Supervising

This document is a continuously updated manual supposed to help students (doctoral and master students) in parasitology (but probably of use for others as well) under supervision in answering frequently asked questions (FAQ). It has been created to ease the work for both supervisors as well as students.



SUPERVISOR -STUDENT ROLE
[author: Brian Lassen, version: 7th October 2013]

Many misunderstandings and frustrations arise from misunderstandings of the supervisor – student role. 
To help clarify, here are a few points:

  • A supervisor expects some level of independence from the student, but has to be aware of the personal needs of each student.
  • A supervisor is obliged to provide assistance regarding research and project problems.
  • A supervisor is obliged to try and provide a stable working environment for the student to excel in. This includes assisting in finding finances for student salary, projects, and education.
  • A student is expected to develop his/her own ideas and formulate hypotheses and projects around them.
  • A supervisor should be available to the student, but the student should try to find solutions themselves and possibly bring them to a discussion.
  • Discussions is strongly encouraged on the research topic but also science in general. 
  • Both parties should respect the need for private life, and aim not to load each other with work other than mutually agreed. This can easily be avoided by respecting deadlines and keeping realistic agreements. 


Tools:
Oxfords Learning Institute: Stages of the doctorate


WORK GUIDE FOR WRITING AND PREPARING A SCIENTIFIC WORK PLAN
[author: Brian Lassen, version: 17th January 2014]

The purpose of this guide is to give students who are planning a project a check-list for designing a successful study, avoid loss of work, and ease the following writing process when presenting the results.
Every project should have its own project plan. The most common study types are: cross-sectional (a snapshot picture of a situation), cohort (study over time), case-control (difference of an action to a unchanged status: eg. drug treatment), case report (interesting finding), and meta-analysis (summary of studies).

Idea
An original idea may not be difficult to get, but formulating it and restricting it to become a project is. Usually ideas are foggy and involve ”something about...”. The more specific the better. It is easier to expand on an idea than restrict a started project that turns out to be massive in work load. The main effort in the idea phase is: limits!
  • What is your idea (2-5 lines)?
  • Who/what is it relevant to (who will benefit)?
  • How is the idea fitting with the current knowledge (find min. 5 references and write them down stating how it connects with your idea)?
Hypothesis
Each project should have at least one hypothesis. Spend a lot of time on refining your hypotheses! The clearer you are about this step before proceeding, the better the project!

A hypothesis is a question that can be answered with data. It can be written as a formula (X = Y or what is the number (N1) of animal species T (eg. sheep) in location F (eg. Estonia) compared to total population N2?). Some hypotheses are confirmed/rejected by showing there is a difference (p-value<0.05, showing you are less than 5% chance of being wrong) while others are indicated by an accuracy of the obtained number (such as a prevalence for example).
  • What is the main hypothesis?
  • What are the secondary hypotheses (if any)?

Units: Science is observations of measurements. List what your ”units” measured are (pigs, fields, E. coli colonies, farms). Make sure the units answer your hypothesis. Each hypothesis has one unit (but if having several hypothesis' the same unit can be used). The unit is ”what you examine” and determines how many samples you need to make a reliable result based on the method you choose to use.
Note that two animals raised in the same farm may have many risk factors in common and thus be more similar to each other than two animals from two separate farms.

Measurements: Once you know your units, define what you are measuring from each unit (grams, gene sequence, ecological: yes/no...)! The measurements make it possible to part your units in groups that can be compared to see if there are differences (the hypothesis).
  • For each hypothesis, list the unit and the measurements associated
  • Write down additional information that needs to be recorded that will be helpful but not necessarily used in the dataset (date sampled, date sample was examined, contact persons of farm, address, earmark number of cow...)

Methods
The methods you need are the ones that can give you measurements accurate enough to prove your hypothesis and allow it to be repeated by others.
  • Identify which method you are going to use. Read how it is used in literature and product manuals.
    Any reported problems with using the method that is required consideration before use (availability, cross reactivity, access to analysing apparatus, special working environment needed...)?

Randomization and blinding: Some studies have a reading bias. For example knowing which is your group A and group B may make you favour one over the other. To avoid subjectivity, blinding the study is needed (a code system). To make sure your investigation is not going to be influenced by ordering your samples/sampling (farm1, farm2, farm3...) a randomization of samples can be made (mixing or ordering codes of samples in according to a random number list). Sometimes randomization can not be achieved practically (eg. loose heifers in large group impossible to catch systematically).
  • Will your method involve reading or sampling bias?
  • If yes, how and what methods will you in-cooperate to avoid or handle such bias?

