Experimental Design Flashcards
Experimental design is perhaps the single most important topic in MCAT P/S. Use this deck to master types of variables, reliability and validity, relevant statistical concepts, and more.
In an experiment, which variable is manipulated by researchers?
independent variable
The independent variable is the variable that is controlled or changed by the experimenters. The goal is to assess whether these manipulations impact the dependent variable.
In the graphed results of a typical experimental study, which variable is on the y-axis?
dependent variable
The dependent variable is the variable that changes in response to the independent variable. This variable is usually plotted on the y-axis, which helps visualize the magnitude of the changes.
In an experiment, which variable is measured?
dependent variable
The dependent variable is the variable that is measured by the experimenters. The goal is to assess how differences in the dependent variable correspond to the intentional changes made by the researchers to the independent variable.
In the graphed results of a typical experimental study, which variable is on the x-axis?
independent variable
The independent variable is the variable that is manipulated by the experimenters. Since this variable is typically varied in consistent increments, it makes sense to plot it on the x-axis.
A researcher in Singapore is studying the impact of PTSD on burnout rate in police officers. If these study results were graphed, what would be on the y-axis?
Burnout rate
Here, burnout rate is being measured, making it the dependent variable. The dependent variable is typically plotted on the y-axis.
A researcher in Singapore is studying the impact of PTSD on burnout rate in police officers. If these study results were graphed, what would be on the x-axis?
PTSD status
Here, PTSD status is something that the experimenters believe may influence the dependent variable (burnout rate). Therefore, PTSD status is the independent variable, which is typically plotted on the x-axis.
True or false.
Experiments can have only one independent and one dependent variable.
False
Experiments, including on the MCAT, can have multiple independent and/or multiple dependent variables. This is usually clear from the results, which (if in graph form, for example) might show multiple bars at each data point or have multiple x- or y-axes.
In an experiment, which type of variable can explain the relationship between the independent and dependent variables?
mediating variable
As its name implies, a mediating variable essentially “goes between” two other variables to explain their relationship. For instance, imagine that an increase in variable A leads to an increase in variable C (put symbolically, A → C). If this is found to be true because an increase in A leads to an increase in B (A → B), which then leads to an increase in C (A → B → C), then variable B is a mediating variable.
In an experiment, which type of variable can explain variations in the strength of the relationship between the independent and dependent variables?
moderating variable
This concept can be tricky! Just imagine that an increase in an independent variable corresponds to an increase in a dependent variable. Under certain circumstances, it might correspond to a large increase, but under others, it might only produce a small increase. These circumstances, which “moderate” the strength of the relationship, are moderating variables.
What is socioeconomic status (SES) called when it impacts both the independent and dependent variables in a study, even if it is not the focus of the study?
confounding variable
This relationship can cause researchers to erroneously believe that the independent and dependent variables are related, when in fact their results are due to the presence of the confounding variable.
True or false.
If a researcher is studying the impact of the provision of medical information on patient adherence to doctor advice, and gender is found to play a role, gender must be a confounding variable.
False
It’s easy to assume that just because gender isn’t being studied (that is, it isn’t the independent or dependent variable), that it must be a confounding variable! However, that is not accurate; it could be a mediating or moderating variable instead.
A study is designed to assess the impact of dog ownership on stress. One post-bac student believes this study is flawed because it does not also assess the impact of cat ownership. Is the student correct?
no
This question relates to study scope. A study design does not need to include the assessment of every possible factor, including those outside its scope! Here, the study aims only to assess dog ownership; adding another independent variable of cat ownership would fall outside this scope.
Imagine that across the country, flights tend to be delayed more on Mondays than on any other day. However, in the southwest U.S., flights are delayed far more often on Mondays, while in the eastern U.S., flights are only delayed slightly more often. Here, region of the country is what kind of variable?
A moderating variable
In this example, the region of the country “moderates” (or affects the strength of) the existing relationship between the day of the week and the likelihood of flight delays.
