Gain in Student Understanding of the Role of Random Variation in Evolution Following Teaching Intervention Based on Luria-Delbruck Experiment

Abstract

Undergraduate students in introductory biology classes are typically saddled with pre-existing popular beliefs that impede their ability to learn about biological evolution. One of the most common misconceptions about evolution is that the environment causes advantageous mutations, rather than the correct view that mutations occur randomly and the environment only selects for mutants with advantageous traits. In this study, a significant gain in student understanding of the role of randomness in evolution was observed after students participated in an inquiry-based pedagogical intervention based on the Luria-Delbruck experiment. Questionnaires with isomorphic questions regarding environmental selection among random mutants were administered to study participants (N = 82) in five separate sections of a sophomore-level microbiology class before and after the teaching intervention. Demographic data on each participant was also collected, in a way that preserved anonymity. Repeated measures analysis showed that post-test scores were significantly higher than pre-test scores with regard to the questions about evolution (F(1, 77) = 25.913, p < 0.001). Participants’ pre-existing beliefs about evolution had no significant effect on gain in understanding of this concept. This study indicates that conducting and discussing an experiment about phage resistance in E. coli may improve student understanding of the role of stochastic events in evolution more broadly, as post-test answers showed that students were able to apply the lesson of the Luria-Delbruck experiment to other organisms subjected to other kinds of selection.

INTRODUCTION

Even after taking college-level introductory biology courses, most students continue to hold multiple misconceptions about the process of evolution (1, 10). Prevalent among these misconceptions is a Lamarckian view of evolutionary change rather than a Darwinian one. This view, that change by organisms over time is goal-directed and occurs at the level of the individual, is more intuitively appealing than the correct Darwinian view that selection acts on random variation within a population of individuals (7). Classroom instruction in evolutionary principles has not been shown to be effective in replacing students’ Lamarckian explanations for evolutionary change with Darwinian ones (1, 10). Rather, students with pre-existing Lamarckian ideas about the mechanism of evolution have been shown to try to reconcile their earlier misconceptions with new information they have been taught (1). This cognitive dissonance may deepen students’ confusion about evolution.

Inquiry-based methods of teaching science have been shown to be more effective in increasing student understanding of scientific principles than traditional pedagogical approaches (6). Some have argued that replacing pre-existing misconceptions about science with more accurate, but more counterintuitive, ideas can only occur through such active learning (11). This study tested the hypothesis that an inquiry-based approach to teaching college students about the mechanism of evolution could result in improved student understanding of the role of randomness in evolution.

The phage resistance experiment conducted by S. E. Luria and M. Delbruck in 1943 ultimately convinced the scientific community that the Darwinian explanation for evolutionary change was the correct one, rather than the Lamarckian explanation (5). This experiment is relatively inexpensive and easy to replicate in an undergraduate lab class (2). In this study, a teaching intervention was tested that combines the persuasive power of the Luria-Delbruck experiment with two separate avenues for active learning to dispel students’ Lamarckian misconceptions about evolution and replace them with the more accurate Darwinian view.

METHOD

Participants

Eighty-two students (18 men, 64 women) enrolled in a sophomore-level microbiology course at a small, private, Methodist-affiliated, midwestern, liberal arts college volunteered to participate in this study and completed all phases of the research. Participants ranged in age from 18 to 35 years (mean = 19.87, SD = 2.33), and the majority (90.5%) were of Caucasian ethnicity. Participants were asked to identify religious affiliation and report how often they attended religious services. Over eighty percent (83.2%) self-identified as belonging to a Christian denomination, and 33.7% reported attending religious services at least weekly. Participants were also asked to select from five different statements and choose the one that most closely reflected their beliefs on how life came to be as it is now on Earth. Four students did not answer this question. Of the remaining 78, two-thirds (n = 52) reported some kind of creationist beliefs. Participants’ pre-existing beliefs about the origin of life on Earth are summarized in Table 1 below.

TABLE 1.

