A Special Kind of Ambition: The Role of Personality in the Retention of Academically Elite Teachers
by Brady K. Jones - 2018
Background: Creating greater stability in the teacher labor force and improving teacher quality is an important education policy priority in the United States. While there is a robust literature on the external, environmental reasons teachers stay in or leave the occupation, little is known about the role internal, person-level factors play in teacher retention, especially among academically elite teachers.
Focus of Study: This study explores the role of personality, holistically defined, in teacher commitment.
Participants: The sample for this study consists of 107 graduates of a single teacher preparation program. They are classified as “academically elite,” as this preparation program is very selective and demands high GRE scores.
Research Design: Discriminant function and regression analyses are used to test which of a rich set of personality measures, both traditional self-report measures and coded narrative accounts of life and career high points, predict long-term commitment to teaching in this sample.
Results: Discriminant function analysis exploring differences between very long-term committers (15+ years) and short-term committers (7- years) suggests that long-term committers are distinguished by a “special kind of ambition”: they set goals that are both more difficult and more prosocial than their counterparts with a shorter commitment to the occupation, and in personal narratives they more often show “enlightened self-interest,” a combination of self-interest/self-promotion with concern for and connection to others. In addition, regression analyses show that these personality variables significantly predict retention in the sample as a whole, even when controlling for school advantage.
Conclusions: These results provide evidence that personality does play an important role in teachers’ occupational commitment, call into question pervasive stereotypes in the United States of teachers as unambitious, and suggest ways academically elite teachers might be able to shift the ways they think about their work in order to sustain themselves in the occupation.
When a teacher leaves a school, especially when she exits the occupation entirely, it exacts a toll on a number of stakeholders. To begin, hiring and training new staff members places a substantial financial burden on districts (Barnes, Crowe, & Schaeffer, 2007; Benner, 2000; Milanowski & Odden, 2007) and can destabilize school communities (Guin, 2004; Ronfeldt, Loeb, & Wyckoff, 2012). Teacher attrition also tends to be much higher in schools that serve disadvantaged students, leaving the most vulnerable students to be educated by a less experienced and less stable faculty (Barnes et al., 2007; Borman & Dowling, 2008; Hanushek, Kain, & Rivkin, 1999; Ingersoll, 2003; Ingersoll & May, 2012). Finally, attrition affects teachers themselves. They often experience frustration and disappointment when the job does not live up to their expectations (Costigan, 2005; Hamman, Gosselin, Romando, & Bunuan, 2010), and job burnout and attrition have been linked to high levels of guilt, stress, and even illness (Hakanen, Bakker, & Schaufeli, 2006; Hargreaves & Tucker, 1991; Kyriacou, 2001).
The consequences are even worse when especially talented teachers leave. Teacher quality is the most important school-based predictor of student achievement (Rivkin, Hanushek, & Kain, 2005; Rockoff, 2004), and excellent teachers seem to have an impact even on nonacademic outcomes many years down the road (Chetty, Friedman, & Rockoff, 2014). Although predicting who will become an excellent teacher is notoriously difficult (Goldhaber, 2002; Rivkin et al., 2005), the best evidence available suggests that those with high cognitive ability, especially verbal ability, are most likely to help their students achieve (Goldhaber, 2002; Jacob, 2007). It has prompted a fair amount of concern, then, that for several decades scholars have noted a decline in scores on standardized tests like the SAT for teachers (Bacalod, 2007; Goldhaber, 2002; Guarino, Santibañez, & Daley, 2006; Murnane, Singer, Willett, Kemple, & Olsen, 1991) and that the highest scorers tend to leave the classroom the most quickly (e.g., Podgursky, Monroe, & Watson, 2004).
There is still much to learn about how we might reverse this trend and retain especially academically talented teachers. We know something about the external factors that aid in teacher retention in general, especially the organizational characteristics (e.g., class size, administrator quality) that influence a teachers satisfaction and economic tools (e.g., salary structures) that might be used to make the job more enticing (Macdonald, 1999). For example, we know that teachers tend to leave schools that are low-achieving, have a high proportion of non-white students, pay lower salaries, have larger class sizes, lack resources (Loeb, Darling-Hammond, & Luczak, 2005), have a negative work context (Johnson, Kraft, & Papay, 2012; Simon & Johnson, 2015) and disciplinary issues (e.g., Friedman, 1995; Tsouloupas, Carson, Matthews, Grawitch, & Barber, 2010), and are subject to strong accountability pressure (Clotfelter, Ladd, Vigdor, & Diaz, 2004; Day & Gu, 2010). But little of this research is specific to academically elite teachers, who may or may not show the same patterns as the general teaching population. For example, individuals who have done very well academically likely have a range of other career options available to them (Bacalod, 2007) or high movement capital (Kelly & Northrop, 2015). They may, therefore, be more likely to stay in teaching because they find it enjoyablea good fit with their personalities and preferencesnot because they have few other choices.
