DOCSLIB.ORG

New Techniques for Time-Activity Studies of Avian Flocks in View-Restricted Habitats

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
New Techniques for Time-Activity Studies of Avian Flocks in View-Restricted Habitats j. Field Ornithol., 60(3):388-396 NEW TECHNIQUES FOR TIME-ACTIVITY STUDIES OF AVIAN FLOCKS IN VIEW-RESTRICTED HABITATS MICHAEL P. LOSITO,• RALPH E. MIRARCHI, AND GUY A. BALDASSARRE 1 Departmentof Zoologyand WildlifeScience AlabamaAgricultural Experiment Station Auburn University,Alabama 36849-5414 USA Abstract.--Focal-animal sampling was comparedto a newly developedtechnique, focal- switchsampling, to evaluatetime budgetsof Mourning Doves(Zenaida macroura) in habitats where the field of view was restricted.Focal-switch sampling included a formula employed to weighthabitat use and testfor restrictionsof habitatstructure on behavior,and a standard waiting period to decidewhen to end sampling or continue pursuit of lost flocks. Focal- animal samplingbiased estimates of activebehavior downward, whereas estimates of inactive behaviorwere similar for both methods.Focal-switch sampling increased research efficiency by 12%. The standardwaiting period saved24% of samplesfrom prematuretermination and reducedobserver bias by maintaining equal sampling effort per flock. Weighting of habitat use also reducedsampling bias. Focal-switch sampling is recommendedfor use in conjunctionwith focal-animalsampling when samplingin view-restrictedhabitats. NUEVAS TP•CNICASPARA EL ESTUDIO DE ACTIVIDADES DE CONGREGACIONES DE AVES EN HABITATS CON VISIBILIDAD RESTRINGIDA Resumen.--Lat6cnica de observaci6ndirecta de animales(focal-animal) es comparada con un nuevom6todo en dondese evaluan los presupuestosde actividadesen Zenaidamacroura en habitatsen dondeel campode visi6nestaba restringido. La nuevat6cnica (cambio-focal) incluyeuna f6rmulapara "pesar"la utilizaci6nde habitaty medirlas restriccionesque imponenla estructuradel habitaten la conducta.Incluye ademfis,un periodode espera estandarizadoque permite al observadordecidir cuando terminar susobservaciones o conti- nuar las mismas.E1 muestreo de un soloanimal (focal-animal)tiene un sesgopara estimar la conductaactira, mientras que los estimados para conducta inactiva rueton similares para ambosm6todos. E1 nuevo m6todo increment6 la eficienciadel estudioen un 12%.E1 periodo de esperaestandarizado evit6 que se terminara prematuramente con el 24% de lasmuestras observadasy redujo el sesgoal mantenetel mismoesfuerzo de muestreopot agregaci6nde aves.Se recomienda la utilizaci6ndel nuevo m6todo en conjunci6n con el previamentedescrito cuandose muestreeen habitatdonde la visibilidadpueda set obstruida. There are many effectiveand widely used methodsto study activity budgetsof avian flocks (e.g., Altmann 1974, Baldassarreet al. 1987). Most studieshave focusedon speciesthat congregatein open habitats (e.g., shorelineand open water) where large numbersof individualscan be sampledreadily (Maxson and Bernstein1984, Paulus 1984a, Quinlan and Baldassarre1984, Tamisier 1976). However, problemsof visibility biasarise in studiesof speciesin densehabitats (Bradley 1985), because the investigatoris confrontedwith samplingconstraints that often cause data variability and biases(Kessel 1976, Verner 1965). • Currentaddress: Environmental and ForestBiology, College of EnvironmentalScience and Forestry,State University of New York,Syracuse, New York 13210. 388 Vol.60, No. 3 NewTime-Activity Methods [389 Focal-animalsampling is the observationof an individualfor a pre- determinedperiod of time (Altmann 1974). It may be the best method for detailedanalysis of behaviorand the only methodavailable to sample small numbersof individuals.Focal-animal samplingis thereforeappro- priate for studiesof speciesin densevegetation where the number of potentialsample individuals is frequentlyreduced by vegetation. A major, unresolvedproblem with focal-animalsampling occurs when focal individualsdisappear from view (hereafterreferred to as "out-of- view";see Lehner 1979). For example,Verner (1965) arbitrarilyrecorded Marsh Wren (Cistothoruspalustris) out-of-view data as feeding,whereas Maxson (1977) broadlymeasured Ruffed Grouse(Bonasa umbellus) ac- tivity by radiotelemetry.Other observers(e.g., Paulus 1984b, Rave 1987) had biasedresults because they were unableto focal-sampleflying birds. Suchlimitations may discouragetime-activity studies of flockingupland species.Accordingly, techniques must be developedto addressproblems associatedwith: (1) focal