The Play’s the Thing: Experimentally Examining the Social and Cognitive Effects of School Field Trips to Live Theater Performances

Abstract

Field trips to see theater performances are a long-standing educational practice; however, there is little systematic evidence demonstrating educational benefits. This article describes the results of five random assignment experiments spanning 2 years where school groups were assigned by lottery to attend a live theater performance or, for some groups, watch a movie version of the same story. We find significant educational benefits from seeing live theater, including higher levels of tolerance, social perspective taking, and stronger command of the plot and vocabulary of those plays. Students randomly assigned to watch a movie did not experience these benefits. Our findings also suggest that theater field trips may cultivate the desire among students to frequent the theater in the future.

Keywords

arts education educational policy educational reform experimental design instructional practices museum education student behavior/attitude

Student field trips to see live theater performances are a long-standing educational practice. Like many common school practices, however, there is little systematic evidence demonstrating educational benefits. Field trips to see plays continue mostly with the support of the wisdom of educators and a sensible deference to prior practice. With the rise of test-based accountability, however, many traditional school practices are under pressure, including school field trips to theaters and other cultural institutions (Gadsden, 2008; Rabkin & Hedberg, 2011). Cultural field trips face severe cutbacks if they cannot show improved performance on tested subjects or other important educational benefits.

This article describes the results of a series of five experiments conducted over 2 years in which school groups from a variety of grades were assigned by lottery to see live theater. We find significant educational benefits from taking students to a theater performance. In particular, students in our full sample randomly assigned to see live theater demonstrate higher levels of tolerance and social perspective taking (SPT) as well as stronger command of the plot and vocabulary of those plays. When results are broken out for each of the five plays in our study, point estimates for each of these outcomes remain positive but sometimes fall short of statistical significance due to smaller sample sizes. Our findings also suggest that theater field trips may cultivate the desire among students to frequent the theater in the future, although this result is only marginally significant.

In addition, for two of the five experiments, a second treatment condition was added in which some students were randomly assigned to see a movie comparable to the play seen by other groups of students. Leaving school to see a movie did not produce the same benefits as viewing live theater. The evidence suggests that there are educational benefits to the traditional practice of school field trips to see plays and that those benefits are unlikely to be replicated by showing students movies instead.

Previous Research

Attending arts-focused school field trips is a longstanding tradition in American public education; however, there is mounting evidence that students are receiving less exposure to the arts, both from in-school and out-of-school activities. Cultural institutions in large cities, such as Chicago and Cincinnati, report a marked drop in visits from school groups (Greene, Kisida, & Bowen, 2014). A survey administered by the American Association of School Administrators also reported that schools were increasingly cancelling previously planned field trips (Ellerson & McCord, 2009). In addition, a Government Accountability Office (GAO) report on access to arts education found 7% of public school teachers reported a decline in the amount of instructional time dedicated to arts education in schools—a modest decrease (GAO, 2009). Of more concern, the GAO found that schools with high numbers of minority students and greater academic challenges were significantly more likely to report a decline in arts instructional time, including arts-focused field trips (GAO, 2009). Further, principals reported struggles with decreased budgets and demands placed on instructional time as a result of accountability standards and voiced concerns that arts education might suffer cuts (GAO, 2009). The GAO recommended further study of these issues to ensure that all students receive a well-rounded education. Similarly, the National Endowment for the Arts (NEA) reports a decline in adult attendance and participation in the arts and that the greatest predictor of future arts consumption and participation is arts education (Rabkin & Hedberg, 2011).

While there has been little rigorous research that speaks directly to the effects of seeing live theater on students, there is a growing literature on related topics. For example, a recent, large-scale experiment found that a single school field trip to tour an art museum caused significant effects that could be observed nearly 2 months following the visit (Greene et al., 2014). Students randomly assigned to receive the art museum tour were significantly more likely to be interested in visiting cultural institutions in the future and actually did so at a higher rate than students randomly assigned to the control group that had not toured the museum (Kisida, Greene, & Bowen, 2014). Students who toured the art museum also scored significantly higher on a measure of their ability to think critically about works of art (Bowen, Greene, & Kisida, 2014). In addition, students who visited the art museum displayed higher levels of content knowledge, tolerance, and historical empathy as a result of their tour (Greene et al., 2014). All of these benefits were more likely to be realized by students from more disadvantaged backgrounds, suggesting that advantaged parents can more easily substitute with their private efforts if schools fail to take students to cultural institutions. It appears that schools may play an essential role in providing equal access to cultural institutions and any benefits they produce.