In a study of the influence of poverty on alcoholism, poverty is found to be associated with stress, which in turn is associated with increased rates of alcoholism. Here, what kind of variable is stress?
mediating variable
Here, stress explains the relationship between poverty and alcoholism, making it a mediating variable.
A Ph.D. student develops an experiment in which she measures the impact of eating breakfast on both cortisol levels and subject-reported alertness. In this study, whether a subject ate breakfast is what kind of variable?
independent variable
Since the Ph.D. student is evaluating the effects of eating breakfast, breakfast-eating status is the independent variable. It is likely that certain subjects are told to eat breakfast and other subjects to skip it.
A Ph.D. student develops an experiment in which she measures the impact of eating breakfast on both cortisol levels and subject-reported alertness. In this study, how many dependent variables are present?
two
Here, both cortisol levels and subject-reported alertness are being measured by the researcher in response to changes in the independent variable. Therefore, both are dependent variables.
True or false.
Age is a common confounding variable.
True
As age has extremely widespread effects, it often correlates with independent and dependent variables even when it is not being studied itself. Therefore, it is a classic confounding variable, and studies often use age-controlled/same-age subjects to control for it.
What is the term for causality in which variable A impacts variable B, but variable B has no effect on variable A?
unidirectional causation
This is common across the board in scientific research, to the extent that we usually assume relationships to be unidirectional unless indicated otherwise.
What is the term for causality in which two variables each impact the other?
reciprocal causation
While less common than unidirectional causation, this certainly still exists. For example, if researchers are evaluating the impact of depression on alcoholism, they may need to consider that alcoholism may also worsen some of the symptoms of depression.
What is the term for the group that is compared to the experimental group in nearly all scientific experiments?
control group
Virtually all well-designed studies include at least one control group, which is a group included for the purpose of comparison to the experimental group(s). Control groups allow the researchers to minimize the impact of confounding variables.
A control group that is expected to exhibit no change (in other words, that represents no effect) is what kind of control?
negative control
For example, a negative control group may simply not be treated with anything, while the experimental group receives the experimental treatment.
A control group that is expected to display a result based on existing understanding about the results in that group is what kind of control?
positive control
For example, if experimenters are testing whether a new treatment decreases levels of an enzyme, they may compare it with a control treatment that is already known to decrease levels of that enzyme.
Fill in the blank.
In human studies, researchers often go to great lengths to ensure that the negative control group believes that they may be receiving the experimental treatment. This best relates to the ________ effect.
placebo
The placebo effect refers to the impact of simply believing that one is receiving a treatment. As such, the negative control group is typically given a placebo treatment (such as a sugar pill) rather than being given nothing.
What does the epidemiologist ensure by carefully writing the Methods section of his paper so other researchers can understand and repeat his experiment?
replicability
It is crucial that experiments be replicable, meaning that other scientists can repeat the experimental method and steps and see if they get the same results.
What term describes an experiment that can be repeated and yields the same results?
reliable
Reliability can apply to the same researchers repeating their experiment on the same subjects, to the same researchers repeating the experiment on different subjects, or to entirely different researchers attempting to repeat the experiment.
Fill in the blank.
A team of scientists in Stockholm are able to repeat an experiment originally conducted by a team in Atlanta. However, the results are dramatically different. This experiment was ________, but not ________.
replicable, reliable
Since the experiment could be repeated at all, it was replicable. However, since its results were found to be inconsistent, it was not reliable.
Define:
validity
In the context of research, validity is the extent to which a study truly measures what it intends to measure and has results that are applicable outside the exact circumstances of the study.
Put simply, validity can be thought of as the extent to which a study’s results are genuine and generalizable.
Which type of validity refers to the soundness of a study with regard to being able to support claims of causality?
internal validity
If a study is internally valid, that means that if a causal claim is determined (as in, “thing A causes thing B”), that causal claim is likely to be accurate/sound.
In a research study, the best way to improve internal validity is to do what?
Control for confounding variables and rule out sources of bias.
Since internal validity refers to the ability to draw accurate causal conclusions from a study, it is important to reduce or eliminate the impact of confounding variables or bias, both of which can lead to inaccurate interpretations of causality.