individuals disappearingfrom view; (2) the amount of time a focal individual should be observed;(3) the spatial distributionof flock individualsamong heterogeneoushabitats; and (4) the eliminationof observer-expectancybias when a sampleflock departs beforethe terminationof the sampleperiod. We reporta new technique,focal-switch sampling, developed to address thesesampling problems during an activity-budgetstudy of Mourning Dove (Zenaida macroura) flocks in northern Alabama. We also report someresults of the activity study by relating behavioralfrequencies to broad habitat categorieswhile employingboth focal-animal and focal- switch sampling. METHODS Focal-switchsampling was testedagainst focal-animal sampling using radio-fitted hatching-year(HY) Mourning Doves to lead observersto dove flocks on the Swan Creek Wildlife Management Area, Wheeler National Wildlife Refuge, and surroundingareas in northern Alabama. Dove flockswere locatedfrom July to September1987. Radio-marked individualswere not observedfor behavioraldata to eliminate potential bias inducedby the radio transmitter.Flocks were sampledduring three observationperiods (OPi = sunriseto 3.5 h after sunrise,OP2 = 3.5 h after sunriseto 3.5 h before sunset,OP3 = 3.5 h before sunsetto sunset) to block-out variation causedby differing periods of Mourning Dove activity (Deuver and Fatora 1968). The closestreadily visible individual in the flock was selectedas the first bird to be sampled(Altmann 1974). Two observers(one per method) collecteddata on eachsample flock. Each methodwas usedby all four observersduring the study.The total data include30.5 h (OP1 -- 10 h, 10 min; OP2 = 10 h; OP3 = 10 h, 20 min). Behavioral categorieswere: feeding,locomotion, comfort, resting, alert, and out-of-view. Focal-switchsampling, unlike focal-animalsampling, allowed switch- ing to a new focal individual (nearestneighbor) if the original focal 390] M. P. Lositoet al. J. Field Ornithol. Summer 1989 T^BLE 1. Demonstrationof weighting focal samplesaccording to habitat use (Y•) where nA = 50, nB = 25, nc = 25, N = 100, T = 60 min, and f= 5 min.a Focal-unit (sample {s}; no sample {--}) Habitat 1 2 3 4 5 6 7 8 9 10 11 12 A s -- -- s -- -- s -- -- s s s B -- s -- -- s -- -- s .... C -- -- s -- -- s -- -- s -- -- -- a y, = (n,/N) x (T/J) where: Y, = numberof focal-units(f) sampledin habitat i; n, = number of individuals in habitat z; N = total number of individuals in flock; T = duration of sampleperiod;f = duration focal-unit. Then YA= n,•/N x T/f = 50/100 x 60/5 = 6, YB=3, and Yc = 3. individual went out-of-view. To avoid observer-expectancybias (Balph and Romesburg1986), two consecutiveout-of-view recordingswere al- lowed before focal-switching.Focal-switching also was restrictedto in- dividuals in the same habitat as the out-of-view individual. This controlled the effectsof habitat structureon behavior.For example, if a focal in- dividual in habitat A moved to habitat B and was lost from view, focal- switchingwas allowed only on individualsin habitat B. The null hypothesistested was that the two methodsresulted in the collectionof the sameamount of data per behavioralcategory during a given unit time. We used Hotelling's multivariate t-tests (SAS Institute, Inc. 1985) to determinethe overall differencein resultsobtained by use of the two methods.Because the data were normally distributed,we used pairedt-tests (SAS Institute,Inc. 1985) to comparebehavioral frequencies betweenmethods for each observationperiod. A probability level of 0.05 was used for all statistical tests. The sampleperiod (T), the maximum time limit for samplinga flock, was dividedinto focal-units(f) where: T = 60 min, andf = 5 min. The focal-unitwas the predeterminedtime within which behaviorof the focal individual(s)was sampledinstantaneously every 20 s (Altmann 1974) as dictatedby a tape recorderused as a metronome(Paulus 1984a, 1984b). The number of completefocal-units (focal-switching not needed)versus the number of incompletefocal-units (focal-switchingnecessary) was testedfor equalityusing paired t-tests(SAS Institute, Inc. 1985). To beginsampling a flock,the observersestimated visually the number of individualsusing major habitat types (e.g., upland field, forestcanopy, willow thicket) and focal-unitswere distributedproportionately (Table 1). Habitat weightingwas testedby comparingactivities sampled in perch (e.g., forestcanopy + willow thicket canopy)versus non-perch habitats (e.g.,upland field + mudflat). Unpaired t-tests(SAS Institute,Inc. 1985) were usedto