Teaching generations of PhD students in our lab how to do stats. You are a legend here!
@kamcgnt
10 жыл бұрын
Hi Marta. Thanks! I am pleased the videos are helpful.
@sakottomitav
2 жыл бұрын
Sorry to bother, If someone can help me. I have 3 biotic matrix of 3 different basin ( abundance of Macroinvertebrates) is there a way to analyze the 3 at the same time ( do for example a Cluster). I don’t know how to generate a big matrix of the 3 in the program. Thanks in advance
@lorenzioscf6840
4 ай бұрын
Hi can I send you a dataset and you make a pca or permanova? I would be nice 🙂
@lorenzioscf6840
4 ай бұрын
Hi can I send you a dataset and you make a pca or permanova? I would be nice 🙂
@valquiriagarcia8512
7 жыл бұрын
Hi Keith, I'm analysing fish larvae distribution and in some sampling sites my abundance is zero. To create a resemblance matrix and do some analysis in primer, do I need to delete the samples (rows in my spreadsheet) where I have these zeros?
@kamcgnt
11 жыл бұрын
Short answer: no. Long answer: also no! I have never used that option and, unless I had a large data set and few missing values, probably never would. I note that it assumes multivariate normality and that looks to be more important with this option than others. Can't help. Sorry!
@getrudetshithukhe5259
3 жыл бұрын
Hello. I am a primer beginner and have been given a data set to analyze, and told to do so in primer. I have tried to put it in the right format but obviously got it wrong. Confused: Any time I try and add some of the worksheets to primer I get "Bad number Row 2 Column 2" I cannot figure out how to correct this. I have re-arranged everything, but the message always pops up when I import that 1 specific sheet. Thanks in advance :)
@kamcgnt
3 жыл бұрын
Sorry, I am retired and it is at least 3 years since I have used PRIMER. The usual problem is non-numerical data in the body of the worksheet. Labels must be after the last row or column, separated from the data by a blank row or column. (See other videos.)
@kamcgnt
3 жыл бұрын
Missing data should be blank.
@getrudetshithukhe5259
3 жыл бұрын
Thank you Keith, I removed non-numeric data to the last rows after the blank row and it worked, thank you so much
@dongdong19760420
10 жыл бұрын
Thank you so much! It is very useful!
@kamcgnt
10 жыл бұрын
Great. Thanks.
@MiltonJinner
9 жыл бұрын
how about w statistics (w, clarke, 1990) with primer 7.
@slashsharma
11 жыл бұрын
Hi Keith, can you tell me why I get "undefined" values for the resemblance matrix? Thanks
@polgy2
6 жыл бұрын
I have a question: if PERMANOVA finds a significant difference in multivariate location among the levels (e.g. in a single factor design), but also PERMDISP finds a significant difference in dispersion among levels, I read (Anderson 2005, PERMANOVA manual) that I should check pairwise tests with both analyses to determine the nature of the difference between each pair of groups, whether in location, spread, or a combination of the two. Should I just compare the results of the two analyses for each pair, e.g. if the comparison A-B is significant in both PERMANOVA and PERMDISP, then there is difference both in location and in spread?
@kamcgnt
6 жыл бұрын
Follow advice from Anderson who is the expert and understands it better than I do. But my understanding is that differences in dispersion can cause significant effects in PERMANOVA but conclusions about location can be tricky.
@CH3COOHpt
10 жыл бұрын
Hi there Keith. First of all, I would like to congratulate you on you easy-to-follow videos, they have been very helpfull to me and other. I would like to ask you if there is a way in PRIMER (or +PERMANOVA) to do something similar to BEST or DistLM, in which we can find the best set of environmental variables that explain variation in biological data, but to build predictive equation/model. Thank you in advance
@kamcgnt
10 жыл бұрын
Hi and thanks! BEST is in PRIMER 6 and DistLM is in PERMANOVA. Be aware that menu options depend on what type of sheet you are looking at. You need to select a data sheet to see BEST (on Analysis menu) but select a similarity matrix to see DistLM (on PERMANOVA menu).
@kamcgnt
11 жыл бұрын
Thank you. Normalisation gives all variables same mean (0) and variance (1) but does not alter the shape of the distribution. According to the PRIMER people, transformation may be helpful for variables which are strongly skewed. They are the experts; I follow their advice.
@ClaudiaTamezisalive
11 жыл бұрын
Hi Keith, Do you have any tips for why the "missing" tool might be generating values that are all the same when the tool is expected to generate unique values for missing values?
