Different sampling methods in agricultural and biological sectors
2006
Shaterian, Javad | Bahar, Massood | Mozafari, Javad | Hajilooie, Saeid
Accuracy and precision of sampling from plant population, field (soil) and explants from storage or from experimental materials under laboratory conditions should be follow an university level standards. These samples are used when evaluating treatment effects or accessing quality of traits related to the imported or exported agricultural items. Sample size (n) has a decisive effects on research expenses and with enlarging sample size amount of standard error decreases (decreasing denominator of ‏σ/n), extend of estimation becomes narrower and precision of judgment increases. There is no universal sampling method to apply on every circumstances, sampling method and size varies, depend on the nature of trait and variance of traits within population. Random sampling techniques have different types as follow : - Simple random method is used, where address of each member within population is available and only one type of variation is present. - Randomly ordered method, each individual in population selection follows a fixed distance from each other, this distance is identified by rounded ratio of N/n, - Random multi-steps method, when there is no one common type of variation within population, different sampling method in different period of sampling is adopted. - Random classic (layer) method, number of sampling units (n) is arrived by multiplication of number of layers in each plot ( K layers) by number of layers ( m sampling units), (m) × (k). When a large variation among individual within population exists and source of variation follows an uniform pattern, this method of sampling method is useful. In this method, sampling precision increases by grouping individuals into different layers, that should make within group variation smaller than between groups&rsquovariation. - Random multi-process / classified / bunch-form method, is a combination between random layer and random multi-step methods. - Subsidiary sampling method, when trait X is very variable, to arrive at adequate precision large sample size is needed. In this case, an independent and easier to record and smaller variation Z which is linked to the X variable is identified and recorded. By setting a regression program between dependent variable X and independent variable Z could give an adequate estimate of variable X.
Показать больше [+] Меньше [-]