognitive dissonance after purchase: A multidimensional scale. Psychology and Marketing, 17(5), 369-385.

Mishra, S. K., & Kumar, K. K. (2016). Minimizing the cost of emotional dissonance at work: a multi-sample analysis. Management Decision, 54(4), 778-795.

Menasco, M. B., & Del. I. Hawkins. (1978). A field test of the relationship between cognitive dissonance and state anxiety. Journal of Marketing Research, 650-655.

Thanking you in advance.

Tobi

Hi. I am working on avifaunal assemblage of an estuary and I have 3 seasons data across 5 different habitat types. I used nmds ordination to look for any gradient in the assemblage across the habitats but in both 2-D and 3-D ordinations my stress values were high, almost equal to 0.3.

In such a case, reducing the number of species (using the most abundant species) would lead to a better ordination of the habitats or do I need to transform the abundances of all species before computing the dissimilarity matrix? I tried the second option using fourth root transformation, still the stress was higher than the non-transformed data. Or do I need to use a different ordination technique or a different dissimilarity matrix (I used Bray-Curtis)? If so please give some suggestions. (I use PAST and R for my analysis). 

I am trying to visualize trends of  particulate matter considering, at the same time, several spatiotemporal aspects such as sampling sites, seasons, daytime vs. nighttime, I want them all to be visually illustrated in on big graph

What are the advantages of doing multidimensional scaling analysis in MATLAB?

Hello,

I am studying wether the benthic invertebrate community could be used to estimate the variation and duration of hypoxia in coastal Baltic Sea. I have around 150 sample points for invertebrates and oxygen levels from the past decades for the sites. As a start, I would like to study the species composition.

From what I've understood Bray-Curtis dissimilarity is usually used in this kind of situation. I started by doing a bray curtis clustering and just for comparison a clustering with euclidean distance (both on sqrt-transformed values). I labeled the cluster groups and plotted them in multidimensional scaling, along with the numbers indicating the oxygen state.

After comparing the euclidean distance mds plots to the ones with bray curtis method, it seems that the groups formed with euclidean distance match better to the oxygen values.

Do you have some idea why euclidean distance seems to work here better, and wether it is acceptable to use it in this situation? There is samples (poor oxygen ones) that have only one or two random species. The whole dataset has around 25 taxons in the data. There is 5 species that are dominant in the samples. In the graphs different kind of plots are different clustering groups, numbers are oxygen groups (1= poor, 3= good, second number for variation 1 = few, 2 = much).

If you have some comment on this I would be glad to hear it.

Thank you for your help!

R

Can anybody give me an advice on how to calculate an MDS (Multidimensional Scale) in SPSS based on a corellation matrix (Pearson)? I don't know how to transform correlations into distances. I would be thankful for any advices.

Best Regards

Benedict

This is for my multidimensional scaling ordination and canonical correspondence analysis

I'm trying to conduct a multidimensional scaling analysis on a dissimilarity matrix using SPSS. I am having no problem importing matrices that are up to 200x200. Over that amount however when using the automatic import wizard to import the data as a delimited text file it limits me to 200 columns (not rows) as it assumes these to be variables.

Is there anyway around this - other than not using SPSS?

Secondly, when I come to plot the 2D data's  new coordinates using SPSS's scatterplot plot creator, I don't seem to be able to associate more than one variable to each point. For example I can display information for genera or collection location (as either colour or symbol shape) but not for both  e.g. red, star vs. an red circle (Porites, Pacific vs Porites, Atlantic).

Is this possible in SPSS or should I stop fighting the urge to go back to some plotting specific software like sigmaplot or r.

Many thanks, 

The quality of reduced dimension in classical MDS can be measured by shepard plot. Is there any other techniques such as Shepard plot to measure the quality of reduced dimension obtained from LLE (Locally Linear Embedding)?

