MCAT Physics and Math Review
Chapter 11: Reasoning About the Design and Execution of Research
1. An experimenter is attempting to investigate the effect of a new antibiotic on E. coli. He plates cells and administers one milliliter of the antibiotic. Which of the following is an appropriate negative control in this experiment?
1. A plate with no cells that was coated with one milliliter of antibiotic.
2. A plate with E. coli and no additional treatment.
3. A plate with E. coli and one milliliter of isotonic saline.
4. A plate of epithelial cells treated with one milliliter of antibiotic.
2. Which of the following would best establish a causal link?
1. A cross-sectional study using survey data for hand-washing and colds.
2. A case–control study of an exposure during childhood and development of a certain disease later in life.
3. A randomized clinical controlled trial of a new antipyretic drug.
4. An IQ test where the results are later segregated by gender.
3. An experimenter is attempting to determine the internal energy of a well-known compound He cleans his glassware, completes the synthesis, calibrates a bomb calorimeter, and then uses it to measure the appropriate thermodynamic values. Which of the following errors did he make?
1. He did not determine if the compound was novel or if the information has already been determined.
2. He did not have a specific goal at the beginning of his research.
3. He should not be involved in both the synthesis of the compound and later testing.
4. He should have calibrated the calorimeter before the synthesis of the compound.
4. A researcher wishes to generate a parameter for American women’s mean weight. Which of the following is the most significant concern?
1. Measuring a person’s weight may have psychological consequences and is unethical.
2. Gathering all of the necessary study participants would be prohibitive.
3. Knowing the average weight of women does not provide any useful information.
4. Enough studies have already been conducted on this topic to render it unnecessary.
5. A cross-sectional study in which current smoking status and cancer history are assessed simultaneously cannot satisfy which of Hill’s criteria?
6. After randomization, it is discovered that one group in a study has almost twice as many women as the other. Which of the following is an appropriate response?
1. Move men and women between groups manually so that they have the same gender profile.
2. Check the randomization algorithm; if it is fair, continue with the research.
3. Eliminate all of these subjects because of potential bias and randomize a new cohort.
4. Keep the current cohort and continue randomizing subjects until the gender profiles are equal.
7. An experimenter is attempting to determine the effects of smoking on very low birth weight (VLBW) and of VLBW on IQ. Which of the following statements is correct?
1. Smoking is an independent variable.
2. Smoking is a dependent variable.
3. VLBW is an independent variable.
4. VLBW is a dependent variable.
1. I only
2. II and IV only
3. I, III, and IV only
4. II, III, and IV only
8. A study is performed on a new medication. Subjects in the experimental group are told about the potential side effects of the medication, while subjects in the placebo group are not. The subjects have no contact with each other and do not know in which group they are placed. The side effects end up being significantly more severe in the treatment group, when seen by the same assessor physician. This is most likely caused by which of the following?
1. Physician unblinding only
2. Patient unblinding only
3. Both physician and patient unblinding
4. Both physician and patient blinding
9. Which of the following relationships between measurement error and overall error is correct?
1. Unreliable data leads to confounding.
2. Invalid data leads to confounding.
3. Unreliable data leads to bias.
4. Invalid data leads to bias.
10.A new study of a weight loss drug uses a radio advertisement to generate study participation. What type of error is most likely to result?
1. Hawthorne effect
2. Selection bias
4. Detection bias
11.A researcher designing a study has paid for it to be professionally translated into several languages. She discusses the potential risks and benefits with each participant and allows them to bring documentation home for review before committing to the study. This researcher has put special focus on:
1. justice by explaining potential unnecessary risks.
2. beneficence by describing the potential benefits of the study.
3. respect for persons by acknowledging the subject’s perspective and rights.
4. selection bias, by making the recruitment documents inclusive.
12.Which of the following methods would be most appropriate for an initial assessment of hemoglobin saturation during an experiment about breath holding?
1. A pulse oximeter, which uses a small light on an adhesive bandage.
2. An arterial cannula, which permits repeated blood draws with a single puncture.
3. Repeated venipuncture, because a single puncture would cause data overlap.
4. A Swan–Ganz catheter inserted through the femoral artery, which can measure saturation nearest the heart.
13.A medical student attempting to impress her attending physician refers to a recent article that says that there is a statistically significant difference in pregnancy length with a new therapy. Which of the following is most likely to be a valid criticism of the article?
1. Medical students usually reference articles in the newest journals, which may be unreliable.
2. The effect didn’t change patient outcomes, only a secondary measure.
3. There is a lack of internal validity in the results, despite significance.
4. Selection bias is inherent in the scientific process.
14.Which sample would be the most appropriate participants for a study on hormone replacement therapy for postmenopausal symptoms?