In addition, some long-term analyses find that exposure to cultural activities improves academic outcomes for students years afterwards. For example, Jægar and Møllegarrd (2017) studied a large sample of monozygotic twins in Denmark to see if their cultural activity was related to later educational outcomes. By comparing outcomes within pairs of identical twins, the researchers control automatically for a large set of unobserved environmental and genetic factors. The mothers of these sets of identical twins were asked about each child’s cultural activity at the age of 12, including “how often child went to any type of museum” and “how often child went to the theater or a musical performance” (Jægar & Møllegarrd, 2017, p. 142). The researchers then examined how well each child did according to teacher and independently proctored grades at the age of 15 as well as the rate at which they graduated from high school. There were strong correlations between the relative level of cultural activity in which each twin engaged, their independently proctored grades, and their likelihood of graduating from high school. Other long-term studies find correlations between student involvement in the arts and later academic outcomes (Lacoe, Painter, & Williams, 2016; Ruppert, 2006), but most of these studies have difficulty establishing that this is a causal relationship.

Goldstein and Winner (2012) conducted a set of experiments to examine how students are affected by drama activities and find significant social-emotional benefits measured shortly after the intervention. There is also an extensive literature that examines how drama-based instructional techniques affect students. In a meta-analysis of that research, Lee, Patall, Cawthon, and Steingut (2015) found that drama-based pedagogy can have significant academic effects as well as social-emotional outcomes. While this is a comprehensive review of that research, the authors acknowledged that the 47 studies they examined are quasi-experimental, not experimental, and therefore some of the observed relationships may not be causal. In another meta-analysis of the literature examining the value of arts integration, the American Institutes for Research (AIR) finds that the average child could gain up to four percentage points in achievement from arts integration interventions, but almost none of the studies used to generate this estimate utilized the strongest causal research designs (Ludwig, Boyle, & Lindsay, 2017).

We have some evidence to suggest that students benefit from school visits to art museums, experience long-term academic gains from frequenting museums and the theater, and may learn from drama-based pedagogy and theater activities. But regarding the exact question addressed in this study—whether students benefit from school visits to see live theater—there is little direct evidence. The results of two of the five theater experiments contained in this article were described in an earlier publication (Greene, Hitt, Kraybill, & Bogulski, 2015), but to our knowledge, this is the first large-scale experiment to examine what students learn from seeing live theater.

Research Design

This study addresses the question of whether students benefit from school visits to see live theater. School groups were randomly chosen to receive free tickets to attend one of five live theater performances over a 2-year period. Participating schools were primarily from the Northwest corner of Arkansas. Most schools were in semirural areas with over half of students qualifying for free or reduced priced lunch. Teachers applied for a chance to bring their class to one of five different theater performances. Teachers and students were unaware of the hypotheses being tested at any time during the experiments. We then matched applicant classes based on their similarity in student populations. Typically, applicant classes were matched with others within their same school, but sometimes they were matched with classes in other schools that had similar students. Matched students within the same or neighboring schools were likely to share similar observed and unobserved characteristics. Within the matched set, we randomly assigned one or more classes to receive tickets and one to serve as a control group, further ensuring that treatment and control groups were similar. While treatment groups received a field trip, the control groups did not and continued with their normal school activities. In total, we conducted 47 lotteries, creating 94 treatment and control groups containing almost 1,500 students. Performances included A Christmas Carol, Hamlet, Around the World in 80 Days, and Peter and the Starcatcher, all performed by an award-winning professional company, and Twelfth Night, performed by university theater students.

For the final two plays, Peter and the Starcatcher and Twelfth Night, we were able to add a second treatment condition in which students would be randomly assigned to see a movie that was similar to what the theater treatment group saw. For the Twelfth Night experiment, applicant groups were randomly assigned to see the play, to see the 1996 film of the same Shakespeare story, or to serve in the control group, which saw neither the play nor the movie. While the mode of delivery was different, and indeed that is the point, the play and the movie both used virtually the same script for the performances. For the Peter and the Starcatcher experiment, applicant groups were randomly assigned to see the play, to see the 1991 film Hook, or to serve in the control group that saw neither. In this case, the theater performance and the movie used different scripts but had similar content, characters, and themes. This addition of a movie treatment allowed us to test whether any effects of seeing a play were derived from the subject matter of the play or from the experience of seeing live theater. The fact that both play and movie treatment groups left school on the same bus, at the same time, and only differed in whether they walked into a ballroom to see the movie or into the theater to see the play allows us to examine whether any observed outcomes could be caused simply by leaving school on a field trip. While the play and movie may not have been identical in both cases, the policy-relevant question we address is whether students are differentially affected by the two modes of delivery. If watching a movie is as effective as taking a bus to see a play, then students might simply watch movies. But if the play is more effective than a virtually identical or similar movie, then the allocation of resources in both time and money are justified for field trips to see live theater.