testthe null hypothesisthat the two habitat structuresaffected behaviorequally. When a sampleflock departed before termination of the sampleperiod, we allowed a standardperiod of 10 min for a new samplingsituation to Vol.60, No. 3 NewTime-Activity Methods [391 TABLE2. Number of instantaneousrecordings (mean + SE) of each Mourning Dove behaviorcollected using focal-animaland focal-switchmethods simultaneously during eachobservation period, northern Alabama, 1987. Method b Observation Behavioral category perioda Focal-animal Focal-switch Locomotion 1 4.5 + 1.1' 9.5 + 2.3* 2 2.6 + 1.3' 5.2 + 1.9' 3 6.9 _+ 1.4' 11.9 _+ 2.4* Feeding 1 21.0 + 5.0* 32.3 + 8.5* 2 1.4 + 0.5 4.5 + 2.3 3 38.6 + 8.0 49.0 ___11.0 Resting 1 13.8 + 3.1 12.3 +
Load more

Recommended publications
  • Downloads Automatically
    VIRGINIA LAW REVIEW ASSOCIATION VIRGINIA LAW REVIEW ONLINE VOLUME 103 DECEMBER 2017 94–102 ESSAY ACT-SAMPLING BIAS AND THE SHROUDING OF REPEAT OFFENDING Ian Ayres, Michael Chwe and Jessica Ladd∗ A college president needs to know how many sexual assaults on her campus are caused by repeat offenders. If repeat offenders are responsible for most sexual assaults, the institutional response should focus on identifying and removing perpetrators. But how many offenders are repeat offenders? Ideally, we would find all offenders and then see how many are repeat offenders. Short of this, we could observe a random sample of offenders. But in real life, we observe a sample of sexual assaults, not a sample of offenders. In this paper, we explain how drawing naive conclusions from “act sampling”—sampling people’s actions instead of sampling the population—can make us grossly underestimate the proportion of repeat actors. We call this “act-sampling bias.” This bias is especially severe when the sample of known acts is small, as in sexual assault, which is among the least likely of crimes to be reported.1 In general, act sampling can bias our conclusions in unexpected ways. For example, if we use act sampling to collect a set of people, and then these people truthfully report whether they are repeat actors, we can overestimate the proportion of repeat actors. * Ayres is the William K. Townsend Professor at Yale Law School; Chwe is Professor of Political Science at the University of California, Los Angeles; and Ladd is the Founder and CEO of Callisto. 1 David Cantor et al., Report on the AAU Campus Climate Survey on Sexual Assault and Sexual Misconduct, Westat, at iv (2015) [hereinafter AAU Survey], (noting that a relatively small percentage of the most serious offenses are reported) https://perma.cc/5BX7-GQPU; Nat’ Inst.
    [Show full text]
  • Human Dimensions of Wildlife the Fallacy of Online Surveys: No Data Are Better Than Bad Data
    Human Dimensions of Wildlife, 15:55–64, 2010 Copyright © Taylor & Francis Group, LLC ISSN: 1087-1209 print / 1533-158X online DOI: 10.1080/10871200903244250 UHDW1087-12091533-158XHuman Dimensions of WildlifeWildlife, Vol.The 15, No. 1, November 2009: pp. 0–0 Fallacy of Online Surveys: No Data Are Better Than Bad Data TheM. D. Fallacy Duda ofand Online J. L. Nobile Surveys MARK DAMIAN DUDA AND JOANNE L. NOBILE Responsive Management, Harrisonburg, Virginia, USA Internet or online surveys have become attractive to fish and wildlife agencies as an economical way to measure constituents’ opinions and attitudes on a variety of issues. Online surveys, however, can have several drawbacks that affect the scientific validity of the data. We describe four basic problems that online surveys currently present to researchers and then discuss three research projects conducted in collaboration with state fish and wildlife agencies that illustrate these drawbacks. Each research project involved an online survey and/or a corresponding random telephone survey or non- response bias analysis. Systematic elimination of portions of the sample population in the online survey is demonstrated in each research project (i.e., the definition of bias). One research project involved a closed population, which enabled a direct comparison of telephone and online results with the total population. Keywords Internet surveys, sample validity, SLOP surveys, public opinion, non- response bias Introduction Fish and wildlife and outdoor recreation professionals use public opinion and attitude sur- veys to facilitate understanding their constituents. When the surveys are scientifically valid and unbiased, this information is useful for organizational planning. Survey research, however, costs money.