Which type of validity can be thought of as the extent to which a study’s results are generalizable?
external validity
Unlike internal validity (which deals with the soundness of causal conclusions), external validity refers to the extent to which a study’s results can be generalized to contexts outside the specific circumstances of the study.
What are three subtypes of test validity in research, which refers to the extent to which a test or experiment measures what it intends to measure?
- criterion
- construct
- content validity
As one might predict from its definition, test validity is very broad. As such, it is sometimes divided into these three slightly more specific subtypes.
Which of the following is not typically considered to be part of test validity?
- Construct validity
- External validity
- Criterion validity
External validity is not part of test validity. Rather, external validity measures something separate (the generalizability of the study).
However, both construct validity and criterion validity are types of test validity.
Define:
criterion validity
It refers to the extent to which the results of a given test correspond to those of another well-respected, established, and/or relevant measure.
For instance, imagine that job applicants to a large tech firm are typically given a six-hour exam. If the leaders of that firm decide they want to replace it with a 30-minute exam, they will likely first confirm that the results on the shorter exam correspond with the criterion of the longer test.
Define:
construct validity
It refers to whether a given test accurately evaluates the construct it was developed to evaluate.
In this context, a “construct” is a variable that is being assessed but cannot be directly measured or observed. For example, a test may measure the construct of stress, but to do so, the experimenters must evaluate proxies of stress such as heart rate or cortisol levels.
Define:
content validity
It refers to how well a given test actually evaluates the full scope of what it was designed to test.
As such, a test with low content validity might test only a tiny part of the larger phenomenon it was developed to evaluate.
What is the difference between content and construct validity?
- Content validity evaluates whether a test covers the full scope of a construct.
- Construct validity determines if the test measures the correct construct.
Define:
predictive validity
It refers to the extent to which the results of a given test correspond to results on some future measure.
For instance, if high MCAT scores typically correlate with low medical school dropout rates, then MCAT score has high predictive validity with regard to dropout rate.
What is the difference between criterion and predictive validity?
- Criterion validity assesses how well a test correlates with an established measure.
- Predictive validity evaluates how well a test predicts future outcomes.
True or false.
An experiment that produces very different results each time it is repeated has low external validity.
False
Whether an experiment produces the same results over multiple identical administrations is reliability, not external validity. External validity refers to the extent to which a study’s findings can be generalized to different situations.
Fill in the blank.
The final exam in Professor Jones’ chemistry class is meant to test information from the entire semester, but Professor Jones is in a weird mood and includes only questions about stoichiometry. This final exam has low ________ validity.
content
Professor Jones’ final exam fails to test the entire scope of content covered by the class; therefore, it exhibits low content validity.
Fill in the blank.
An experiment on heart rate and PTSD in veterans involves the subjects putting on virtual reality headsets and watching scenes of everyday life, such as visiting a supermarket, with the goal of mimicking everyday life. If the veterans’ heart rates are found to be dramatically higher in actual supermarkets, this study lacks ________ validity.
external
External validity refers to the extent to which the results of a study can be generalized to other situations, often real-life ones. Here, the study lacks externally valid results regarding heart rate.
Fill in the blank.
In many ways, improving the validity of a research study involves trade-offs. For example, as the internal validity of a study is improved, the ________ validity often decreases.
external
External validity refers to the extent to which a study’s results can be generalized to other situations or the real world. Unfortunately, the real world is full of confounding variables, which must be minimized to increase internal validity. Therefore, a common struggle in research design is that increases in internal validity bring with them decreases in external validity.
Fill in the blank.
Controlling for the effects of demographic factors, such as gender and socioeconomic status, is often done with the goal of increasing ________ validity.
internal
Internal validity refers to the ability to draw accurate causal conclusions from a study. Importantly, this is facilitated by properly controlling for potential confounding variables.
Fill in the blank.
An inept teacher intends to test his students’ performance on eighth-grade algebra but actually gives them a spelling test by mistake. Putting the ridiculousness of this error aside, if the teacher decided to still use the spelling test results as a metric of math ability, this reflects a lack of ________ validity.
construct
Here, the teacher is assessing the entirely wrong construct: spelling ability instead of math ability.
