Bio 3.1.PDF

Chapter 3: Designing and Interpreting [xperiments Les$On 3.1 Experimental Design Biological research is conducted via a process commonly referred to as the scientific methoe1, which involves systematic observation and experimentation. Appropriate experimental design is an essentia...

Chapter 3: Designing and Interpreting [xperiments Les$On 3.1 Experimental Design Biological research is conducted via a process commonly referred to as the scientific methoe1, which involves systematic observation and experimentation. Appropriate experimental design is an essential aspect of biological research that produces reliable and valid data, which can be analyzed to reach defensible conclusions. This lesson explores fundamental components of experimental design, including types of variables, controls, and conclusions based on experimental outcomes. Although this lesson focuses primarily on biological research, the fundamentals of experimental design apply to all areas of scientific research. As such, knowledge of the principles of appropriate experimental design may be tested in multiple sections of the MCAT. a 3.1.01 Experimental Approach Biological research typically entails conducting systematic investigations that allow cause-and-effect relationships to be determined among variables through the analysis of data generated via experimentation. This experimental approach typically involves the generation of testable hypotheses, which then guide the process of experimental design. The results of an initial experiment typically guide future experiments, either for replicating the initial results (ie, strengthening an underlying scientifictheory)ortesting a revised hypothesis if the results of the initial experiment did not support the original hypothesis. Figure 3.1 summarizes a common way in which scientific experimentation is used to address a scientific problem. ldentify scientilic problem I * Generate testable hypothesis + Design experiment to test hypothesis Revise tl I hypothesis Replicate experiment to I I + based on results shengthen --.1 Collect and analyze data underlying I theory I f-----------l l*v I L- Hypothesis Hypothesis is supported is rejected I I I Figure 3.1 Typical experimental approach. 71  Chapter 3: Designing and Interprcting Experiments Scientific hypotheses are tested by performing experiments in which researchers deliberately manipulate one or more variables while measuring the effect of this manipulation on a different variable. In properly designed experiments, factors other than the deliberately manipulated variable(s) that could affect the outcome of an experiment (ie, extraneous variables) are controlled, or the effects of such variables are minimized by randomization. To that end, scientific experiments are characterized by the random assignment of test subjects (eg, research animals, human participants) to treatment groups and control groups. A treatment group (also referred to as an experimental group) is exposed to the treatment under study in an experiment, whereas a control group is not subjected to the factor being investigated. By comparing the results observed in the treatment group with those observed in the control group, researchers can determine the extent to which the observed experimental outcome is attributable to the manipulated variable. Certain aspects of experimental design can be optimized to avoid the introduction of systematic error (ie, bias) in experiments involving human subjects. In a single-blind study, the research subjects do not know whether they have been assigned to a treatment group or control group. In a double-blind study, neither the research subjects nor the researchers conducting the experiment know which subjects have been assigned to treatment and control groups. This concealment (blinding) of the allocation of subjects to treatment and control groups helps prevdnt bias caused by human expectations, as does the use of placebos in control groups. gJg?*ryp^esslY.s:r"+"91_es An experimental variable is any factor that can change (ie, take on different values) in an experiment. An independent variable (lV) in an experiment is a factor that is purposefully varied (ie, manipulated) by a researcherto determine its effect on another variable. An lV is independenf in that its variation does not occur in response to another variable (ie, does not depend on another variable). lVs are also referred to as explanatory variables or predictor variables because lV variation is presumed to explain and/or predict change in ahother variable that is measured to determine the outcome of the experiment. A dependent variable (DV) in an experiment is a factor measured for the purpose of observing the experiment's outcome. Consequently, DVs are also called outcome variables or response variables, and changes in a DV typically happen only after manipulation of an lV has occurred in the experiment. In addition to lVs and DVs, experiments typically have multiple controlvariables (which are also known as controlled variables). A control variable is a factor, which could affect the DV, that is held constant in the experiment by the researcher. Figure 3.2 summarizes the characteristics of the main experimental variable types as applied in the context of an experiment to test memory. ts Chapter 3: Designing and Interpreting Experiments For example, in an experiment designed to