Dear researches:

I am analyzing different environmental factors that affect to avian communities in 64 urban and periurban areas. The question is that I have a good fitting with Redundance Analysis (ANOVA p<0,001) and more of 85% of the constrained variability in only two axis. A Canonical Principal Coordinate Analysis (CPCO) give similar values with binomial distance. Finally Non-metric Multidimensional Scaling (NMDS) give a perfect reconstruction (r=0,99) of binomial similarities using only two axis. The problem is that when I use envfit function of R (vegan) my results give very different environmental factors associated in the three analysis. Also I had used Principal Curve Analysis that give me a r=0,54. In this case, only a factor is associated to the curve and this is completely different to the other ordinations. In my opinion this ordination is erroneous. Apparently the best ordination is the NMDS but this is more an intuition based in my previous knowledge on bird species and studied areas. What criteria is the best? ANOVA? Mantel test between distances and predicted similarities for CPCO and NMDS? I am confuse regard this validation problems but I continue thinking in a general methodology. Many thanks

I'm analyzing family planning data in Indonesia. I want to explore the perception of potential client about contraception and their attributes. I want to understand further on how to make a statistical interpretation of multidimensional scaling. Thanks

I am working with classical multidimensional scaling and the configuration matrix for start the iterative process is ill-conditioned. For that reasen I am looking for methods to transform this matrix and obtain a better condition number

The aim is to compact a material to obtain a similar density. The scaling factor(s) should account for side effects between the smaller mould and the larger mould.

I'm analyzing data from a study, where Multidimensional Scale of Perceived Social Support (MSPSS) is included. It seems that there are scores recorded for 12 items, but I don't know how to calculate the score. Do I just take the mean of 12 scores?

I searched online, and found the following statement:

I'm wondering if that means rating from different individuals. Since I only have one set of scores for each subjects, can I assume that's "global perceived support"?

Good to have an agreement on ubiquity of 1/f scaling, I also am of the opinion experimental control is essential (see Hasselman, 2013). I would add that direct confrontation of theoretical predictions is crucial as well:

"In order to advance scientific knowledge about scaling phenomena in living systems a program of strong inference that aims to produce closed theories of principles is needed. In order to reach this goal, empirical inquiries need to go beyond describing scaling phenomena in different populations in the context of impaired performance or pathology (e.g., Goldberger et al., 2002; Gilden and Hancock, 2007; West, 2010; Wijnants et al., 2012a). Several recent studies reveal scaling phenomena can be brought under experimental control, which is essential for a program of strong inference (e.g., Kello et al., 2007; Wijnants et al., 2009; Van Orden et al., 2010; Correll, 2011; Holden et al., 2011; Kuznetsov et al., 2011; Stephen et al., 2012). The diverging theoretical predictions examined in most studies reveal that the observed waveforms are more likely to originate from interaction-dominant complexity than from component-dominant mechanics (also see Turvey, 2007; Kello et al., 2010; Diniz et al., 2011)."

From: http://journal.frontiersin.org/article/10.3389/fphys.2013.00075/full 

At least these articles revealing experimental control over scaling exponents should have been discussed:

Then, there are many more  studies that make risky predictions or directly confront two or more competing predictions (in fact, all predictions in 1/f studies are more risky than mainstream, because they concern interval predictions and not merely > 0). In any case, they do much more than 'just' show another case of 1/f noise in some population.

All the best,

Fred

I am not able to interpret how we are getting these values in MDS either positive or negative for a attribute for all the three dimension . 

I am doing research in measuring the subjective response to interior sound in automobile, and then the subjective response which is the SD rating is included in MDS analysis, since the raw data of SD rating is not in the required format for MDS,  I really wanted to know how the way to do it.

For multivariate analysis, I would like to understand the differences between CCA and NMS.

A species matrice is analyzed in vegan (R statistics) with NMDS. The nonmetric multidimensional scaling technique is not using environmental data for ordination. Anyways the literature is plotting environmental arrows into the diagrammes. Stackoverflow discussions say this is not the right way. But it would be better to plot weighted averages of species or sample scores. What do you think?