Fill in the blank.
The SAT is meant to be indicative of success in college. If it succeeded in that goal, it could be said to have ________ validity.
predictive
Here, college success is a future measure, meaning that predictive validity is the most accurate.
If the results of a new IQ test strongly correlate with the results of the previously accepted, “gold-standard” measure of IQ, the new test has which type of validity?
criterion validity
Here, the results of the new IQ test correspond well to the existing criterion of the gold-standard IQ test.
The use of positive controls best relates to which form of validity?
criterion validity
Essentially, a positive control is an existing criterion to which researchers can compare their new treatment. This therefore relates to criterion validity.
Researchers want to determine whether viral infection decreases levels of a specific protein in cells. How might these researchers distinguish a decrease in the level of this protein alone from a decrease in the level of total protein in the infected cells?
They can normalize the data against total protein.
Often, the results in MCAT experimental passages will contain data that has been “normalized,” or scaled in comparison to a given standard. In the example given here, normalization against total protein will allow the researchers to see whether the decrease in the protein of interest has happened proportionally to a decrease in total protein or whether this protein in particular was impacted.
Fill in the blank.
A(n) ________ measure is one that is unbiased and is based on facts or numbers.
objective
Objective measures are (at least under typical circumstances) unbiased, meaning that they cannot be interpreted differently based on the opinions of the interpreter. A person’s weight is an example of an objective measure.
Fill in the blank.
A(n) ________ measure is one that is subject to opinion. Open-ended questions on surveys exemplify this type of measure.
subjective
Subjective measures are those that are “subject” to opinion, such as open-ended questions and ratings of feelings or perceptions.
What is the difference between quantitative and qualitative measures?
- Quantitative methods produce numbers as results.
- Qualitative methods produce results that are not numbers.
As such, numerical data can be termed quantitative data, while non-numerical data is known as qualitative data.
A study that produces both numbers and verbal descriptions as results is utilizing which methods?
mixed methods
If you answered “quantitative and qualitative methods,” you’re also correct! However, it is important to know that “mixed methods” refers to the use of both quantitative and qualitative methods in a study.
A political poll is conducted in which pollsters ask subjects which political candidate they plan to vote for. This poll is gathering what kind of data?
Choose from qualitative or quantitative data.
qualitative data
All non-numerical data is qualitative. This includes observations, descriptions, words, or phrases, such as the verbal description of the subject’s preferred candidate in this example.
A political poll is conducted in which subjects are asked to rate the perceived importance of a number of issues from 1 to 10 and then to provide a 5-word-or-fewer description of the first thing they think about when they hear about each issue. This study is utilizing which methods?
Choose from qualitative, quantitative, or mixed methods.
mixed methods
Since this study is collecting both numerical (ratings from 1 to 10) and non-numerical (short descriptions consisting of words) data, it is a mixed-methods study.
Fill in the blank.
Ratings of the importance of a political issue from 1 to 10 are both _________ [qualitative/quantitative] and _________ [subjective/objective].
Choose one term from each box above to correctly complete the sentence.
quantitative, subjective
Importantly, just because a study collects quantitative data does not mean that it is an objective study! Here, the numbers being collected are subject to opinion, making them subjective.
Fill in the blank.
Descriptions of the pain a subject is feeling with words like “extreme,” “stinging,” or “unbearable” are both _________ [qualitative/quantitative] and _________ [subjective/objective].
Choose one term from each box above to correctly complete the sentence.
qualitative, subjective
Here, since these descriptions are words (and, as far as we know, do not correspond to any numerical scale), they are qualitative. They are also subjective, since different people experience and describe pain very differently.
Fill in the blank.
When a subject is asked to count the number of dots shown in a clear line of dots on a page, he or she is being asked to provide _________ [subjective/objective] information.
Please choose one term from the box above to correctly complete the sentence.
objective
Here, there is only one right answer: the correct number of dots. This makes this information objective, not subjective. Of course, the subject could always count incorrectly, but that does not make this subjective; rather, to be subjective, different answers would need to be valid depending on the subject’s opinions.