It is important to note that the experimental interventions did not include anything beyond the opportunity to see live theater. The treatment did not provide any additional training, materials for teachers, or supplemental activities for students. Any supplementary activities, such as reading the play or watching the movie in class, were assigned at the teachers’ discretion and could have occurred among treatment or control groups. If teachers in the treatment group were more likely to provide additional instruction or discussion, then that is part of the treatment effect we are estimating. Our lack of information on the extent to which teachers may have prepared their students to see the plays, however, prevents us from knowing with confidence the extent to which outcomes are attributable to the play alone or to any preparation.

To collect outcome measures, we administered surveys to treatment and control students in their classrooms. On average, surveys were administered 54 days after the treatment group had seen the play. There were not differential participation rates among the treatment and control groups. We collected surveys from 77.6% of the students assigned to see a play, 76.0% for those assigned to see a movie, and 76.5% among control group students.

While the basic design—offering free tickets, matching similar applicant groups, and then conducting a lottery within matched sets of applicants—remained the same across all five plays, some important details did change over time. For example, after the first two plays, we changed the survey to add a scale designed to measure students’ SPT (Gehlbach, 2004; Gehlbach et al., 2008) while dropping the Reading the Mind in the Eyes Test (Baron-Cohen, Wheelwright, Hill, & Raste, 2001). We believed that SPT would be a better way of capturing potential social-emotional effects of seeing live theater.

In addition to replacing one measure on the survey instrument after the first two plays, we also administered pretreatment surveys to students after the first three plays. In prior administration, we lacked the resources to collect measures both before and after the intervention, so we relied on the lottery to give us equivalent treatment and control groups since we could not control for any pretreatment differences. For the final two plays, however, we were able to administer surveys to all students both before and after the treatment occurred. This allowed us to check whether our treatment and control groups were similar on pretreatment measures of the outcomes. Controlling for pretreatment measures of the outcomes also improves the precision of our estimates of treatment effects, although it reduces the size of the sample, which reduces the precision of estimates. On balance, we prefer the models using the larger sample and without controlling for baseline measures because the larger sample tends to give us smaller standard errors with generally the same point estimates as the smaller samples that control for baseline measures.

Other outcomes measured in the survey remained unchanged across the five plays. Given prior research that had found social-emotional benefits from arts-focused field trips, we used a scale to measure Tolerance (Greene et al., 2014). We suspected that by exposing students to a broader world through theater, they would increase their ability to understand other people’s points of view (SPT) as well as gain greater acceptance of other people (Tolerance) (Gehlbach, 2004; Greene et al., 2014). We also suspected, based on prior work by Greene et al. (2014), that seeing live theater would be an effective mechanism for conveying the plot and vocabulary of these plays, so we included measures of Content Knowledge. In addition, since past research suggested that visiting cultural institutions increased the desire to frequent those institutions in the future (Kisida et al., 2014), we included measures of Theater Consumption in the survey. Lastly, we included measures of the desire to Participate in Theater because we thought it was possible that seeing live theater might inspire students to become more involved in theater by auditioning for plays and taking drama class.

Implementation of Research Design

Examining the background characteristics of our treatment and control groups confirms that randomization was successful in helping ensure that we compared generally similar groups. There were no differences in background characteristics that were significant at p < .05 (see Table 1). There were two instances in which differences are significant at p < .10, but we might expect this by chance given that we were comparing three groups on 10 different variables. On our pretreatment measures of outcomes, we observed no statistically significant differences between the control and treatment groups.

Table 1

Student Characteristics in Treatment and Control Groups

	Control Group Mean	Play Treatment Mean	Movie Treatment Mean	N
Full sample
Female	0.609	0.550*	0.596	1,467
Age	14.751	14.717	15.175	1,463
Grade	8.628	8.605	9.059	1,485
White	0.672	0.660	0.681	1,467
Seen play in last year	0.263	0.241	0.304*	1,467
	Twelfth Night and Peter and the Starcatcher sample with pretest
Tolerance	2.986	2.936	2.959	485
Social Perspective Taking	2.557	2.451	2.471	485
Content Knowledge	0.368	0.394	0.354	485
Theater Consumption	2.229	2.252	2.323	485
Theater Participation	1.961	1.676	1.870	485

Note. The outcomes of interest (Tolerance, Social Perspective Taking, Comprehension, Vocabulary, Theater Consumption, and Theater Participation) were only pretested with students who went to see Twelfth Night and Peter and the Starcatcher.

p < .10. **p < .05. ***p < .01 (two-tailed, relative to control group).

Students in our sample were just shy of their 15th birthday and in the middle of 9th grade on average. However, students varied in age, with some students as young as 4th grade and some as old as 12th grade. Approximately two thirds of our sample identified as white, which reflects the broader community in which the experiments took place. About one quarter of students had seen a play in the previous year, however this might have included school plays, church plays, and holiday shows. This probably reflects a relatively low level of previous exposure to theater and cultural activity.