Planning ahead: When planning an experiment it is a good idea to...
  • List what you need
  • Calculate how much is needed of each item +10-20%.
  • Check what is already available and order what is lacking in due time (min. weeks in advance).
  • Take into account shelf time of products.
  • Familiarize yourself with the method before using it experimentally.

Sample Size: In the planning phase it should already be known how many samples needs to be examined to get a result that can be statistically tested.
If your study is a sub-sample of a population you need to make a power analysis (predictability) of your sample size based on expected observations (estimated from literature), your methods sensitivity and specificity, and the population size examined.
If your experiment does not have a representative population (a laboratory experiment for example), you need to estimate how few samples you need to examine to find an expected difference in the results. The smaller the difference you want to detect and the lower the method can predict an accurate result – the larger the size. Also here a power analysis applies.
  • Describe your population or investigated samples.
  • Describe how your sample size is representing the population (%, power analysis). Add some extra samples to be sure (10-30% depending on confidence in power analysis).
Relevant tools: 
OpenEpi, Animal recording center: www.jkkeskus.ee, population statistics: www.stat.ee

Laboratory time: By trying the methods it is possible to estimate how quickly the analysis is going. After doing this, estimate if you will have to plan your experiment to account for too many samples accumulating at once.
  • Can you handle the samples in the time you planned?
  • Is what you are measuring affected by waiting to be analysed (hatching eggs, dying parasites...)?
  • How should you store samples to avoid losses?
  • Is someone else using the laboratory and tools at the same time?
  • Are you samples useful to somebody in the future (exotic, rare, first look)? Should you arrange long term storage of extra samples (eg. freezing of blood/faeces for genotyping)?

Working protocol: To better visualize what needs to be done, draw your project, step by step. Draw the cups, the shovel, the sample etc. Often this way you will find materials or problems you previously overlooked .
  • Draw what you are going to do in the laboratory, step by step.

Ethics
Some projects will involve contact with animals. If this is the case you need to evaluate (with your supervisor) if your study needs an approval from the Ethical Committee in Estonia. Use of laboratory animals, human sampling, and consent to share private information is likely to need approval. Sampling from animals (blood, feces, lavage, tissue...) is often a good idea to apply for approval even if the pain and danger to the animal is estimated to be minimal. If in doubt, apply. 


Statistics
Knowing beforehand what analysis' that needs to be done can make the difference of a successful or failing project, make the work more certain of reaching a publication, and certainly will reduce work load dramatically. To do this one needs to know A) the hypothesis, B) the units and measurements, C) something about statistics. As a rule – the better the study design the more simple statistics is needed (and the other way around) to get a reliable result!
If not confident in statistical analysis, consult a statistician regarding the hypothesis (recommended: Tanel Kaart or Toomas Orro).
For each hypothesis:
  • How will you group your data?
  • Do you need to randomize/blind your investigation (is there reading bias)?
  • What statistical method should be applied to analyse data?
  • By what criteria will you omit data from your dataset (proof of contamination, incomplete questionnaire, unidentifiable observation...)?
  • How will you record, store, and backup your data, and transfer data to a file readable by a program?
Free relevant tools:
R-statistics program, RStudio (user interface for R), OpenEpi, WinEpiscope, Survey Toolbox.

Funding and Budget
For a project to become reality it is important there is financial backup. Confirm with supervisor that it is possible to get what you need (in advance). Present the needs in an ordered form (excel), including the information of ordering and billing.
Before ordering confirm with your supervisor how bills should be handled when they arrive. Avoid letting your supervisor to take care of materials (ordering, receiving, price comparisons) 
Common items overlooked in budgets are: transportation costs, disposables (cups, gauze, slides...), special equipment (adaptors for needles, storage racks/boxes...), buffers, etc.

Human Resources
If you need to involve collaborators make sure you have a clear agreement and contact information. Respect you are taking other peoples time (even if for an important research project to their benefit). Announce your arrival and how long time you expect your visit to take or how much work load you expect giving to a person (analysis for example). Sometimes it is best to consult with them to evaluate the work load. Remember to thank for peoples’ time and keep them informed what you are doing.
  • Know who you need to help you.
  • Contact them in advance to let them know you need their assistance and negotiate for help.
  • If setting up a meeting with one person: have your questions prepared.
  • If setting up a meeting with several persons send an agenda beforehand of the meeting and if you need specific persons to provide something for the meeting, state so in the agenda. Be short and precise (bullitform) what will be discussed in the agenda.

Relevant tools: 
Meeting planner: Doodle.