I would like to find a correlation in the dataset represented by three dimensional vector field (35x observed displacement vector field). I thought about the PCA analysis, but I am not sure how I can properly construct the covariance / correlation matrix on the given dataset. Please, is PCA a good way to start? I appreciate any suggestion. Thank you, .

Petr Henyš

Is there anybody who has seen implementation of MDS-MAP? I have done a Classical MDS implementation, but I am unable to find the MDS-MAP implementation.

I collected samples in streams and, in each of them sampled 10 riffles and 10 pools. To check for differences in the composition of the fauna I performed a ANOSIM two-way crossed, the factors were streams and mesohabitats; Another analysis that was performed to verify the difference between mesohabitats within each stream was the beta PCoA. In both analyzes I used the data of abundance and the Bray-Curtis coefficient.

However, in the analysis of similarity, a difference was found in the assemblage composition between streams and mesohabitats. And beta PCoA showed no difference between mesohabitats of each stream.

I'm unable to understand the differences in the results. What factors may be related these divergent results?

Andrés Baselga; Patricia Koleff; Adriano Melo

Dear vegetation community and multivariate users,

I liked discussions here related to selection of different ordination methods. In this connection I have a query related to using NMDS score value as environmental variable and regression against a response variable. Bot NMDS axes were taken as latent environmental variable and done regression against response separately. I have a strong reason why I did that. Because sampling vegetation is hard to distinguish a clear controlling gradient. 

I shall be grateful of some feed back and discussions related to this topic.

I am using all of these analyses in R and vegan.

Thank you all in advance.

Chitra Baniya

Hello, everyone! I have a data of community composition, and I do a NMDS ordination to see how different treatment groups distributed, and after that, I want to make some analysis to test whether two groups are different for their community composition and which species contribute more to the difference. I know that software PRIMER can achieve this goal, but PRIMER is not a free software, and its function SIMPER has been criticized for some problems (see Warton et al. 2012). So, anyone who can give me some suggestions to finish this task in R software? Thanks in advance.

Is there anyone who knows about this methodology. I am considering employing this methodology to analyze speech perception of sounds. Thanks a lot!

Minghui,WU from Radboud University Nijmegen /Shanghai International Studies University

Testing dimensionality of scale I have found out that it differs based on original

For one country it is unidimensional, but for another - it appeared to be multidimensional.

What can we conclude out of this about usage of the scale? Is it bad or good or nothing? 

In general what is better (or when) - unidimensional scale or multidimensional scale?

In my research I need to use automatic scaling in both dimensions: horizontal and vertical. I found some works regarding horizontal scaling performance in cloud providers and hypervisors but nothing about vertical scaling overhead. I need scaling the VMs at run time.

Can anyone assist me and tell me if they are aware of any advances in multidimensional data analysis in the behavioural sciences?

What's the procedure to do multidimensional scaling in spss, and in which case shall I do it?

Is there somebody with experience using Multidimensional Scaling for Java (MDSJ) libraries to produce 3D diagrams from dimensional / dissimilarity matrices or using another java open source or free java library?

I am using the java library mdsj.jar from http://www.inf.uni-konstanz.de/algo/software/mdsj/ version 0.8 2008 (there is a newer one from 2009) but I am having problems with verifying the results against results provided using R.

I know Multidimensional Scaling uses only a distance matrix, but Self-Organizing Map requires coordinates of points in the original space. What are some other dimensionality reduction techniques, such as Multidimensional Scaling, that need only a distance matrix rather than point coordinates?

Recently when I tried to visualize similarities between different groups in a social network, I discovered an interesting phenomenon which I couldn't explain. Can someone give me some insight on this?

The story is as follows: Users can create different groups on our social network website, and we'd like to visualize the similarities between different groups in 2D. We use the group members and their page views within the group as the feature vector of each group, and the similarity between groups is computed as the cosine similarity between their feature vectors. I used multidimensional scaling ( cmdscale in R ) to reduce the data into 2D and visualized the data.

The result of the MDS is points lined up with some lines orthogonal to each other. Can someone explain why this is happening?