Pseudoreplication and Ecology - Columbia University

Pseudoreplication and Ecology - Columbia University

Pseudoreplication and Ecology Dr. James A. Danoff-Burg Columbia University Purposes of Replication Controls for random or stochastic error E.g., untested independent factors may otherwise determine the outcome of the experiment Increases the precision of the test Increases the generalizability of the test If you test across many sites you can safely

generalize to many others Some Definitions Replicate = Sample Maximize these in your experimental design Greatest number possible, given logistical limitations If you are a professional, use a power analysis Subsample = Pseudoreplicate Only true if the subsamples are incorrectly treated as true replicates for statistical analysis Subsamples: useful to increase the accuracy of the data estimate for that replicate A special type of statistical analysis are therefore

possible Pseudoreplication - Defined Incorrect replication Replicating samples, not treatments Replicates are not independent Problem is that it violates a key assumption of statistical analysis: Independence of replicates Increasing precision of studies if independent Approximates truth better if independent Prevalence of

Pseudoreplication 48% of all studies had pseudoreplication (Hurlbert 1984) 71% of studies using ANOVA (a common statistical test) had design errors (Underwood 1981) Particularly acute in studies with logistical problems Rare animals Transportational or financial limitations

Many of ours! Examples Many samples from a single site These are actually subsamples Only a single sample for each treatment condition These are actually replicates, but cannot do statistics on a sample size of one Single samples from a single site, but replicated in time Would be true samples if the experimental question

is time-dependent If not, it is pseudoreplication Pseudoreplication Example Treatment A Treatment B Site 1 Site 3 Site 2

Site 4 Question What is the affect of treatments A & B? Pseudoreplicati on = treating stars of the same color as replicates Replication = include only a single star of each color

Controlling Pseudoreplication I Know your question Question determines whether design includes pseudoreplication Taxonomic level Ecological hierarchy level Clearly define your independent and dependent variables Controlling Pseudoreplication II What constitutes a unit of data?

Plant branch? Individual? Population? Etc.? Identify what is the unit of replication Individual? Population? Community? Site? Replicate accordingly sites are often the level of replication for our projects Randomize your sampling design Helps to decrease sampling errors For Our Class We will frequently use pseudoreplication Limitations on time (only 4 weekends for our work!) Limitations on transportation (only 1 van!)

Limitations on effort (only you!) Consequently We will frequently treat our pseudoreplicates as true replicates However be aware of this and you will be fine when you design more robust research projects in the future

Recently Viewed Presentations

  • NYC4SEC June 11, 2014 Meet-up Group at John Jay College

    NYC4SEC June 11, 2014 Meet-up Group at John Jay College

    NYC4SEC Meet-up Group - June 11th, 2014 . NYC4SEC Meet-up Group‎John Jay College-Criminal Justice899 10th Avenue, New York, New YorkWednesday June 11th, 2014 06:30pm. Introduction to the Microsoft exFAT File System. Robert Shullich . CPP, CISSP, CRISC, GSEC, GCFA, CEH,...
  • We have not answered the question where is

    We have not answered the question where is

    "We have not answered the question 'where is God?' But we have suggested how to go about answering the question. Search and you will find. But be careful.
  • Maryland Department of Health / Marylands Commitment to

    Maryland Department of Health / Marylands Commitment to

    Since 1919, Easter Seals has been serving the community by providing assistance with physical, mental, and emotional health issues. In 2017, we formed a partnership with the Cohen Veterans Network to provide on-site mental health services in DC metro area.
  • www.iflr.com

    www.iflr.com

    Excludes closed-end funds, REITs, units, best efforts offerings, non-U.S. offerings and offerings less than $10 million. Proceeds include over-allotment amount. Note: Growth sectors defined as aerospace and defense, alternative energy, consumer, health care, technology and telecommunications. A.
  • ACL Data Analysis - Cost Management Data Requirements Review

    ACL Data Analysis - Cost Management Data Requirements Review

    Applications / Data Sources Pathways Financial Management (PFM) Accounts Payable data Pathways Materials Management (PMM) Purchasing data Lawson Payroll & Employee data Star Accounts Receivable data Third Party Sources Credit Card vendor transaction data System Diagram ACL Software PFM PMM...
  • The use of PDMS micromodels to study CO2

    The use of PDMS micromodels to study CO2

    PDMS surface modification by UV-Ozone 1. Berdichevsky Y, et al, Sensors and Actuators B-Chemical 2004, 97, (2-3), 402-408. Ozone [1] Wettability control by water immersion Wettability maintenance by keeping UV-ozone-treated PDMS (1-hour curing at 80 °C) surface in contact with...
  • Public service restructuring in the UK: the English NHS

    Public service restructuring in the UK: the English NHS

    No narrative - rationale. 2010 White Paper Equity & Excellence: Liberating the NHS referred to . clinicians in the driving seat, creating the largest social enterprise economy in the world, patient voice 'no decision about me without me' and choice...
  • East Asia Iii

    East Asia Iii

    THE JAKOTA TRIANGLE CHARACTERISTICS Great cities Enormous consumption of raw materials State-of-the-art industries Voluminous exports Global links Trades surpluses Rapid development CHALLENGES Social problems Political uncertainties Vulnerabilities OUTLINE OF JAPANESE HISTORY 600 - 800 Chinese ...