Compiling Data
Order your collected data for the project in ONE excel sheet. Save it to at least three different locations. When working on it, always save with another name, for example add the date, and don’t delete all old versions – this is your backup if you accidentally mess up the file. First column is your sample codes (1, 2, 3,...). It is easiest if you assign a column with your group codes too (example, if the hypothesis is comparing observations in autumn and spring one code for column “season” would be 0=autumn and the other 1=spring). Make a separate file or work sheet with all the explanations of the codes. Remember programs may not be able to read letters as data and all information should be attempted to be converted into numbers. If you prepared well in your project plan your answers will already be in the form of numeric data (eg. a questionnaire can be designed to say (“<10 cm” (0), “10-20 cm” (1), >20 cm (2)) already in the design phase.
  • Compile data sheet with measurements in columns starting with sample number.
  • Transform non-numeric data to numeric data if needed.
  • Is there missing data or observations that need to be omitted?
  • List all your measurements in your excel sheet and describe them in a separate document (makes it much easier for a new person to use your data later).
Relevant tools: 
Dropbox (online synchronizing tool to avoid loss of data), LibreOpffice, OpenOffice.

Analysis of Data
Test your hypothesis as planned. 
  • Does the result reject or prove your hypothesis as set up?
  • How confident is your analysis (errors / p-value / confidence interval)?

Publishing
Retaining data (not publishing) is unethical if the study is any practical use or academic interest! In worst case not making new information public also means consequences (e.g. people or animal lives that could be saved will not be saved).
There are two kinds of publishing that both should be considered.
Internationally (peer reviewed paper): for scientists to know what you have done. From an ethical perspective one should consider when choosing a journal: is the work within the scope of the journal? What is the review time (if you need the article published soon)? Is impact factor important? Is it important the article is public accessible (to everyone or pay per view – an African university may never see your article if it is pay per view)?
Locally (popular science): to let public or target groups who will benefit from your work know what you have done. Ideally both international and local publications should be done.
Making data public. Once published the ethical aspect should be considered to make the data-set publicly available on a website (ethical because research is often paid by tax money and can be used in other contexts once created). The data-set should be uploaded with explanation of variables (see “compiling data”), a reference to your publication(s) and contact information to the principal investigator / author. Sometimes it is possible to add supplementary data to the article, in electronic form. Another place is ResearchGate.
Relevant tools: it will be increasingly difficult to keep track of literature. To avoid spending time on finding literature and citing it in the work, use a program for keeping track of the articles. Options is the free Mendley or Endnote.
Who are authors and who are acknowledged: Please read and follow the guidelines of the Vancouver Convention for justifying authorship. If some potential co-authors has only participated in 1-2 parts of the study they should be thanked in “Acknowledgements”. If you think potential co-authors in the acknowledgements should be given a chance to qualify for the position one option is to invite to become a co-author granted they actively take part in writing and editing the manuscript.
Project Plan and Time-line
Once you know the above it is time to break it down into modules.
  • Make a deadline for the finished paper.
  • Break down your expected time periods into “preparations”, “execution/collecting data”, “analysis”, “writing”. Add 10-20% time for every step (you will always be short on time).
  • Is it realistic? If not, where can adjustments/sacrifices be made?

Defending Data
When defending a work your work is critised by an opponent because of the concept of science is that is it tested by being falsifiable or not. This is the strength of science. Trying to prove a theory wrong (even your own) should eventually only leave the option of “being true” if you have examined all aspects possible. The opponents is thus not working to humiliate the work but to find flaws in the design that could be improved for further research in proving the hypothesis. Consider it as a brainstorming-process with you. You can/should therefore yourself (in the “discussion” part of the thesis/article work) point out errors and weaknesses that could improve the study of your hypothesis. Likewise, if there are similar observations (also in other animals or similar conditions) they can be drawn in as evidence for supporting your results.
Your success in a defense is weighed in the integrity of the work, meaning both systematic use of the scientific method but also honesty.

Relevant tools: 
Gapminder (mainly cohort data), Prezi (presentation tool)

Feedback
Reporting back to involved parties and interested individuals is perhaps more important than defending the data. If nobody knows the data exist it will not be used and the efforts are wasted. Do not expect people to pick up your work and act on it because it exists! Use your network and politely make it aware you have published something that might be of their interest. If governmental agencies should know about your information – you should contact them with a copy!
  • Make a list of people who contributed and who should benefit from receiving your work once completed. Spread on Facebook and other social media.
  • Always thank funding sources and people who helped you.
  • Self-reflection: think back what could have been done even better in the study - how your next project is going to be better.