Two problems occurred during implementation of the research design that caused some applicant groups not to see the play despite being assigned to the play treatment. Severe winter weather forced the cancellation of a performance of A Christmas Carol. Additionally, during the Around the World in 80 Days experiment, the theater made an error that caused the actors not to be available to perform when students arrived. To be very conservative, we report results for Intention to Treat (ITT), in which we count all students randomly assigned to the treatment group as if they received the treatment even if they failed to do so because of weather or a scheduling error. To provide a more realistic estimate of the treatment effect, we also report the Impact on Treated (IOT) derived from a two-stage model in which the first stage uses assignment to the treatment group as a predictor of whether students actually received the treatment.

Outcome Measures

The Tolerance scale consisted of seven items. Students were given four options, from strongly disagree to strongly agree, to respond to a series of statements designed to capture their general acceptance of other people and different opinions. The scale was adapted from Greene et al. (2014) and included statements such as “People who disagree with my point of view bother me” or “I think people can have different opinions about the same thing.” The Cronbach’s alpha for this scale was .71, suggesting that there is an acceptable amount of internal consistency within this scale.

The SPT scale also consisted of seven items and was adapted from Gehlbach, Brinkworth, and Wang (2012). Students had five response options, from almost never to almost all of the time, to questions like “How often do you try to figure out what motivates others to behave as they do?” and “Overall, how often do you try to understand the point of view of other people?” The Cronbach’s alpha in our study for this scale was .85, indicating strong internal consistency in students’ responses.

Our measure of Content Knowledge consisted of six questions about the plot of each play and five questions about vocabulary drawn from the play. For example, for students in the Hamlet experiment, we asked, “What happens to Ophelia?” or asked them about the definition of “countenance” and provided four response options to each question. It is reasonable to expect that students who attended the play or movie would be more likely to know the plot and vocabulary than students who saw neither. However, it is not guaranteed that simply taking students to the theater means that they will acquire knowledge about what they saw. We are able to determine if students actually learn and retain content knowledge from attending a play or watching a movie by comparing treatment and control students. In addition, it is important to measure the content knowledge of control group students because students may acquire information about the plot and vocabulary of these shows from other sources without seeing a play or movie. Given that the questions were different for each play, calculating a combined Cronbach’s alpha is not possible, but we are confident that we captured meaningful variation in knowledge about each play’s plot and vocabulary.

The Theater Consumption scale consisted of 10 items adapted from Kisida et al. (2014). Students had four response options to questions like “How interested are you in seeing live performances in a theater?” or statements like “I plan to see live theater performances when I am an adult.” The Cronbach’s alpha for this scale was .92.

The Theater Participation scale was also adapted from Greene et al. (2014) and consisted of four items. Students were asked questions like “How interested are you in taking a drama class?” and “If your school were having auditions for a new play, how interested would you be in trying to get a role in that play?” The Cronbach’s alpha for this scale was .93.

In Table 1 all of these outcomes are expressed as the mean of a 0–3 or 0–4 scale, except for Content Knowledge, which is expressed as the percentage of questions answered correctly. For the purposes of the outcome analyses, all scales were converted into z scores with a mean of 0 and a standard deviation of 1. The reported results, therefore, are the effect sizes expressed as a percentage of a standard deviation.

Analyses

Because the randomized controlled trial research design used here has the important feature of generating comparable treatment and control groups (which we confirmed with the analyses presented in Table 1), we can use a straightforward set of analytic techniques, designed for use in social experiments, to estimate the impact of a school field trip to see live theater on student outcomes. In its simplest form, this technique can estimate mean differences using the following equation for outcome $γ$ of student i in matched set m:

γ_{i m} = α + β_{1} P l a y_{i m} + β_{2} M o v i e_{i m} + β_{3} M a t c h_{m} + ε_{i m},

where the binary variable $P l a y_{i m}$ is equal to 1 if the student is in the treatment group that was randomly assigned to receive free tickets for a field trip to see one of the five plays and is equal to 0 otherwise. The binary variable $M o v i e_{i m}$ is equal to 1 if the student is in the treatment group that was randomly assigned to take a field trip to see a movie and is equal to 0 otherwise. Because the groups were created using a randomization procedure within matched applicant sets, $M a t c h_{m}$ is also included in the model as a vector of binary variables that have the statistical effect of estimating within, as opposed to across, matched groupings. Finally, $ε_{i m}$ is a stochastic error term clustered at the classroom level to take into account the spatial correlation from students nested within classrooms.