Study participants’ pre-existing beliefs about the origin of life on Earth

Statement About Belief of Origin of Life on Earth	% (n) of participants reporting belief (N = 78)
All life, including humans, evolved from earlier life forms over many millions of years.	33% (26)
Most organisms evolved from earlier life forms over many millions of years, but humans did not evolve from other animals.	10% (8)
Kinds of organisms were created, largely in their present form, by God, and that there is only minor change within those kinds over time. This happened many millions of years ago or I have no opinion about when this happened.	36% (28)
Kinds of organisms were created in their present form by God in the past several thousand years. There is only minor change within kinds over time.	21% (16)
Kinds of organisms were created in their present form by God in the past several thousand years, and organisms do not change at all over time.	0% (0)

Study participants were asked to select one statement out of the five listed above that most closely represented their views about how life came to be as it is now on Earth.

Materials

In addition to completing a voluntary informed consent form and a demographic survey, participants also completed two short questionnaires with multiple-choice questions about evolution and genetics (see Appendix, Survey Questionnaires). One questionnaire was administered at the beginning of the semester, prior to the teaching intervention (referred to here as the pre-test), while the other was given at end of the semester, at least ten weeks following the intervention (referred to here as the post-test). Each question relevant to the aims of this study asked students to pick the most likely explanation for evolutionary change described in a short scenario. Of the three answers from which students could choose, two were Lamarckian explanations and one was a Darwinian explanation. More than half of the questions on both the pre-test and the post-test were not relevant to the role of randomness in evolution, and were designed to distract study participants from the aims of this study. Reliability of the questionnaires was acceptable. The pre-test had a Cronbach’s Alpha measure of reliability of 0.774 and the post-test scored at 0.703.

Neither participation in this study nor answers on questionnaires in any way affected students’ grades in the microbiology course. Students were made aware prior to choosing whether or not to participate that the course instructor would have no way of knowing whether particular individuals were enrolled in the study or not.

In addition to answering these questionnaires, participants were asked to participate in normal class activities that were required of all students enrolled in the course.

This study was approved by the Human Subjects Committee of the Institutional Research Review Board of Morningside College.

Teaching intervention

Students in five separate sections of a sophomore-level microbiology class conducted an experiment to determine whether E. coli strain B resistance to T-4 bacteriophages occurs randomly or as a result of exposure to the phages. The protocol for this experiment was adapted from that of the 1943 experiment of S. Luria and M. Delbruck (5) and from D. Bozzone and D. Green’s teaching notes on the Luria-Delbruck experiment (2). Briefly, a single colony of E. coli strain B is used to inoculate a flask of sterile nutrient broth. The same colony is also streaked onto a nutrient agar plate previously inoculated with T-4 bacteriophages to confirm that the progenitor bacterium is susceptible to T-4 phages. After 18 hours of incubation at 37°C, a 10⁻⁵ dilution of this broth culture is made. The 10⁻⁵ dilution is then used to inoculate many test tubes containing sterile nutrient broth (hereafter referred to as parallel cultures), and to inoculate a single flask containing an equivalent amount of nutrient broth as the sum of all the test tubes (hereafter referred to as the replicate culture). After 18 hours of incubation at 37°C with shaking, 100 microliter samples of each of the parallel cultures are plated on plates of nutrient agar that have been inoculated with no less than 10⁵ T-4 phages per plate. An equal number of replicate culture plates are prepared in the same way. These plates are then incubated at 37°C for 18 to 24 hours, after which time T-4-resistant E. coli colonies are counted. The variance in resistant colony counts from the parallel plates is compared to the variance in resistant colony counts from the replicate plates. A few representative parallel cultures and the replicate culture are also used to prepare 10⁻⁷ dilutions that are plated on nutrient agar free of phages to confirm that the concentration of viable bacteria is similar across all cultures.

Prior to the experiment, students were asked to predict in writing whether resistance to phages would occur, and if it did occur, whether bacterial resistance to phages occurred randomly or because of the action of the phages on the bacteria. At no time did the class instructor express any opinion on this question, and when directly asked by students which prediction was the correct one, the instructor feigned ignorance.