    [Show full text]
  • Levy, Marc A., “Sampling Bias Does Not Exaggerate Climate-Conflict Claims,” Nature Climate Change 8,6 (442)
    Levy, Marc A., “Sampling bias does not exaggerate climate-conflict claims,” Nature Climate Change 8,6 (442) https://doi.org/10.1038/s41558-018-0170-5 Final pre-publication text To the Editor – In a recent Letter, Adams et al1 argue that claims regarding climate-conflict links are overstated because of sampling bias. However, this conclusion rests on logical fallacies and conceptual misunderstanding. There is some sampling bias, but it does not have the claimed effect. Suggesting that a more representative literature would generate a lower estimate of climate- conflict links is a case of begging the question. It only make sense if one already accepts the conclusion that the links are overstated. Otherwise it is possible that more representative cases might lead to stronger estimates. In fact, correcting sampling bias generally does tend to increase effect estimates2,3. The authors’ claim that the literature’s disproportionate focus on Africa undermines sustainable development and climate adaptation rests on the same fallacy. What if the climate-conflict links are as strong as people think? It is far from obvious that acting as if they were not would somehow enhance development and adaptation. The authors offer no reasoning to support such a claim, and the notion that security and development are best addressed in concert is consistent with much political theory and practice4,5,6. Conceptually, the authors apply a curious kind of “piling on” perspective in which each new paper somehow ratchets up the consensus view of a country’s climate-conflict links, without regard to methods or findings. Consider the papers cited as examples of how selecting cases on the conflict variable exaggerates the link.
    [Show full text]
  • 10. Sample Bias, Bias of Selection and Double-Blind
    SEC 4 Page 1 of 8 10. SAMPLE BIAS, BIAS OF SELECTION AND DOUBLE-BLIND 10.1 SAMPLE BIAS: In statistics, sampling bias is a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others. It results in abiased sample, a non-random sample[1] of a population (or non-human factors) in which all individuals, or instances, were not equally likely to have been selected.[2] If this is not accounted for, results can be erroneously attributed to the phenomenon under study rather than to the method of sampling. Medical sources sometimes refer to sampling bias as ascertainment bias.[3][4] Ascertainment bias has basically the same definition,[5][6] but is still sometimes classified as a separate type of bias Types of sampling bias Selection from a specific real area. For example, a survey of high school students to measure teenage use of illegal drugs will be a biased sample because it does not include home-schooled students or dropouts. A sample is also biased if certain members are underrepresented or overrepresented relative to others in the population. For example, a "man on the street" interview which selects people who walk by a certain location is going to have an overrepresentation of healthy individuals who are more likely to be out of the home than individuals with a chronic illness. This may be an extreme form of biased sampling, because certain members of the population are totally excluded from the sample (that is, they have zero probability of being selected).