Proper randomization generates experimental groups that are comparable but not necessarily identical. The basic regression model can, therefore, be improved by adding controls for observable characteristics to increase the reliability of the estimated impact by accounting for minor differences and improving the precision of the overall statistical model. This yields the following equation:

γ_{i m} = α + β_{1} P l a y_{i m} + β_{2} M o v i e_{i m} + β_{3} M a t c h_{m} + β_{4} G e n d e r_{i m} + β_{5} A g e_{i m} + β_{6} W h i t e_{i m} + ε_{i m},

where $G e n d e r_{i m}$ is a binary variable equal to 1 if the student is a female and 0 otherwise, $A g e_{i m}$ is a continuous variable indicating precise age of student i at the time posttreatment surveys were administered, and $W h i t e_{i m}$ is a binary variable equal to 0 if the student does not identify as being white and is 1 otherwise. In this model, $β_{1}$ and $β_{2}$ are the parameters of interest and represent the effect of a class field trip for students in the play and movie treatment groups. Equation 2 is our preferred model and was used to produce the ITT results presented in Table 2. The ITT estimate describes what the effect was for all students assigned to the treatment, regardless of whether they actually attended the field trip. It is the most conservative estimate of the treatment effect because it could be biased downward due to some students not seeing the play despite being assigned to treatment.

Table 2

Impacts of Play and Movie Treatment on Five Primary Student Outcomes

	Tolerance		Social Perspective Taking		Content Knowledge		Theater Consumption		Theater Participation
	Intention to Treat	Impact on Treated	Intention to Treat	Impact on Treated	Intention to Treat	Impact on Treated	Intention to Treat	Impact on Treated	Intention to Treat	Impact on Treated
Play Treatment	0.142**	0.190**	0.169**	0.222**	0.101***	0.135***	0.127*	0.170*	−0.109	−0.147
Play Treatment	(0.065)	(0.087)	(0.080)	(0.110)	(0.016)	(0.018)	(0.069)	(0.092)	(0.082)	(0.105)
Movie Treatment	−0.038	−0.009	−0.045	−0.014	−0.015	0.006	0.084	0.110	0.008	−0.015
Movie Treatment	(0.095)	(0.098)	(0.087)	(0.097)	(0.027)	(0.026)	(0.104)	(0.108)	(0.120)	(0.125)
Female	0.290***	0.289***	0.407***	0.406***	−0.016*	−0.016*	0.359***	0.358***	0.414***	0.415***
Female	(0.058)	(0.057)	(0.093)	(0.090)	(0.009)	(0.009)	(0.053)	(0.052)	(0.058)	(0.056)
White	0.231***	0.235***	0.090	0.102	0.060***	0.063***	0.035	0.038	0.013	0.009
White	(0.073)	(0.071)	(0.071)	(0.071)	(0.014)	(0.013)	(0.062)	(0.061)	(0.061)	(0.059)
N	1,441	1,441	791	791	1,442	1,442	1,442	1,442	1,442	1,442

Note. Standard errors—clustered within classes—are presented in parentheses. Coefficients for age and for the matched set parameters are not presented. Outcome measures for baseline were taken for students who saw Twelfth Night and Peter and the Starcatcher only.

p < .10. **p < .05. ***p < .01 (two-tailed).

Due to the noncompliance to treatment assignment during the first and third play experiments caused by weather and a scheduling error on the part of the theater, we are also interested in generating an IOT estimate. The IOT estimate describes what the effect would have been had all of the applicant groups actually seen the plays to which they were randomly assigned. The model used to generate that estimate is a two-stage least squares model in which the second stage is identical to Equation 2 except that $P l a y_{i m}$ is derived from the first stage in which lottery assignment is used to predict treatment compliance.

Because we do not have strong theoretical expectations that different plays should produce different effects and because the sample size for each individual play is relatively small, we present in Tables 2 and 3 the results of all five plays combined. The outcomes for SPT in Table 2 only contain the results for students who saw the last three plays since SPT was not added to the survey instrument until that time. All of the other outcomes in Table 2 represent the results of students across all five plays.

Table 3

Impacts of Play and Movie Treatment on Five Primary Student Outcomes Controlling for Baseline

	Intention to Treat
	Tolerance	Social Perspective Taking	Content Knowledge	Theater Consumption	Theater Participation
Play treatment	0.182**	0.056	0.154***	0.130	−0.040
Play treatment	(0.071)	(0.076)	(0.025)	(0.081)	(0.064)
Movie treatment	0.074	−0.064	0.051*	0.082	−0.005
Movie treatment	(0.074)	(0.064)	(0.028)	(0.066)	(0.057)
Pretest	1.046***	0.681***	0.424***	0.977***	0.581***
Pretest	(0.066)	(0.051)	(0.057)	(0.045)	(0.025)
Female	0.071	0.312***	−0.013	0.042	0.084
Female	(0.081)	(0.094)	(0.015)	(0.076)	(0.073)
White	0.052	−0.005	0.040*	0.039	−0.027
White	(0.099)	(0.062)	(0.021)	(0.065)	(0.073)
N	475	475	475	475	475

p < .05. ***p < .01 (two-tailed).