OBTAINING EXPERTISE
[author: Brian Lassen, version: 12th January 2013]

Generalized view on what "expertise" means: To obtain an ”expertise” in a field, the international measure of such is 5 publications within the same area (EU funding applications for example).
Niels Bohr said "An expert is a person who has made all the mistakes that can be made in a very narrow field." Or in other words: acknowledge you have learned something, but stay humble.



WORK GUIDE FOR WRITING A SCIENTIFIC ARTICLE
[author: Brian Lassen, Estonian University of Life Sciences, version: 22st Jan 2014]



The purpose of this guide is to help students with scientific writing: a doctoral thesis/diploma work/journal article.
Choosing a journal: 
The journal you choose should be one covering your topic (read the journals scope & aim). Weigh how important you think your data is according to the impact factor. If your work is mostly of local relevance, a more local journal (eg. Scandinavian Journal of....) to publish in as many journals are focused on research relevant for the international community. Many journals take fees for publishing - avoid those! 

Relevant tools: 

Scientific language:
The purpose of scientific language is to be convincing, clear, and honest (make your point using your evidence). It is thus required to be precise and concise. Because most journals want to reduce the amount of space used per article and their readers lack extra time, the text should be as short (but clear) as possible. Practising – reading, writing, and editing - improves these skills.

Plagiarism is a serious ethical issue that should be considered. NEVER copy paste others or your own text when writing a new text.

Relevant tools:
Plagiarism Finder
Turnitin
iThenticate
WriteCheck 

Types of journal publications:
Scientific article/original paper (research project, explained in detail below)
Rapid communication (research needing fast publication due to novelty or importance)
Short comment/brief comment (research that do not qualify for a full research project and can be communicated in a concentrated form)
Case study/case study series (interesting and relevant observations, often associated with medical practice)
Letter (to the editor) (comments to the journal, and in some cases on general scientific topics)
Opinion (scientific knowledge or expertise communicated in a more narrative form)
Review (summary of advances in a field)

Title:
Should basically be your main hypothesis in an interesting headline form. Can also include the conclusion if one is confident enough in the results. 

Authors:
Rights for authorship and main author should be agreed BEFORE beginning a project. Exceptions can be made if the need of expert knowledge is noticed only after the planning (difficulties with statistical analysis, interest in genotyping of samples in other laboratory, etc.) that requires a considerable amount of work.
General rule of authorship: As a rule of thumb, an author must have contributed to at least 3 of the following: idea, design, implementation, conducting the experiment, data analysis, and writing, and funding (idea and writing being the most important).
VERY IMPORTANT: Please take note: supervisors DO NOT qualify for authorship if she/he has not contributed to the work with anything other than funding! If you feel forced by a supervisor to include her-/himself or others in work you have performed, and you ethically do not think they qualify, you should consult with your student director, science director, or the person dealing with intellectual property at the university! The same goes if data you have collected is used without your knowledge and consent, or the data being used without you being acknowledged! This is ethically a very serious matter, but unfortunately not something that has an easy solution. However, most supervisors do take actively part of the study and the writing.
Abstract
Purpose: To give a short overview of the whole article so that the reader gets the most important message and can evaluate if further reading is needed: question your text is addressing, methods you used to answer the question, your main results, and conclusions and/or implications. Please note this may be the only part of your article that gets read.
References: no (some conferences and journals may allow it, but unless stated so, don't include them)
Grammar: same as the individual sections of an article described below. Past tense for theories, methods, and results. Present tense for conclusions and implications.
Structure: Include all the most important elements of Introduction, Materials and Methods, Results, and Discussion/Conclusion in short form.

Introduction
Purpose: To give the reader the background needed to understand how the hypothesis for the experiment was derived.
References used: Yes. Each piece of information should be supported with 1 - 3 references, unless accepted as common knowledge (e.g. cats prey mice, the suns rays can be harmful, etc.).
Grammar: Present tense for general known information “Toxoplasma gondii is a protozoan parasite.”. Past tense for theories and earlier results.
Structure: Write as a reversed pyramid (broad towards focus). Start with what organism, topic/field your text addresses (not broader what should be known for readers of the journals or thesis work scope: e.g. veterinary medicine). Journals are not interested in a very detailed description of the parasite and its life cycle - get to the point quickly! Narrow down the topic to your specific interest (e.g. Zoonotic parasite: Toxoplasma – main host: cat – risk groups: kittens – risk behaviour: outdoor hunting – likely host of carrying parasite: mice – specific interest/hypothesis: mice in urban areas have higher infection levels with Toxoplasma due to high concentration of cats compared to rural areas). Identify clearly the gap in knowledge your study is going to fill: what makes your hypothesis original, what makes this study needed, what has been missed in previous research in the field. The introduction should end in a clearly stated aim/objective/question/hypothesis. If more than one, list in point form (1, 2, 3...). Keep this order throughout the text: give results in the same order, and discuss them in the same order. This is the heart of the study. Only this is examined in the study and attempted proved (to the scientific reader)! Everything else should be removed from the manuscript.