    [Show full text]
  • Cognitive Biases in Software Engineering: a Systematic Mapping Study
    Cognitive Biases in Software Engineering: A Systematic Mapping Study Rahul Mohanani, Iflaah Salman, Burak Turhan, Member, IEEE, Pilar Rodriguez and Paul Ralph Abstract—One source of software project challenges and failures is the systematic errors introduced by human cognitive biases. Although extensively explored in cognitive psychology, investigations concerning cognitive biases have only recently gained popularity in software engineering research. This paper therefore systematically maps, aggregates and synthesizes the literature on cognitive biases in software engineering to generate a comprehensive body of knowledge, understand state of the art research and provide guidelines for future research and practise. Focusing on bias antecedents, effects and mitigation techniques, we identified 65 articles (published between 1990 and 2016), which investigate 37 cognitive biases. Despite strong and increasing interest, the results reveal a scarcity of research on mitigation techniques and poor theoretical foundations in understanding and interpreting cognitive biases. Although bias-related research has generated many new insights in the software engineering community, specific bias mitigation techniques are still needed for software professionals to overcome the deleterious effects of cognitive biases on their work. Index Terms—Antecedents of cognitive bias. cognitive bias. debiasing, effects of cognitive bias. software engineering, systematic mapping. 1 INTRODUCTION OGNITIVE biases are systematic deviations from op- knowledge. No analogous review of SE research exists. The timal reasoning [1], [2]. In other words, they are re- purpose of this study is therefore as follows: curring errors in thinking, or patterns of bad judgment Purpose: to review, summarize and synthesize the current observable in different people and contexts. A well-known state of software engineering research involving cognitive example is confirmation bias—the tendency to pay more at- biases.
    [Show full text]
  • Correcting Sampling Bias in Non-Market Valuation with Kernel Mean Matching
    CORRECTING SAMPLING BIAS IN NON-MARKET VALUATION WITH KERNEL MEAN MATCHING Rui Zhang Department of Agricultural and Applied Economics University of Georgia [email protected] Selected Paper prepared for presentation at the 2017 Agricultural & Applied Economics Association Annual Meeting, Chicago, Illinois, July 30 - August 1 Copyright 2017 by Rui Zhang. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies. Abstract Non-response is common in surveys used in non-market valuation studies and can bias the parameter estimates and mean willingness to pay (WTP) estimates. One approach to correct this bias is to reweight the sample so that the distribution of the characteristic variables of the sample can match that of the population. We use a machine learning algorism Kernel Mean Matching (KMM) to produce resampling weights in a non-parametric manner. We test KMM’s performance through Monte Carlo simulations under multiple scenarios and show that KMM can effectively correct mean WTP estimates, especially when the sample size is small and sampling process depends on covariates. We also confirm KMM’s robustness to skewed bid design and model misspecification. Key Words: contingent valuation, Kernel Mean Matching, non-response, bias correction, willingness to pay 2 1. Introduction Nonrandom sampling can bias the contingent valuation estimates in two ways. Firstly, when the sample selection process depends on the covariate, the WTP estimates are biased due to the divergence between the covariate distributions of the sample and the population, even the parameter estimates are consistent; this is usually called non-response bias.
    [Show full text]
  • Quantifying Aristotle's Fallacies
    mathematics Article Quantifying Aristotle’s Fallacies Evangelos Athanassopoulos 1,* and Michael Gr. Voskoglou 2 1 Independent Researcher, Giannakopoulou 39, 27300 Gastouni, Greece 2 Department of Applied Mathematics, Graduate Technological Educational Institute of Western Greece, 22334 Patras, Greece; [email protected] or [email protected] * Correspondence: [email protected] Received: 20 July 2020; Accepted: 18 August 2020; Published: 21 August 2020 Abstract: Fallacies are logically false statements which are often considered to be true. In the “Sophistical Refutations”, the last of his six works on Logic, Aristotle identified the first thirteen of today’s many known fallacies and divided them into linguistic and non-linguistic ones. A serious problem with fallacies is that, due to their bivalent texture, they can under certain conditions disorient the nonexpert. It is, therefore, very useful to quantify each fallacy by determining the “gravity” of its consequences. This is the target of the present work, where for historical and practical reasons—the fallacies are too many to deal with all of them—our attention is restricted to Aristotle’s fallacies only. However, the tools (Probability, Statistics and Fuzzy Logic) and the methods that we use for quantifying Aristotle’s fallacies could be also used for quantifying any other fallacy, which gives the required generality to our study. Keywords: logical fallacies; Aristotle’s fallacies; probability; statistical literacy; critical thinking; fuzzy logic (FL) 1. Introduction Fallacies are logically false statements that are often considered to be true. The first fallacies appeared in the literature simultaneously with the generation of Aristotle’s bivalent Logic. In the “Sophistical Refutations” (Sophistici Elenchi), the last chapter of the collection of his six works on logic—which was named by his followers, the Peripatetics, as “Organon” (Instrument)—the great ancient Greek philosopher identified thirteen fallacies and divided them in two categories, the linguistic and non-linguistic fallacies [1].