The final step in this experiment is counting the T-4-resistant E. coli colonies from parallel and replicate E. coli cultures and comparing the variance in resistant colony counts from the parallel culture plates to that of the replicate plates. Before they were allowed to examine the plates, students participated in a class discussion about what different potential results would mean. Students were provided with the written predictions they had made earlier, to help facilitate discussion. Discussion continued until there was consensus among the students that greater variance in resistant colony counts within the parallel plates than within the replicate plates would indicate that resistance occurs randomly; but if resistance is caused by the phages, there would be the same variance in resistant colony counts regardless of whether bacteria on plates were from parallel or replicate cultures. The instructor did not express an opinion as to the correct interpretation of the data, although the discussion was led in such a way as to direct students to the idea of variability among the parallel and replicate cultures. When students had reached a perfect consensus about the interpretation of data, resistant colonies were counted and class data pooled. An F-test for variances was run on the pooled class data.

All students enrolled in the class were required to write up the results of this experiment as part of a formal lab report on bacterial genetics, regardless of whether or not they had chosen to participate in the study.

Results

Average post-test scores were significantly higher than pre-test scores, with respect to the questions about the role of randomness in evolution (t(81) = −7.041, p < 0.001) (see Fig. 1).

FIGURE 1.

Average percentage score on relevant questions on pre-test and post-test, for participants in all five sections. Error bars represent standard error of the mean (SEM = 3.8).

A repeated-measures ANOVA was run with time (pre- vs. post-test) as the independent variable and scores on relevant questions as the dependent variable. When comparing test performance by class section, a significant effect of pre- vs. post-test (F(1, 77) = 25.913, η² = 0.316, power = 1.00, p < 0.0001) was found, as was a marginally significant interaction between pre vs. post and the class section from which the data was gathered (F(4, 77) = 2.375, p = 0.059).

Both men and women performed significantly better on the post-test than on the pre-test in regard to relevant questions. No interaction between sex of study participants and improvement on post-test scores compared to pre-test scores was observed (Mixed-Design Repeated Measures ANOVA, F(1, 80) = 0.239, p = 0.626). There was a nearly significant effect of sex on overall performance (F(1, 80) = 3.901, p = 0.052), largely accounted for by women’s poorer performance on the pre-test (see Fig. 2).

FIGURE 2.

Comparison of men’s and women’s average percentage score on relevant questions on pre-test and post-test. While no significant difference was observed in improvement on the post-test compared to the pre-test for either sex, men overall slightly outperformed women. Error bars represent standard error of the mean.

Pre-existing beliefs about the origin of life on Earth did not influence participants’ improved understanding of the role of randomness in evolution, as judged by comparison of pre- and post-test scores. There was no significant interaction between which of the five statements about the origin of life on Earth was selected by participants and their individual improvement on the post-test compared to the pre-test (Mixed-Design Repeated Measures ANOVA, F(3, 74) = 0.481, p = 0.696), nor were participants’ pre- and post-scores significantly different among those with different pre-existing beliefs about the origin of life on Earth (F(3, 74) = 1.836, p = 0.148).

Discussion

The teaching intervention based on the Luria-Delbruck experiment was followed by a significant gain in student understanding of the role of randomness in evolution for students across demographic groups and course sections. An extensive literature search using Google Scholar uncovered no previous studies showing any teaching intervention to be effective in disabusing students of a Lamarckian interpretation of evolutionary change. Prior to the teaching intervention, students chose Lamarckian explanations for evolutionary change in more than two out of three cases, and Darwinian explanations less than one-third of the time. This is the pattern of student answers on the pre-test that would be predicted if students picked answers randomly. After the intervention, students chose Darwinian explanations for scenarios describing evolutionary change more than two times as often as they did previously. This significant gain in student understanding of the role of randomness in evolution shown in comparing pre-test and post-test scores was underscored by observed changes in students’ written explanations of evolutionary change on class assignments. Consistent with the requirements of the Institutional Research Review Board that the course instructor not know the identity of research participants, such assignments were not included as data for this study. However, it was observed that students overwhelmingly provided Lamarckian explanations for microbiological phenomena such as antibiotic resistance on short written assignments prior to the teaching intervention, but overwhelmingly gave Darwinian explanations following it.