Define:
precision
It refers to the degree to which multiple measurements are similar to each other.
For example, if a person measures the volume of a sample and obtains results of 1.01, 1.02, and 1.00 mL, these results are precise, even if they are nowhere near the actual volume of 0.50 mL.
Define:
accuracy
It refers to the degree to which measurements are similar to the correct value.
For example, if a person measures the volume of a sample and obtains a result of 0.50 mL, and the actual volume is 0.50 mL, then this result is extremely accurate.
What can be said about the reliability and validity of census conducted by Anthropologists results showing 101,592, 101,594, and 101,595 people if the actual population is 101,593?
Choose from “accurate,” “precise,” or “both accurate and precise.”
both accurate and precise
Since these measurements are both very close to each other and very close to the actual value, they are both accurate and precise.
In the context of scientific research, is reliability more similar to accuracy or precision?
precision
Reliability refers to whether repeating a study yields similar results to the original study, while precision refers to whether repeating a measurement yields similar results to the original measurement.
In the context of scientific research, is validity more similar to accuracy or precision?
accuracy
Validity refers to whether the results of a study were genuine and actually reflect what they were designed to measure. This is analogous to accuracy, which refers to whether a measurement actually reflects the true value of that meaurement.
How do reliability and validity differ from precision and accuracy?
- Reliability and validity refer to the methods and design of an entire study.
- Precision and accuracy refer to a given set of measurements or data.
What should a study ideally demonstrate in terms of reliability and validity?
Choose from valid but not reliable, reliable but not valid, both valid and reliable, or neither valid nor reliable.
both valid and reliable
This means that the study results are both reflective of the accurate/real-life results and that the results will be similar if the study is repeated over time. Both are desirable characteristics of a study.
What is the validity and reliability of a maternal and child health study in which the results appeared realistic and closely correlated with established measures, but differed significantly when repeated?
Choose from valid but not reliable, reliable but not valid, both valid and reliable, or neither valid nor reliable.
valid but not reliable
The description makes it clear that this study was valid, at least to a degree (note the implied reference to criterion validity!), but since its results were dissimilar upon repetition, it was unreliable.
What is the validity and reliability of a study that poorly correlates with the gold-standard index, only tests a portion of its intended scope, and produces inconsistent results when replicated?
Choose from valid but not reliable, reliable but not valid, both valid and reliable, or neither valid nor reliable.
neither valid nor reliable
The first portion of this description indicates a lack of validity (specifically, both criterion and content validity), while the second indicates a lack of reliability.
What is the replicability and validity of an experiment with extremely flawed methods that are described in excruciating step-by-step detail?
Choose from valid but not reliable, reliable but not valid, both valid and reliable, or neither valid nor reliable.
reliable but not valid
Since the methods are described extremely carefully, it is likely that when other researchers attempt the experiment, they will get similar results (constituting reliability). However, since those methods are highly flawed, the experiment is not valid.
What is the empirical system of learning and evaluating information that has been used in science for at least the past 200 years?
Hint: It is often depicted as a series of steps, including the formulation of a hypothesis.
scientific method
You don’t need to memorize the steps of this method for the MCAT, as it is meant to be more of a cohesive system of practice than a rigid step-by-step manual.
What must any hypothesis developed as part of the scientific method be?
testable
This is a critical fact about the scientific method! If a hypothesis is not testable, it effectively doesn’t matter that there even is a hypothesis, because the subsequent steps of the method cannot be followed and the hypothesis cannot be validated or undermined.
What are the two types of errors that can arise in research, statistically?
Type I and Type II errors
That’s easy enough to remember, but you should also understand what these errors are. Type I errors are often termed “false positives,” while type II errors are known as “false negatives.”
An individual goes into a clinic for a liver biopsy. The biopsy showed no cancerous cells, but unfortunately, the individual did have liver cancer. This exemplifies what type of error?