    [Show full text]
  • Thesis with Row Frequencies (RF) and Column Frequencies (CF)………………………...173
    SAMPLING BIASES AND NEW WAYS OF ADDRESSING THE SIGNIFICANCE OF TRAUMA IN NEANDERTALS by Virginia Hutton Estabrook A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Anthropology) in The University of Michigan 2009 Doctoral Committee: Professor Milford H. Wolpoff, Chair Professor A. Roberto Frisancho Professor John D. Speth Associate Professor Thomas R. Gest Associate Professor Rachel Caspari, Central Michigan University © Virginia Hutton Estabrook 2009 To my father who read to me my first stories of prehistory ii ACKNOWLEDGEMENTS There are so many people whose generosity and inspiration made this dissertation possible. Chief among them is my husband, George Estabrook, who has been my constant cheerleader throughout this process. I would like to thank my committee for all of their help, insight, and time. I am truly fortunate in my extraordinary advisor and committee chair, Milford Wolpoff, who has guided me throughout my time here at Michigan with support, advice, and encouragement (and nudging). I am deeply indebted to Tom Gest, not only for being a cognate par excellence, but also for all of his help and attention in teaching me gross anatomy. Rachel Caspari, Roberto Frisancho, and John Speth gave me their attention for this dissertation and also offered me wonderful examples of academic competence, graciousness, and wit when I had the opportunity to take their classes and/or teach with them as their GSI. All photographs of the Krapina material, regardless of photographer, are presented with the permission of J. Radovčić and the Croatian Natural History Museum. My research would not have been possible without generous grants from Rackham, the International Institute, and the Department of Anthropology.
    [Show full text]
  • 730 Philosophy 101, Sect 01 Logic, Reason and Persuasion
    730 Philosophy101, Sect 01 Logic, Reason and Persuasion Instructor: SidneyFelder,e-mail: [email protected] Rutgers The State University of NewJersey, Fall 2009, Tu&Th2:15-3:35, TH-206 I. Fundamental Philosophical Terms and Concepts of Language (week 1) Extension and Intension; Puzzles of Identity and Reference; Types and Tokens; Language and Metalanguage; Logical Paradoxes. Course Notes: Fundamental Terms and Concepts of Language.Schaum’s pps. 16-17 II. Elements of Set Theory (weeks 1, 2 and 3) Sets and Subsets. Union, Intersection, and Complementation. Ordered Pairs, Cartesian Products; Relations, Functions. Equivalence Relations. Denumerable and Non-denumerable Infinite Sets. Course Notes: Elements of Set Theory. III. Linear and Partial Orderings (weeks 3 and 4) Linear Orderings; Partial Orderings. Upper Bounds and Lower Bounds; Least Upper and Greatest Lower Bounds. Course Notes: Linear and Partial Orderings. IV.Truth-Functional Propositional Logic (weeks 4, 5, and 6) Symbolism. The Logical Constants. Logical Implication and Equivalence. Consistency, Satisfiability,and Validity. Course Notes: Truth Functional Propositional Logic. Schaum’s pps. 44-68 V. Quantification (weeks 6and 7) ‘All’ and ‘There exists’ — The universal and existential quantifiers. Free and Bound variables; Open and Closed Formulæ (Sentences). Interpretations and Models; Number. Logical Strength. Course Notes: Quantification Logic. Schaum’s ch. 5; ch. 6 pps. 130-149; ch. 9 pps. 223-226 Midterm VI. Fallacies (week 8) Strawman. Ad Hominem;Argument from Authority; Arguments from Pervasiveness of Belief. Arguments from Absence of Information. Schaum’s ch. 8 Syllabus Page 1 Philosophy101, Sect 01 Logic, Reason and Persuasion VII. Probability and Statistical Arguments (weeks 9 and 10) Elements of Axiomatic Probability.