@raiynenald
11 жыл бұрын
Hello..thank you for this great video tutorial..I just have a question about the pre-treatment environmental variables. Why do you have to transform first the data in HC where in fact you also did normalization of the entire variables including again the HC..would it be possible to directly normalize all the variables?
@kamcgnt
11 жыл бұрын
Hi. That is usually due to missing/blank/empty cells in your data sheet. Some functions work with missing values; others don't.
@alejandrodonnangelo7001
9 жыл бұрын
Does anyone knows where to find the primer permanova manual? I do have doubts about permanova results reading? I dont really know the meaning of some results and I looking for som help. Thanks in advance. .I really appreciate any help!
@kamcgnt
9 жыл бұрын
Alejandro Donnangelo You can only get the PERMANOVA+ manual from Primer-E: www.primer-e.com/index.htm.
@elismo3521
7 жыл бұрын
Excellent explation! Thank you very much!!!!
@polgy2
6 жыл бұрын
Great contribution! Finding the manual of PERMANOVA+ is a nightmare and your videos are the only piece of information I found: THANK YOU SO MUCH :-)
@kamcgnt
6 жыл бұрын
Happy to help.
@cynssc07
12 жыл бұрын
Hi! What are your version of Primer? Thank you for you kind!
@MiltonJinner
9 жыл бұрын
Does anyone knows where i can find the program primer 6 + permanova , thank you all
@velutharanish6385
9 жыл бұрын
thank you for uploading this video...very useful to me and also i have one doubt about why u r taking log from hydrocarbons....
@kamcgnt
9 жыл бұрын
Pollutants are often log-normally distributed, so take log to give more normal distribution.
@juancarloscarrasconavas-pa9493
4 жыл бұрын
Muy bueno, me ayudo mucho
@Owlli
9 жыл бұрын
Great tutorial - have referred to it many times over the last two years. Any idea as to why PRIMER might be opening an excel sheet and interpreting the column names as factors? I have everything set up just like in the "Bio" sheet, samples as rows, and a blank column in between my count data and factors. Yet, when I open "edit factors" it shows the "label" column, then all of my species columns. The intended factors aren't even present ...
@kamcgnt
9 жыл бұрын
PRIMER is very particular about the format of the Excel file. I have the first row as the variable and factor names and the first column as labels. Data must be numeric. And when opening the file, uncheck "Title", check "First column as labels" and check "Samples as rows". It sounds like you are doing all this so I don't know what is going wrong.
@Owlli
9 жыл бұрын
Keith McGuinness Um ... yup, I have it all like that ... I'll keep working on it and let you know if I figure it out. Thanks for such a quick response!
@leonym2214
3 жыл бұрын
Hi Keith, I will be using PRIMER working on a table of IRI values (Index of Relative Importance) which takes into account frewuency, number and size of a given set of microplastics I found in my samples. Do you think this can be added to PRIMER just like abundance data? I am not even sure if I can use PRIMER with this kind of dataset..
@kamcgnt
3 жыл бұрын
Yes. (I am retired.) Can be either ENV or BIO, depending on objectives.
@leonym2214
3 жыл бұрын
@@kamcgnt Thank you so much for even answering out of your retirement :D
@kamcgnt
3 жыл бұрын
@@leonym2214 You're welcome! (It was an easy one, that's why.)
@njfl4
11 жыл бұрын
very good video! it helped a lot! thank you
@sginther12
8 жыл бұрын
Hi Keith, I am would like to create a Bray-Curtis dissimilarity matrix for my normalized data. My normalized data has negative values, and when I attempt to create a B-C matrix I am told that I cannot do so (because of the negative values). Do you have any explanation for why PRIMER doesn't let me create a dissimilarity matrix from negative values? Thanks!
@kamcgnt
8 жыл бұрын
+Sam Ginther Bray-Curtis is for count (or percentage) data which is not usually normalised. It may be standardised, or transformed, which will not give negative values.
@sginther12
8 жыл бұрын
+Keith McGuinness Thanks for the quick response, Keith. Ok. What is the usefulness of standardizing vs. normalizing (according to PRIMER's definitions of the 2 terms) data? I've been searching through the PRIMER manual and online, but can't find a straight forward answer. Essentially, I am trying to create a similarity matrix out of both percentage and density data - both types apply towards habitat parameters (% cover of algae and substrate, and density of other algal species). Either way, it appears that I should be standardizing these habitat variables so that I can create a matrix, correct?
@kamcgnt
8 жыл бұрын
Standardisation converts counts to proportions if sample size volume. Normalisation is for when you have different kinds of variables.
@ninoposadas5582
7 жыл бұрын
Can I perform any analysis on unlicensed Primer7?