Choose from a type I or a type II error.
type II error
A type II error, also known as a false negative, occurs when a phenomenon or relationship actually is present, but testing fails to detect that phenomenon or relationship (here, cancer).
A woman visits her gynecologist for a mammogram. The mammogram initially showed results that were indicative of breast cancer, but further testing showed the woman to be cancer-free. Here, the mammogram exemplifies what type of error?
Choose from a type I or a type II error.
type I error
A type I error, also known as a false positive, occurs when a phenomenon or relationship is absent, but testing produces an inaccurate result indicating that it is present.
True or false.
Researchers and medical professionals should strive to minimize type II errors.
True
This is absolutely accurate! The occurrence of type II errors, or false negatives, mean that we mistakenly believe that a relationship or medical phenomenon (for example, cancer) is absent. This can have catastrophic results in any field.
Fill in the blank.
Researchers and medical professionals should strive to maximize ________.
Choose from type I errors, type II errors, both, or neither.
neither
Researchers and medical professionals should not strive to maximize any type of error! It’s easy to think that type II errors (false positives) are less catastrophic than type I errors, at least in a medical context. However, these “false positives” typically create distress and require additional testing that can be expensive, painful, or inconvenient.
What are the two types of methodological errors that can arise in research?
random and systematic errors
These differ from Type I and Type II errors in that they refer to errors in the method of the experiment or its measurements, rather than to the relationship between the results and what exists in reality.
Define:
random errors
Random errors are fluctuations in a measurement, often due to the inherent lack of precision of the measurement apparatus.
If measurements are taken many times, random errors will result in some errors that are below and some that are above the “actual” value.
Define:
systematic errors
These are regular, consistent errors in a measurement, often resulting from miscalibration or from other mistakes that are made for all trials.
If measurements are taken many times, systematic errors will produce results that are always above or always below the “actual” value.
What is the default position in scientific research that assumes no relationship is present or that the independent variable has no effect on the dependent variable?
null hypothesis
Rejection of the null hypothesis therefore means that a relationship is actually present.
What is the type of experimental design in which subjects serve as their own control group?
within-subjects design
In such a design, participants are exposed to both the experimental condition (or conditions) and the control condition. This reduces the impact of confounding variables, as it does not introduce the undesirable variations that come with comparing different individuals.
What is the type of experimental design in which participants in the experimental group are different from those in the control group?
between-subjects design
While such a design may be necessary in certain situations, it does carry the potential of more confounding variables than a within-subjects design.
In a single-blind study design, who is “blinded” to information regarding the experimental vs. control groups?
Only the subjects (participants) are blinded.
In other words, the subjects do not know whether they are in the experimental or the control group. However, the researchers do know which subjects are in which group.
In a double-blind study design, who is “blinded” to information regarding the experimental vs. control groups?
Both the subjects (participants) and the researchers are blinded.
In other words, neither of these groups is informed regarding which subjects are in the experimental vs. the control group.
What is a data point that is wildly different from the points comprising the rest of the dataset termed?
outlier
For instance, if a dataset contains values of 5.5, 5.7, 5.2, and 14, then 14 exemplifies an outlier. Outliers can be indicative of experimental error.
What are the three measures of central tendency?
- mean
- median
- mode
According to the AAMC, these are also the measures of central tendency that you must understand for the MCAT.
What is the measurement of central tendency obtained by arranging the values of a dataset in order of magnitude and identifying the middle value?
median
You may remember using median as early as elementary school, but it is still relevant on the MCAT, particularly with regard to Reasoning Skill 4 (Data-based and Statistical Reasoning).
What is the measurement of central tendency obtained by summing the values of a dataset and dividing by 𝑛, where 𝑛 represents the number of values?
mean
“Mean” is another term for “average.”
What is the value that appears most commonly in a dataset?
mode
For instance, if ten children are sampled about their favorite number between 1 and 10, and six children say 7, two children say 3, and two children say 5, then 7 is the mode.
When assessing salary or income data, which is typically used as the measure of central tendency?