    [Show full text]
  • Misunderstandings Between Experimentalists and Observationalists About Causal Inference
    J. R. Statist. Soc. A (2008) 171, Part 2, pp. 481–502 Misunderstandings between experimentalists and observationalists about causal inference Kosuke Imai, Princeton University, USA Gary King Harvard University, Cambridge, USA and Elizabeth A. Stuart Johns Hopkins Bloomberg School of Public Health, Baltimore, USA [Received January 2007. Final revision August 2007] Summary. We attempt to clarify, and suggest how to avoid, several serious misunderstandings about and fallacies of causal inference. These issues concern some of the most fundamental advantages and disadvantages of each basic research design. Problems include improper use of hypothesis tests for covariate balance between the treated and control groups, and the conse- quences of using randomization, blocking before randomization and matching after assignment of treatment to achieve covariate balance. Applied researchers in a wide range of scientific disci- plines seem to fall prey to one or more of these fallacies and as a result make suboptimal design or analysis choices. To clarify these points, we derive a new four-part decomposition of the key estimation errors in making causal inferences. We then show how this decomposition can help scholars from different experimental and observational research traditions to understand better each other’s inferential problems and attempted solutions. Keywords: Average treatment effects; Blocking; Covariate balance; Matching; Observational studies; Randomized experiments 1. Introduction Random treatment assignment, blocking before assignment, matching after data collection and random selection of observations are among the most important components of research designs for estimating causal effects. Yet the benefits of these design features seem to be regularly mis- understood by those specializing in different inferential approaches.
    [Show full text]
  • An Investigation of the Moral Stereotype of Scientists Master Thesis Alessandro Santoro
    Evil, or Weird? An investigation of the moral stereotype of scientists Master Thesis Alessandro Santoro Student Number 10865135 University of Amsterdam 2015/16 Supervisors dhr. dr. Bastiaan Rutjens Second Assessor: dhr. dr. Michiel van Elk University of Amsterdam Department of Social Psychology Evil, or Weird? An investigation of the moral stereotype of scientists Alessandro Santoro University of Amsterdam A recent research by Rutjens and Heine (2016) investigated the moral stereotype of scientists, and found them to be associated with immoral behaviors, especially purity violations. We developed novel hypotheses that were tested across two studies, inte- grating the original findings with two recent lines of research: one suggesting that the intuitive associations observed in the original research might have been influenced by the weirdness of the scenarios used (Gray & Keeney, 2015), and another using the dual-process theory of morality (Greene, Nystrom, Engell, Darley, & Cohen, 2004) to investigate how cognitive reflection, as opposed to intuition, influences moral judg- ment. In Study 1, we did not replicate the original results, and we found scientists to be associated more with weird than with immoral behavior. In Study 2, we did not find any effect of reflection on the moral stereotype of scientists, but we did replicate the original results. Together, our studies formed an image of a scientist that is not necessarily evil, but rather perceived as weird and possibly amoral. In our discussion, we acknowledge our studies’ limitations, which in turn helped us to meaningfully interpret our results and suggest directions for future research. Keywords: Stereotyping, Moral Foundations Theory, Cognitive Reflection How far would a scientist go to prove a theory? In 1802, et al., 2015) has shown a lack of interest in people to a training doctor named Stubbins Ffirth hypothesized pursue a science related career.
    [Show full text]
  • Data Fallacies to Avoid
    DATA FALLACIES TO AVOID CHERRY PICKING DATA DREDGING SURVIVORSHIP BIAS Selecting results that fit your claim and excluding Repeatedly testing new hypotheses against the same Drawing conclusions from an incomplete set of data, those that don’t. set of data, failing to acknowledge that most because that data has ‘survived’ some selection criteria. correlations will be the result of chance. COBRA EFFECT FALSE CAUSALITY GERRYMANDERING Setting an incentive that accidentally produces the Falsely assuming when two events appear related Manipulating the geographical boundaries used to opposite result to the one intended. Also known as a that one must have caused the other. group data in order to change the result. Perverse Incentive. SAMPLING BIAS GAMBLER’S FALLACY HAWTHORNE EFFECT Drawing conclusions from a set of data that isn’t Mistakenly believing that because something has The act of monitoring someone can affect their representative of the population you’re trying to understand. happened more frequently than usual, it’s now less behaviour, leading to spurious findings. Also known as likely to happen in future (and vice versa). the Observer Effect. REGRESSION TOWARDS THE MEAN SIMPSON’S PARADOX MCNAMARA FALLACY When something happens that’s unusually good or When a trend appears in different subsets of data but Relying solely on metrics in complex situations and bad, it will revert back towards the average over time. disappears or reverses when the groups are combined. losing sight of the bigger picture. OVERFITTING PUBLICATION BIAS DANGER OF SUMMARY METRICS Creating a model that’s overly tailored to the data you Interesting research findings are more likely to be Only looking at summary metrics and missing big have and not representative of the general trend.
    [Show full text]