assess how the diffculty of a vocabulary list affects the number of words participants can remember: Independent variable Dependent variable (explanatory variable, (outcome variable, predictor variable) responsevariable) i is manipulated by the researcher is the measuredout@me : i Vocabulary list word dfficulty Number of words recalled i Simple wotd grqjp WordA WordA./.,.. Word B Word B./ Word C Word C "/ Word D Word D Word E Word E / Control variables are held constant by the researcher lo help ensure that the cfran{es in the dependent variable are due only to the manipulations of the independent variable Room temp€rature, lighting, subject age, etc $& rii & ,ir, t Figure 3.2 Independent, dependent, and control variables. Experimental variables can de classified into two main groups by statisticaltype. Quantitative variables have numerical values that represent quantities (ie, amounts or counts), such as the length of an object or the number of offspring produced. Gategorical variables have values assigned to a limited number of distinct categories (ie, groups) based on some characteristic, such as blood type or educational level. Both quantitative variables and categorical variables can be further classified into subtypes. Quantitative variables are classified as being either continuous or discrete. Gontinuous quantitative variables can assume a potentially infinite number of numerical values obtained practically only by performing measurements (eg, height). Alternately, discrete quantitative variables take on only certain numerical values (eg, integers) that are determined by counting (eg, number of vertebrae). Categoricalvariables may be nominal or ordinal. Nominal categoricalvariables describe characteristics grouped rnto categories that do not have a natural order but are simply distinguished by arbitrary names (eg, round, wrinkled). Alternately, ordinal categoricalvariables describe characteristics grouped into three or more categories that exhibit an intrinsic order (eg, low, medium, high). Figure 3.3 depicts experimental variable categorization by statistical type. 73  Chapter 3: Designing and Interpreting fixperiments Variables , fiategorical varisbles Take on numerical Take on values that are values that represent assigned to a limited quantities number of categories Nominal fJrd!n*l variables var'i, hl*s Can assume any Can assume only Grouped into Grouped into three numerical value certain numerical categories that or more categories within a range of values, sich as lack a natural that have an values (obtained integers (obtained order (categories intrinsic order (eg, by performing by performing are defined by low, medium, high) measurements) counts) arbitrary names) Examples: Examples: Examples: Examples:. Mass. Population size. Fur color. Pain level. Temperature. Number of teeth. Place of birth. Performance at Olympic Games Figure 3.3 Categorization of experimental variables by statistical type' : m Concept Check 3.1 For the following variables, identify the statistical type (ie, quantitative or categorical) and subtype (ie, continuous, discrete, nominal, or ordinal). blood type, pH, test grade (letter grade), reaction time, cell count $olution Note: The appendix contains the answer' 3.1.03 Relationships Among Variables The purpose of most biological research is to investigate relationships among variables. In particular, (ie, biological research that involves experimentation seeks to determine cause-and-effect relationships causality) among vartables. lf a relationship exists between two variables, then the two variables correspond to each other in some way. Correspondence that is in the form of a linear relationship between variables (ie, the variables puritiv change together at a uniform rate) is called correlation, which may be positive or negative. In a 1ie, direct correlation), the change in both variables is in the same direction, whereas, in a "orro.lution negative correlation (ie, inverse correlation), the direction of change of one variable is opposite that of tl" other variable. The strength of a correlation between two variables is represented by a correlation coefficient (eg, pearson's correlation coefficient), which ranges from -1 to +1, with -1 indicating a perfect negative 74  Chapter 3: Designing and lnterprethrg Experiments between the variables and +1 indicating a perfect positive correlation. A correlation coefficient 0 indicates that no linear relationship exists between the variables. Visualization of correlations variables requires that experimental data be commonly displayed in scatterplots. Figure 3.4 representative scatterplots and associated correlation coefficients (r), which quantify the strengths directions of the correlations. No conelation -1.0 0 Strong Weak Weak r. t 3.4 Representation of correlation strength and direction. I existence of a correlation befoeen two variables indicates that the variables are statistically ted; however, the existence of a correlation does nof indicate whether a change in one variable the change in the other variable. ln other words, the existence of a causal relationship (ie, cause- #ect relationship) between variables cannot be inferred solely based on an observed correlation, if the correlation is strong. Figure 3.5 shows an example of two variables that are strongly but that do not exist in a causal relationship.  Chapt*r 3: ilcsigning un