Neither pre- nor post-test questions asked about the specific case of evolution shown in the Luria-Delbruck experiment (i.e., evolution of bacteria from phage susceptibility to resistance), illustrating that students were able to apply the understanding they gained to new situations. Although the time delay in pre- to post-assessment could also affect change in understanding of evolution (because of students’ maturation, history, etc.), because of the good effect size (η² = 0.316) and near perfect power (power = 1.00) of this analysis, we attribute the improvement on scores on the post-test to the teaching intervention.

This study utilized multiple-choice questionnaires to measure student understanding of the role of randomness in evolution. These questionnaires were written by the study authors, as no validated instrument for assessing student learning of this concept has been described in the literature. While the reliability of these questionnaires was acceptable, with Cronbach’s Alpha measuring above 0.70 for both the pre-test and the post-test, future studies could help develop better instruments for assessing student understanding of this concept.

The effectiveness of this teaching intervention may be because of its emphasis on active learning. Through conducting a simplified version of the Luria-Delbruck experiment, students hypothesized whether E. coli resistance to phages arose randomly or because of phage exposure, and discovered for themselves what was true. Students then debated their interpretation of possible experimental results until a consensus was formed, allowing for active engagement with these concepts on multiple levels.

Previous studies have shown active learning approaches to be useful in teaching evolutionary biology concepts. In particular, student-conducted experiments that allow students to witness evolution have been shown to be successful (10). Further, pedagogical approaches emphasizing discussion and debate have been shown to improve student understanding of evolution (1). Finally, some studies suggest that conceptual change of the kind that displaces deep misconceptions, such as the naïve Lamarckian view of evolutionary change addressed in this study, requires cognitive conflict. That is, displacement of a deep misconception in science may only result from students being actively confronted with the knowledge that their previous understanding of phenomena was wrong (4).

In this intervention, students not only witnessed evolution and debated its causes, but were forced to confront their incorrect written predictions that mutations in E. coli to viral resistance occurred because of exposure to the viruses. Students were disallowed from examining experimental results until after the entire class section had reached a consensus about what potential results would mean. This provided a further social incentive for students to follow those results to their Darwinian conclusion, rather than falling back to a post hoc Lamarckian interpretation consistent with their original misconceptions about the workings of evolution.

Student gain in understanding of the role of randomness in evolution was not influenced by students’ religious affiliation or degree of religiosity as measured by attendance at worship services. Chi square contingency analyses indicated no significant relationship between students’ reported religious affiliation and their post-score performance (χ²(24) = 17.194, p = 0.840). Additionally, regression analyses reveal students’ reported frequency of church attendance did not significantly predict their post-score percentage (β = 0.136, p = 0.196 [Model F(1, 90) = 1.69, p = 0.196)]. Student gain in understanding was also not influenced by students’ beliefs about how life came to be on Earth, despite a majority of participants (67%) expressing at least some disbelief in evolutionary change. These findings contrast with studies showing that underlying creationist beliefs inhibit students’ abilities to learn about evolution (9), as well as with data that indicate that degree of Christian religiosity correlates negatively with understanding of evolution among Americans (8). While Ingram and Nelson (3) have shown similar levels of achievement in learning evolutionary concepts between students who accept and reject evolution, both they and McKeachie et al. (9) found creationists’ learning to be more strongly motivated by grades than that of students who accepted evolution. It is unlikely that the strong performance of evolution-doubting students on the post-test in this study can be explained by such external motivators, however. Study participants were repeatedly informed that the instructor grading them would never know whether they had chosen to participate in this study. As well, both the pre- and post-test were worded in such a way as to encourage reporting of personal beliefs about the reasons for evolutionary change, not merely reiteration of a perceived “right answer” (see Appendix: Pre- and Post-Test Questionnaires).

Whether these gains in understanding the role of randomness in evolution will be lasting has yet to be seen. Smith (12) notes that, while several promising interventions to improve understanding of evolution have been published in the past decade, student reversions to earlier-held misconceptions are common. Students who consented to participate in this study also agreed to be contacted in years following the intervention. While a high drop-out rate is expected, it is hoped that enough participants from this study will respond to our future questionnaires that it can be determined whether the intervention described in this paper affects a lasting improvement in students’ understanding of evolutionary change, or whether the gains shown here are short-term.

SUPPLEMENTARY MATERIALS

Appendix: Pre- and Post-Test Questionnaires