@kamcgnt
7 жыл бұрын
I only answer questions about legal versions of PRIMER. The limitations of the trial version are available from e-PRIMER.
@stefankazakov5302
8 жыл бұрын
If we import as a sample data (calculated diversity indices, numerical or biomass ratio between the different taxa), in regard of looking for relation between indices and rations with the measured env. variables, how should we treat the sample data in data type tab: abund, biom, env or other? and even further in pretreatment process, because it is somehow quite transformed biological data already?
@kamcgnt
8 жыл бұрын
+Stefan Kazakov I would use other but the main effect of stating this when importing data seems to be for PRIMER to pre-select (or recommend) appropriate options. As you can usually override these, it is not a big deal. For pre-treatment, PRIMER analyses are said to be more reliable if data are not strongly skewed, so that's where you look at transformations.
@stefankazakov62
8 жыл бұрын
+Keith McGuinness OK. Thank you!
@kamcgnt
12 жыл бұрын
Thank you. PRIMER V6 with PERMANOVA+ add-in.
@felipemolina1384
6 жыл бұрын
Can I mix biomass and abundance results in the biological data? Or would be better to work with similarity data? Im dealing with this, someone can help me please!
@kamcgnt
6 жыл бұрын
I would not do that because they are not comparable. I would try to change the abundance into biomass, although doing the reverse (biomass to abundance) would be better. I have not used biomass myself or read much about it.
@felipemolina1384
6 жыл бұрын
Thank you so much Keith
@f4lc0n88
8 жыл бұрын
Hello Keith, First at all thanks for the tutorial. I would like to ask you how can I change a variable into a factor in PRIMER. I have an experimental design with 2 preset levels of temperature, nutrients and light and when I import my data and go to Edit -> Factors there is nothing there and I am really getting headaches because I cannot make PRIMER to change those variables into factors. THank you very much.
@kamcgnt
8 жыл бұрын
To make this easier, you need to setup the file structure in Excel before doing the import into PRIMER. If columns are variables, and rows are samples, then the first column should have labels for the rows/samples. The next columns have the data, either environmental or biological, THEN there should be a blank column: this is critical. After the blank column, have the columns with the factors, with rows/samples labelled with the appropriate levels for each factor. The Edit-->Factors options in PRIMER are there for adding factors in, after importing data; for fixing errors identified after the import; and for doing other things such as combining factors.
@f4lc0n88
8 жыл бұрын
Thank you very much Keith, the BLANK COLUMN was the key, now it is working perfectly. Have my upvote :)
@tajamulislam9565
3 жыл бұрын
How to download primer and Parmanova software
@robertgillan7653
5 жыл бұрын
Hi Keith, love the video, do you know how to go about doing this in R?
@kamcgnt
5 жыл бұрын
I recommend "Numerical ecology with R" (I think that is the title) as it shows how to do everything.
@loosealliance
8 жыл бұрын
Very handy description. One question if I may; in your example if you then repeated the sample across a number of years how would you setup the spreadsheet for import into primer? Would you had another column year?
@kamcgnt
8 жыл бұрын
+loosealliance Yes. This would enable, for instance, ANOSIM or PERMANOVA including YEAR as a factor. In PRIMER 7, which I will be looking at soon, ANOSIM is greatly enhanced and capable of multifactor analyses like PERMANOVA.
@loosealliance
8 жыл бұрын
+Keith McGuinness Thanks for the reply. I had a few problems getting it to load into primer but I think I may have it now(ill see how I go haha). You tutorials are very useful so far. Much appreciated!
@kamcgnt
8 жыл бұрын
+loosealliance Excellent. Thank you. PRIMER is fussy about the file format.
@kamcgnt
12 жыл бұрын
Thank you.
@dexterdelacruz660
8 жыл бұрын
Hi again, Keith. in your video you transformed the bio data prior to the Bray-Curtis analysis. I read papers that did the same. However, many of the manuals and blogs suggest that using Bray-Curtis for abundance, transformation is not necessary because the analysis is "self-transforming". What is your comment about this? Many thanks again.
@kamcgnt
8 жыл бұрын
Hi. I don't know what 'self-transforming' means in this context and the authors of PRIMER, and the manuals that go with it, clearly discuss transformation of the biological data. In any case, transformation certainly affects the results as you can see in my video 'Second stage analysis in PRIMER': kzitem.info/news/bejne/kY2f3XVohoCWiKA. It's a 'user choice' situation and this partly depends on the purpose or aim of the analysis.
@dexterdelacruz660
8 жыл бұрын
Thank you very much again, Keith.
