MES 503 Research Methods

Sennai's phone- 7274820113 office hours Friday 10-1 get Experimental design book for next semester october 7th is first exam (24 hours) can use any recourse except talking to people  exam 2 is december 2th save syllabus to MES folder, put schedule in calender Joe Townsend tutoring sunday evenings- save zoom meeting. discussion board on black board  read grittoli article, discussion, start page in notes for next week, read first chapter in book pre course assesment    Tips: stay organized- download data sets beforehand keep programming notebook  work on things little bit everyday     Parametric statistics? Intro to experimental design Treatment is conditioned imposed on experimental unites- the thing that you are manipulating  Experimental unit(or sample)- unit that did or did not recieve the treatment  control- the unit that did not receive the treatment but otherwise identical  confounding factor- not of interest but may change nature of results if not controlled (like water flow in recruitment studies) randomization- factors that cant be controlled can be randomized to remove bias, balances effects of potential confounding factors replication- assess consistency, allows measurement of variability. more replication gets closer to natural pop statistical independence- experiemental unites are independent if its response does not affect how another unit will respond. one does not affect the other. if you cannot show independence, you have to account for it in stats Data - type, scale of measurement, role     can be discrete- only take certain values eg:small medium large    continuous-    any values,  no gaps eg: size weight temp salinity    scales of  measurement for discrete  data Nominal- name, classification, no ordering,  arbitrary can only count number of occurances central tendency used is  mode Ordinal- ordered name, but not continuous data eg restaurant ratings  can rank ordinal data but cannot quantify  differences between two ordinal values  central tendency is median or mode  Roles of data names (classes) strings, character, numeric, integers can be used to group observations ??                    

Lipids and stable carbon isotope in two species of hawaiian corals

scientific hypothesis- Does lipid concentration and chlorophyll a levels decrease with higher levels of coral bleaching? Does this response vary among species? Is skeletal carbon, measured in delta C 13, effected? what is the corelation between coral bleaching and lipid concentration, lipid class, and clorophyll a concentrations among Porites compressa and Montipora verrucosa.    statistical hypothesis predicts trends- Bleached corals will have lower levels of chlorophyll a, lipids, and    Grittoli et al. (2004) investigates several hypothesis on the impacts of bleaching on coral species Porites compressa and Montipora verrucosa. Corals mainly obtain energy through their symbiotic relationship with algae, which can be measured in chlorophyll a levels, but can also subside on lipid concentrations stored from excess carbon. Depending on where this energy is derived from is evident in the delta C 13 levels in coral skeleton. During bleaching events, when symbionts are not photosynthesizing to feed the coral, there is evidence that lipid composition changes. The main scientific hypothesis are 1) Lipid concentration and type, chlorophyll a, and delta C 14  levels are affected by coral bleaching, and 2) how do P. compressa and M. verrucosa each respond to bleaching in these regards. These address cause-and-effect relationships, therefore represent the scientific hypothesis. Statistical hypothesis predict trends. In this study, the alternate hypothesis, supporting that a pattern does exist, states that 1) lipid concentration, chlorophyl a levels, and delta C 14 decreases with increased bleaching, and 2) these results will differ depending on species. The null hypothesis would state that these variables are all random and have no effect. 

Summary Statistics, Framing and Testing Hypotheses

location- where is majority of data found (mean, median, mode) spread- how variable data are (sd, variance, standard errors) Y stands for random variables subscript i indicates observation n is sample size, i can be any integer between 1 and n greek letters show unknown parameters of distributions  unbiased estimator must meet three conditions: 1) random 2)independent  3) drawn from larger pop that has normal random variable Law of Large Numbers - as sample size n increases, the arithmetic mean of Y  approaches the expected value of Y, E(Y)  ?? Geometric mean ??