For the last two plays, we surveyed all students prior to the treatment and again after the intervention, which allows us to control for the pretreatment measure of the outcome. For example, we can control for students’ score on the Tolerance scale prior to the intervention when estimating the effect of the treatment on their Tolerance score collected after the intervention. The model we used to generate these results (as presented in Table 3) can be expressed as:

γ_{i m} = α + β_{1} P l a y_{i m} + β_{2} M o v i e_{i m} + β_{3} M a t c h_{m} + β_{4} P r e T e s t_{i m} + β_{5} G e n d e r_{i m} + β_{6} A g e_{i m} + β_{7} W h i t e_{i m} + ε_{i m} .

This is identical to Equation 2 except that it adds $P r e T e s t_{i m}$ , which is the pretreatment measure of the outcome $γ_{i m}$ , including Tolerance, SPT, Content Knowledge, and Theater Consumption and Participation.

Results

As shown in Table 2, providing students with the opportunity to leave school on a field trip to see a live theater performance produced a number of significant effects. Students given the opportunity by lottery to see a play scored .142 of a standard deviation higher on the Tolerance scale than if they were in the control group. If we adjusted for the noncompliance produced by bad weather and a scheduling error, the estimated effect of actually seeing a play on Tolerance increases to .190 of a standard deviation. Being assigned to see a movie instead of a play appears to have no effect on Tolerance. When results are broken out for each of the five plays in our study, point estimates for Tolerance remain positive but sometimes fall short of statistical significance due to smaller sample sizes.

SPT increases by .169 of a standard deviation for students randomly assigned to go on a school field trip to see a play. Again, the effect is larger for those who actually saw the play. Being assigned to see a movie, however, has no effect on SPT. Student’s Content Knowledge of the plot and vocabulary in these stories is also increased when students see the play. Watching a movie did not convey this Content Knowledge as effectively as seeing the live performance. When results are broken out for each of the three plays in our study for which we measured SPT, however, the point estimates remain positive, but none of them are statistically significant due to smaller sample sizes.

Taking a field trip to see a play may strengthen student interest in consuming theater in the future but that effect was only significant at p < .10. Being assigned to the movie treatment produced a similar estimated effect, but it was not statistically significant even at p < .10. Neither the play nor movie treatments had any detectable effect on student interest in participating in future theater activities.

Results for each play are presented in Table 4; the point estimates are roughly consistent across all plays. That is, if the overall effect is positive and statistically significant, the individual play estimates are almost all positive and many are also statistically significant. The only interesting result from the analyses of each individual play has to do with the effect of the movie treatment on Content Knowledge. Seeing a movie of Twelfth Night increased understanding of the plot and vocabulary of that story relative to the control group by .045 of a standard deviation but that effect is only statistically significant at p < .10 and is almost one third as big as the play treatment effect. Seeing the movie Hook had no effect on the Content Knowledge for Peter and the Starcatcher but that is not too surprising given that the stories are not as well aligned as are the movie and play of Twelfth Night.

Table 4

Intention to Treat Impact on Each Outcome of Interest for Each Play

	Combined Plays	A Christmas Carol	Hamlet	Around the World in 80 Days	Twelfth Night	Peter and the Starcatcher
Impact on Tolerance
Play treatment	0.142**	0.231	0.261**	0.288**	0.337***	0.076
Play treatment	(0.065)	(0.145)	(0.118)	(0.100)	(0.110)	(0.176)
Movie treatment	−0.038				0.144	−0.126
Movie treatment	(0.095)				(0.124)	(0.193)
N	1,441	340	310	260	269	262
Impact on Social Perspective Taking
Play treatment	0.169**			0.239	0.227	0.026
Play treatment	(0.080)			(0.143)	(0.133)	(0.114)
Movie treatment	−0.045				0.046	−0.241*
Movie treatment	(0.087)				(0.104)	(0.119)
N	791			260	269	262
Impact on Knowledge
Play treatment	0.101***	0.062	0.121***	0.067***	0.122***	0.237***
Play treatment	(0.016)	(0.051)	(0.024)	(0.016)	(0.020)	(0.046)
Movie treatment	−0.015				0.045*	−0.005
Movie treatment	(0.027)				(0.024)	(0.048)
N	1,442	341	310	260	269	262
Impact on Interest in Theater Consumption
Play treatment	0.127*	0.035	0.260*	0.329**	0.217	0.319
Play treatment	(0.069)	(0.097)	(0.146)	(0.129)	(0.150)	(0.255)
Movie treatment	0.084				0.250	0.106
Movie treatment	(0.104)				(0.155)	(0.186)
N	1,442	341	310	260	269	262
Impact on Interest in Theater Participation
Play treatment	−0.109	−0.108	−0.051	−0.090	0.148	−0.373***
Play treatment	(0.082)	(0.130)	(0.182)	(0.152)	(0.238)	(0.117)
Movie treatment	0.008				0.236	−0.033
Movie treatment	(0.120)				(0.242)	(0.121)
N	1,442	341	310	260	269	262

Note. Standard errors—clustered within classes—are presented in parentheses. Coefficients for age, grade, gender, and class are not presented. The movie treatment only occurred for Twelfth Night and Peter and the Starcatcher.

p < .10. **p < .05. ***p < .01 (two-tailed).