@kamcgnt
8 жыл бұрын
I had an idea about 'self transforming'. If the biological data are standardised -- effectively becoming percentages or proportions -- then the transformations may well have less effect (but I have not tested this yet). This is because differences in absolute abundance are removed and comparisons are then on relative abundances and these are constrained and may be less variable than the original absolute abundances.
@dexterdelacruz660
8 жыл бұрын
That is exactly the argument of some of the people in the forums I read. I am playing around now how the results of my data will change if I transform or not. Thanks again, Keith. :-)
@kamcgnt
8 жыл бұрын
Hi. New video looks at transformation on raw and standardised data.
@cynssc07
12 жыл бұрын
And perfect video!
@janevanessaa3523
4 жыл бұрын
hi. is there any way i can get the primer free version?
@kamcgnt
4 жыл бұрын
Sorry. I am retired and no longer have the resources.
@janevanessaa3523
4 жыл бұрын
@@kamcgnt oh too bad. Thanks for the videos it help alot!
@dexterdelacruz660
8 жыл бұрын
Thank you very much for this very useful video. May I know how to deal with missing data in biological and envi data?
@kamcgnt
8 жыл бұрын
To my knowledge, in the past, multivariate analyses have usually not been able to work with missing values: in other words, the entire sample would have to be excluded. This is changing and PRIMER 7 and PAST 3 are both able to handle missing values in some situations. (I do not know about other software.) For example, PAST 3, for PCA will use either 'mean value imputation' or 'iterative imputation' if there are missing values. PRIMER 7 has a 'missing tool' which 'Replaces missing values in a data worksheet. If the data is biotic then missing values can be replaced by zero. Commonly zero counts are originally entered as missing values. For environmental data there is the EM (expectation maximum likelihood algorithm) which assumes a multi-normal distribution model for the data (so is unsuitable for biota!).' In any situation in which there are missing values, you would need to check that the app, if it worked with missing values, dealt with them appropriately. I would probably run the analysis with the missing values, then exclude the sample with missing values and run the analysis again, then compare the two. If the results with, and without, missing values were very different, I would have to think carefully about how to proceed.
@fgo77
8 жыл бұрын
To follow this up, can SIMPER be calculated to give meaningful results with missing data?
@kamcgnt
8 жыл бұрын
To be clear, missing data here means some values for some samples are missing: it does not mean entire samples are missing (that would be referred to as unbalanced). According to the manual, PRIMER 7 automatically corrects resemblance calculations for missing values, if it is required. This appears to refer to all situations where resemblances are calculated, which would include SIMPER. Beyond that, and for PAST, I cannot say.
@kamcgnt
7 жыл бұрын
I don't know the answer to this and it would depend on how much data was missing. Clearly, if some values are missing the results would differ from those obtained if the values were not missing. In such situations, you may want to experiment: delete the entire row/column and see how results are affected. If an 'important' species (or sample) was removed then the results are going to be affected. 'Important' here simply means in terms of the SIMPER results.
@kamcgnt
7 жыл бұрын
Missing and empty samples are different. An empty sample (I assume) is a sample that was collected but had no species (I have seen this in highly polluted or disturbed situations: we collect a marine mud sample but there are no animals in it). A missing sample occurs if the sample is not actually collected: in this case, we don't know what the counts would have been. It is generally not appropriate to enter zeros for the species counts in this situation. PRIMER/PA vary among analyses in how they handle missing values. Zero is fine for an empty sample (sample collected but no animals found) but not for a missing sample.
@lizevandermerwe6738
6 жыл бұрын
Thank you for taking the time to make these videos, Keith. I am sure that I am just one of many people worldwide using these regularly - watching and rewatching them, as one would read and re-read a stats textbook. A question from an academic perspective - is there a publication that I can refer to when citing these methods in a scientific, peer-reviewed publication?
@kamcgnt
6 жыл бұрын
It is usual now to cite the user manual. Clarke, KR, Gorley, RN 2015 PRIMER v7: User Manual/Tutorial. PRIMER-E: Plymouth. That's for PRIMER. For PERMANOVA, the user manual for it.
@annaheii_864
4 жыл бұрын
Thank you, this really helped me
@kamcgnt
4 жыл бұрын
Glad it helped. ☺
@gailclaus
Жыл бұрын
You are the best, your explanations despite the language barrier are very understandable and I was able to analyze my data. Thank you very much for taking the time to make the videos, I felt lost until I found them, I am very grateful.
@kamcgnt
Жыл бұрын
Am now retired and don't do this anymore but I am pleased that the video helped. Good luck.
Пікірлер: 89