1. Prepubescent girls
2. Premenopausal adult women
3. Pregnant women
4. Postmenopausal women
15.Use of a colorimetric assay to determine protein concentration may be subject to all of the following EXCEPT:
1. the use of standards.
2. measurement error.
3. the Hawthorne effect.
4. systematic error.
Answers and Explanations
The purpose of a control is to keep the conditions of two experiments as close as possible to establish causality. In this case, the one milliliter volume addition might have impacted the growth of E. coli; thus, we must control for this by administering an equal volume of a theoretically inert compound to a plate of E. coli.
An experiment will always establish a clearer causal link than an observational study. Choices (A), (B), and (D) are all examples of observational data.
The experimenter has not completed the initial phases of research. There was no data acquisition or refinement, and there was no indication that the question required an experiment to be answered. Were the experimenter doubtful of the validity of the reported value, an experiment could be appropriate—but there is no information to indicate that this is so. Based on the question stem, it is clear that the experimenter had a clear goal, eliminating choice (B). In human subjects research, tasks may be divided to facilitate blinding, but this is generally unnecessary in basic sciences research, eliminating choice (C). As long as the calorimeter was calibrated prior to its use, it does not matter when this calibration occurred relative to the synthesis of the compound, eliminating choice (D).
A parameter is a population measure, so to calculate it, every single member must be measured. Identifying, measuring, and recording data for a population that large—over 150 million—is essentially impossible. Common biometric measures, if not misused, generally neither cause significant psychological harm nor are unethical, eliminating choice (A). Knowing a mean weight could have major ramifications, including public health measures, medical recommendations, and shifting of body image, eliminating choice (C). While a number of studies on weight have been performed, there is not yet a parameter describing the entire population, eliminating choice (D).
Because both the exposure and outcome are measured at the same time, we cannot make any conclusions about temporality. The cancer patient may have only begun smoking after diagnosis, and this type of study doesn’t examine that possibility.
Randomization is based on the idea that the results will only vary as a result of random chance as long as the assignment is proper. The appropriate response to a fair algorithm that assigns groups in an unexpected way is to proceed with the research. Participants should never be assigned by the researcher, nor should one continue randomizing samples to achieve a desired outcome—these are likely to introduce more error than leaving unequal groups, eliminating choices (A) and (D). It is unnecessary to drop this entire cohort, assuming the randomization algorithm was fair, eliminating choice (C).
Two relationships are being assessed. In the relationship between smoking and very low birth weight (VLBW), smoking is the independent variable and VLBW is the dependent variable. In the second relationship, VLBW is being compared to IQ. VLBW is the independent variable here, while IQ is the dependent variable.
Because the same physician sees both the control groups and the experimental groups, there is the potential for the physician to realize which of the groups is receiving which treatment—especially if the subject mentions expected side effects. In this study, patients were not told which group they were in, but if they were assigned to the medication group, they were told about its side effects. If the patients talked to each other they could experience patient unblinding, but we are told they have no communication with each other.
Data that is off in a systematic way (reads at a value that is not the true value) will cause bias. This type of data error is an example of a lack of validity (or accuracy). Unreliable data suffers from random, not systematic, error. Confounding arises from errors in data analysis, not data collection.
Requiring subjects to volunteer for a study and to seek the study out will introduce selection bias. The people who end up volunteering listen to the radio, which the general population may not, and are interested in the topic and willing to volunteer. Most studies suffer from selection bias and it is the most common impediment to generalizability.
The behaviors described in the question stem (informing the patient, providing time to make a decision) are consistent with informed consent and autonomy, which are both part of respect for persons.
While it may appear that this question is asking to determine which method is the most accurate or reliable, this is an ethics question. All of these methods (with the exception of venipuncture, choice (C)) measure oxygen saturation. According to the principle of beneficence, we must minimize potential harms associated with our investigations, thus the noninvasive pulse oximeter should be greatly favored over other measurements for an initial assessment.
Statistical significance is not the same as clinical significance. There are medications that increase the length of pregnancy in preterm labor, but some only do so for a few hours. Because this may not impact patient outcomes, it should not inform treatment decisions.
Samples should always be taken from the target population or population of interest. Given that the target population is postmenopausal women, this group should be used for the sample. Both prepubescent girls and pregnant women are populations with special precautions against coercion, and are not appropriate for this study, eliminating choices (A) and (C). Premenopausal women are unlikely to require hormone replacement therapy unless they have a condition that specifically necessitates it, eliminating choice (B).
The Hawthorne effect—a change in behavior as a result of the knowledge that one is being observed—is only present with human subjects. Basic science research generally does not suffer from the Hawthorne effect.