When we control for pretreatment measures of the outcome, as presented in Table 3, the results generally remain the same. The Tolerance and Content Knowledge effects from the play treatment remain approximately the same size and continue to be statistically significant. The result for SPT shrinks in magnitude and falls short of being statistically significant once we control for a pretreatment measure of SPT. Similarly, the Theater Consumption effect, which was marginal in Table 2, also falls short of being statistically significant. Generally null effects of the movie treatment persist even after controlling for pretest measures of the outcomes with the possible exception of Content Knowledge. When we control for prior Content Knowledge, seeing a movie may increase understanding of the plot and vocabulary of the stories by .051 of a standard deviation, but that effect is only significant at p < .10. It is important to note that in controlling for pretreatment measures, we decrease our sample size. While the addition of the baseline controls improves the precision of our estimates, the loss of sample size decreases the precision even more while point estimates remain largely similar. For these reasons, we prefer the full sample models without controls for baseline measures.

Descriptively, it is interesting to note that female students tend to score higher on the Tolerance, SPT, Theater Consumption, and Theater Participation measures, but these differences mostly dissipate when controlling for pretreatment measures of those outcomes. Similarly, white students score higher on Tolerance and Content Knowledge outcomes, but that entirely disappears when controlling for pretreatment measures of those outcomes. We find no evidence that the play or movie treatments had differential effects on students by gender or race/ethnicity.

Discussion

The experimental evidence presented here clearly shows that students can benefit from school field trips to see live theater. The effects are most robust with respect to measures of Tolerance and Content Knowledge. Whether we control for pretreatment measures of outcomes or not, students experience an increase in Tolerance as well as greater understanding of the plot and vocabulary of stories if they see live theater.

There may also be a benefit from seeing live theater for student ability to engage in SPT, but that effect falls short of statistical significance when controlling for a pretreatment measure of that outcome. We gain precision by controlling for pretreatment measures, but we also lose sample size because pretest measures were only collected for the last two plays. On balance, we prefer the full model without controls for baseline measures, but we provide both sets of results for readers to judge.

There is some indication that students randomly assigned to see live theater become more interested in frequenting the theater in the future, but that effect is only marginally significant and disappears when controlling for the pretreatment measure of that outcome. Again, this loss of statistical significance is likely due to the reduction in sample size and loss of precision. Student interest in participating in theater does not seem to be affected at all by this treatment.

It is also important to emphasize that the movie treatment does not seem to have a robust effect on any of these outcomes. Going on a field trip to see live theater produces benefits that cannot be produced by watching a movie instead. And the fact that students who received the movie treatment also left school for a field trip suggests that the effects we have observed are caused by the experience of watching live theater and not simply caused by leaving school.

While this experiment demonstrates that live theater field trips cause an increase in Tolerance and perhaps in the related concept of SPT, it cannot tell us why these effects were produced. Our best explanation is that theater is a window for students to a broader world. Exposure to that broader world may increase their understanding and acceptance of that broader world, which is why we see increases in Tolerance and SPT. Plays may be more effective than movies in helping students understand and accept that broader world because we react differently to human beings acting out a story in front of us than to representations of human beings on a screen. The in-person experience may create greater emotional connections. This pattern of results is consistent with earlier research by Greene et al. (2014) that finds similar effects when students take field trips to visit an art museum.

It is educationally significant and a bit surprising that watching a movie is not a particularly effective way of conveying content knowledge while watching a play is. Watching movies is an extremely common school practice, but it may produce little learning. Going to see a play, on the other hand, is less common but appears much more effective. This is especially surprising given that many films may be higher quality productions than plays accessible to school groups. But as we saw in our experiment, even seeing a university play of Twelfth Night taught students significantly more plot and vocabulary than seeing the film with award-winning actors like Helena Bonham Carter and Ben Kinglsey. The in-person experience, again, appears to trump the skill of the actors.

Of course, we were only able to observe effects 7 to 8 weeks after students saw the plays, so we do not know if these benefits endure over longer periods of time. And while we saw consistent results across multiple plays, produced by different theaters, and involving different school groups, all of our experiments occurred in one particular place of the country. So we cannot know with confidence that these benefits would be produced for other students.

Nonetheless, this is the first major piece of experimental evidence on how viewing theater affects students and it shows significant benefits. Before we eliminate or further cut these cultural experiences from schools, we should attempt to replicate this experiment in other locations and with longer term measures of outcomes.