Materials and Methods
Purpose: To give the reader transparency of what was done to the extent that it can be replicated for confirmation of the findings.
References: Yes, including brands of materials used (name, company, country)
Grammar: Written in past tense.
Structure: Order according to what was done, starting with defining the study group examined (e.g. Cat population in the examined area). When describing methods, it must be clear what makes the method chosen appropriate in detecting what you examine for (direct (parasite itself detected) and indirect (antibodies detected) means to detect infection: sensitivity, specificity, error in measurement of samples).
End with describing statistics: programs used for analyses, how randomization was included (if applying), statistical methods used. If data was excluded or transformed (e.g. several answers in a questionnaire (“once”, “twice”, “never”) joined to form two answers instead (“never” and “once or more”)) it also has to be described here.
If the method has been used and described earlier, refer to the article describing it, and explain only where your method differs.
If methods are very complex or new it may be useful to present as an illustration/flow-diagram of how you structured you work.
Do not discuss the method in this section, only report it as used!

Results
Purpose: Presentation of outcomes from the measurements in a form that is correctly representing the results and easy to read.
References: No.
Grammar: Past tense.
Structure: If larger amounts of data can be compromised in a table or figure form, it is preferred to do so. Otherwise, describe in text in short and precise language.
Make sure the accuracy of your results reflects the sensitivity of your methods (e.g. It makes no sense to report 123.5 viruses since you cannot have half a virus, or to report a prevalence of 32.361% if your method is 99.9% specific – it should then be 32.4%).
If direct observations were made, images should be provided to give the reader an option of creating an opinion of accuracy.
Generally there are no rules for how to transmit a convincing result. Just as figures can be misleading if used incorrectly (pie diagrams are for example difficult to estimate proportional differences from), even hand drawn innovations may be more useful than words (e.g. circle diagrams overlapping to describe how different pathogens exist in the same group).
Always make sure to include errors (95% confidence intervals, standard deviations, p-values)! Without them, the data is meaningless since the results are a value that is the estimation of the true number with a certain accuracy. Accuracy reflects how well your study is performed.
Exclude irrelevant data from journal articles, only include data relevant to proving your hypothesis. In a thesis/dissertation work, additional data can be added.
Remember negative results (not significant) are also results (often very interesting results)!

Discussion
Purpose: To explain the success of the study and its attempt to prove the hypothesis, and the practical significance of your results. Here you answer your question. Moreover, you relate your findings to findings of other studies, critique them and your own.
References: Yes.
Grammar: Past tense
Structure: Shortly give an overview of the purpose of the study and whether it was accomplished to the authors own satisfaction. If more than one hypothesis, order them as presented in the introduction and discuss them accordingly. Reflect on to the reader how the results support or differ from other available knowledge. Offer an explanation to your result, but resist the temptation to speculate further than there is evidence to support your new theories.
Describe shortcomings of methods and design, and strengths (also compare to other studies, with references). If some of these shortcomings/strengths (e.g. for false negatives) were tested experimentally they can be presented as ”data not shown”.
You may also hint applications for further research or raise questions.

Conclusion:
Purpose: Give the most important and confident outcomes of the study.
References: No.
Grammar: Often in present tense.
Structure: Brief text of only the most important outcome(s) that can be confidently supported by the research done.

Acknowledgement:
Purpose: Credit people who contributed to the project but do not qualify as an author.
References: No.
Grammar: Present tense.
Structure: List people connected to the project not entitled for authorship (laboratory personnel, language editing, experts asked, etc.). Acknowledge contributions of materials from collaborators and funding agencies. 

References
Purpose: Allow the reader to check the original research in the field that is relevant to your work.
References: does not apply.
Grammar: does not apply.
Structure: Stated by the journal or committee.

Cover letter (when writing an article for a journal)
Purpose: Can be considered as a request with arguments for scientific peers to consider the work significant enough to consider for publication.
References: No.
Grammar: Written in present tense.
Structure: Polite, formal. “Dear Editor, yours faithfully.” State the title of your study. State why this manuscript is of novelty (what is new/innovative) and thus worthy of consideration as new knowledge. Thank the Editor and peers for their time (peer reviewers are most likely not paid to review your work).
Write how each authors contributed to the article (see “Authors”).

Further reading:
http://www.phrasebank.manchester.ac.uk/index.htm




No comments:

Post a Comment

Subscribe to: Posts (Atom)