Choose from either the mean or the median.
median
Importantly, the median is far less susceptible to the influence of outliers than the mean. Analyses of salary data often include such outliers. For instance, the mean of a sample that included Jeff Bezos’ salary would be skewed incredibly high, while the median would be much more representative of the sample overall.
True or false.
The mean and median of a dataset can never be the same.
False
An easy way to tell whether this is false is to imagine a fake set of data, such as 3 mg, 5 mg, and 7 mg. Here, the mean and median of the dataset are both 5 mg. Of course, most datasets in reasearch will be much larger, but the same principle holds true.
What are three measures of dispersion used in the context of research data?
- range
- interquartile range
- standard deviation
According to the AAMC, these are also the measures of dispersion that you must understand for the MCAT.
In the context of a dataset produced as part of a research study, what is the range?
It is the numerical difference between the highest and lowest values in the dataset.
For instance, if the highest value was 95 ppm and the lowest was 17 ppm, the range would be (95 − 17) = 78 ppm.
What are the four groups called when a numerical dataset is organized by magnitude and divided into equal parts?
quartiles
For instance, a dataset that included every number from 1 to 100 would have quartiles of 1-25 (lowest), 26-50, 51-75, and 76-100 (highest).
What term describes the numerical difference between the highest value of the lowest quartile and the lowest value of the highest quartile?
interquartile range
(IQR)
Essentially, the IQR describes the numerical spread covered by the middle two quartiles.
In the analysis of data, why might researchers prefer to use the interquartile range instead of simply using the range?
The interquartile range (IQR) is less sensitive to the presence of outliers.
Since the IQR excludes the values in the highest and lowest quartiles, it would not be affected by, say, one extremely high or one extremely low value. In contrast, the range would be impacted.
What is the common indicator of variation in a data set that is mathematically equal to the square root of the variance?
standard deviation
Note that you don’t need to understand the ins and outs of the mathematical calculation of standard deviation for the MCAT, but you should know that it exists and how to interpret it in a figure/table.
True or false.
A high standard deviation indicates that values are relatively spread out, or far from the mean.
True
Standard deviation is a measure of the variation in a data set with respect to the mean. A high standard deviation means that values are spread out (they vary) rather than being clustered near the mean.
The probability of rejecting the null hypothesis relates most closely to what concept?
statistical significance
A statistically significant result is one that is extremely unlikely to have arisen if the null hypothesis were true. As such, statistical significance is tied to the rejection of the null hypothesis.
A result is deemed statistically significant if its probability value (p-value) is less than what threshold?
Please answer this question with the name of the threshold, not with a number.
alpha level
This threshold is termed the alpha level. In most studies, the alpha level is set at 0.05, meaning that p-values < 0.05 indicate statistical significance.
In a study of stress and performance on a memory test, a p-value of 0.49 was determined. Does this constitute a statistically significant relationship?
no
While we were not given a threshold for statistical significance (or alpha level), the typical alpha level is 0.05 (or occasionally 0.10). 0.49, therefore, is much too high to indicate significance. Be sure to read numbers carefully; otherwise, it’s easy to confuse 0.05 with 0.5!
True or false.
If the relationship between viral infection and osteoblast activation is not statistically significant, this means that the experimental (virally infected) group did not have a higher rate of osteoblast activation than the control group.
False
Just because no statistically significant relationship was found does not mean that the experimental group showed no difference at all from the control group! The virally-infected cells may have displayed higher osteoblast activation, just not to a significant extent.
Define:
confidence interval
It is a range of values within which a parameter has a certain probability of falling.
This sounds confusing, so let’s give an example. If the 95% confidence interval for a study of IQ is 88 to 112, then there is a 95% probability that a given subject’s IQ will fall within that range.
Name the effect illustrated by the example below.
Researchers were hired to examine the relationship between light intensity and worker productivity in a factory. Near the conclusion of the study, the researchers found that productivity improved not from the lighting inside the factory, but from the attention the workers received from the researchers.
Hawthorne effect
The Hawthorne effect, also termed subject reactivity, is any change in behavior resulting from the attention study participants perceive they are getting from researchers, rather than resulting from differences in the independent variable(s).