Footnotes

Authors

JAY P. GREENE, PhD, is a Distinguished Professor of Education Policy at the University of Arkansas, 200 Graduate Education Building, Fayetteville, AR 72701; jpg@uark.edu . His research focuses on culturally enriching field trips, school choice, and the effect of schools on noncognitive and civic values.

HEIDI H. ERICKSON is a Doctoral Fellow at the University of Arkansas, 200 Graduate Education Building, Fayetteville, AR 72701; hh018@email.uark.edu . Her research focuses on education policy and school finance with issues ranging from school choice to arts and values education.

ANGELA R. WATSON is a Distinguished Doctoral Fellow of Education Policy at the University of Arkansas, 200 Graduate Education Building, Fayetteville, AR 72701; arw007@uark.edu . Her research interests include arts-related field trips and informal learning, STEM gender gap, and school choice with a focus on homeschooling.

MOLLY I. BECK, MPA, is a Distinguished Doctoral Fellow at the University of Arkansas, 201 Graduate Education Building, Fayetteville, AR 72701; mibeck@uark.edu . Her research interests include the educational outcomes of English Language Learners, academic and nonacademic outcomes from experiential learning, and teacher recruitment, quality, and retention.

References

Baron-Cohen

Wheelwright

Hill

Raste

(2001). The “Reading the Mind in the Eyes” test revised version: A study with normal adults, and adults with Asperger syndrome or high-functioning autism. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 42(2), 241–251.

Bowen

D. H.

Greene

J. P.

Kisida

(2014). Learning to think critically: A visual art experiment. Educational Researcher, 43(1), 37–44.

Ellerson

McCord

(2009). One year later: How the economic downturn continues to impact school districts. American Association of School Administrators. Available online at http://www.aasa.org/uploadedFiles/Resources/files/OneYearLater%20FINAL.pdf

Gadsden

V. L.

(2008). The arts and education: Knowledge generation, pedagogy, and the discourse of learning. Review of Research in Education, 32(1), 29–61.

Gehlbach

(2004). A new perspective on perspective taking: A multidimensional approach to conceptualizing an aptitude. Educational Psychology Review, 16(3), 207–235.

Gehlbach

Brinkworth

M. E.

Wang

(2012). The social perspective taking process: What motivates individuals to take another’s perspective? Teachers College Record, 114, 1–29.

Gehlbach

Brown

S. W.

Loannou

Boyer

M. A.

Hudson

Niv-Solomon

Maneggia

Janik

(2008). Increasing interest in social studies: Social perspective taking and self-efficacy in stimulating simulations. Contemporary Educational Psychology, 33(4), 894–914.

Goldstein

T. R.

Winner

(2012). Enhancing empathy and theory of mind. Journal of Cognition and Development, 13(1), 19–37.

Government Accountability Office. (2009). Access to arts education: Inclusion of additional questions in education’s planned research would help explain why instruction time has decreased for some students. Washington, DC: Author. Available online at http://www.gao.gov/assets/290/286601.pdf

10.

Greene

J. P.

Kisida

Bowen

(2014). The educational value of field trips. Education Next, 14(1), 78–86.

11.

Greene

Hitt

Kraybill

Bogulski

(2015). Learning from live theater: Students realize gains in knowledge, tolerance and more. Education Next, 15(1), 54–61.

12.

Jægar

M. M.

Møllegarrd

(2017). Cultural capital, teacher bias, and educational success: New evidence form monozygotic twins. Social Science Research, 64, 130–144.

13.

Kisida

Greene

J. P.

Bowen

(2014). Creating cultural consumers: The dynamics of cultural capital acquisition. Sociology of Education, 87(4), 281–295.

14.

Lacoe

Painter

G. D.

Williams

(2016). Museums as classrooms: The academic and behavioral impacts of “school in the park.” Working Paper, Sol Price School of Public Policy, University of Southern California.

15.

Lee

B. K.

Patall

E. A.

Cawthon

S. W.

Steingut

R. R.

(2015). The effect of drama-based pedagogy on preK-16 outcomes: A meta-analysis of research from 1985-2012. Review of Educational Research, 85(1), 3–49.

16.

Ludwig

Boyle

Lindsay

(2017). Review of evidence: Arts integration research through the lens of Every Students Succeeds Act. Washington, DC: American Institutes for Research. Retrieved from http://www.air.org/resource/review-evidence-arts-integration-research-through-lens-every-student-succeeds-act

17.

Rabkin

Hedberg

E.C

. (2011). Arts education in America: What the declines mean for arts participation. National Endowment for the Arts, Washington, D.C. Retreived from: https://www.arts.gov/sites/default/files/2008-SPPA-ArtsLearning.pdf

18.

Ruppert

S. S.

(2006). Critical evidence: How the ARTS benefit student achievement. National Assembly of State Arts Agencies. Retrieved from https://nasaa-arts.org/wp-content/uploads/2017/05/critical-evidence.pdf