Furthermore, organizational characteristics and economic structures cannot explain all teacher attrition, and with few exceptions, researchers have not looked beyond these external factors. We know very little, for example, about why some teachers leave excellent schools while others stay at difficult ones, or why two different teachers in the exact same school context make different retention decisions. As Hamman et al. (2010) write, Repeatedly, researchers have pointed to contextual influences on staggering rates of teacher attrition yet among new teachers who remain in the field, some factor must act to insulate them from the difficulties created by nonsupportive administrators, cultural differences between themselves and their students, and limited resources (p. 1358).
CONCEPTUAL FRAMEWORK: MCADAMS MODEL OF PERSONALITY
This study seeks to explore what these insulating factors might be for academically elite teachers. The tools of personality psychology, specifically McAdams holistic model of personality (see McAdams & Olson, 2010; McAdams & Pals, 2006) is a particularly useful way to do so. Over the years, a number of scholars have noted the need for studies on the role of personality in the teacher labor market (Iverson, Olekalns, & Erwin, 1998; Kokkinos, 2007; Mills & Huebner, 1998). Indeed, in the 1960s several research teams administered a series of studies that attempted to link teacher personality with effectiveness (e.g., Belgard, Rosenshine, & Gage, 1968; Getzels & Jackson, 1963). However, very few current researchers do this work, none of them using a holistic conceptualization of personality, and previous efforts have focused on effectiveness, not retention. Work in psychology and education points to this as a potentially fruitful question to explore. Roberts, Kuncel, Shiner, Caspi, and Goldberg (2007), for example, make a convincing meta-analytic case that personality variables have as much impact on important life outcomesincluding occupational attainmentas socioeconomic status and cognitive ability. And several studies suggest that much of the variance in teacher burnout does in fact occur at the individual rather than at the institutional level (McCarthy, Lambert, ODonnell, & Melendres, 2009; Parker & Martin, 2009; Zellars, Perrewé, & Hochwarter, 2000). Finally, Rimm-Kaufman and Hamre (2010) have specifically advocated using McAdams three-layer model of personality to better understand teachers and their careers.
McAdams posits that personality is constructed in three layers. The first layer is traits. Personality traits are individual dispositions that appear early in life and remain relatively stable over the lifespan and across situations (McAdams & Olson, 2010). These basic traits are often operationalized by using measures of the Big Five, or five tendencies that have been shown to distinguish human beings across cultures and over the lifespan. These are: extraversion (the tendency to be social, outgoing, and excitable), agreeableness (the tendency to be cooperative and kind, to get along with others), neuroticism (the tendency to feel many strong negative emotions), openness to experience (the tendency to be curious and show creativity and intellect), and conscientiousness (the tendency to act responsibly, dependably, and dutifully).
A scattered series of studies suggest that some of these traits may predict teacher satisfaction and retention. High neuroticism or negative affect is associated with higher stress levels, lower job satisfaction, and lower retention rates for workers in general (Judge, Heller, & Mount, 2002) and for individuals in caring professions like teaching in particular (Cano-García, Padilla-Muñoz, & Carrasco-Ortiz, 2005; Fontana & Abouserie, 1993; Griffith, Steptoe, & Cropley, 1999; Iverson et al.,1998). And although none of the findings are as robust as those for neuroticism and its correlates, other studies suggest that extraversion (Kokkinos, 2007; Mills & Huebner, 1998; Zellars et al., 2000), conscientiousness (Kokkinos, 2007), agreeableness (Cano-García et al., 2005; Mills & Huebner, 1998; Zellars et al., 2000), and openness to experience (Kokkinos, 2007; Zellars et al., 2000) can also protect against burnout among workers in the caring professions.
McAdams second layer of personality consists of characteristic adaptations. These are the typical ways individuals manage various circumstances, decide what they want, and take action to get it. These variables tend to change much more over the course of the lifespan than traits. Researchers often operationalize characteristic adaptations by measuring peoples values, goals, and psychological strengths.
How might values, goals and psychological strengths predict teacher commitment? To begin, teachers or other individuals who pursue public service careers may be more inclined to hold values like universalism and less likely to hold values like power. Likewise, when they set personal goals, literature suggests those with a long-term commitment to the occupation may tend to set goals that are more prosocial (Beltman, Mansfield, & Price, 2011; Brunetti, 2001; Lortie, 1975). Research also suggests that long-term teachers goals may be less ambitious than those than those who quickly leave the occupation (Dohmen & Falk, 2010; Olsen & Anderson, 2007), as teaching is an occupation that allows little room for advancement, increase in responsibility, and public recognition. Finally, possessing certain work-specific psychological strengths such as resilience, hope, and optimism might also buffer against attrition in the labor market in general and in teaching in particular. For example, a recent study by Robertson-Craft & Duckworth (2014) found grit to be predictive of effectiveness and retention in novice teachers.
McAdams third and final personality layer is personal narratives. These are the ways people think about and recount the important events and circumstances through which they have lived both. Three constructs often used by narrative psychologists readily apply to peoples work lives: agency, communion, and the integration of agency and communion, or enlightened self-interest (Frimer, Walker, Dunlop, Lee, & Riches, 2011).
Agency and communion, analyzed separately, represent two motivations scholars and philosophers consider fundamental in human life. Agency refers to a concern for the promotion or building up of the selfa desire to become stronger, smarter, or wiser, or to make oneself more powerful and autonomous. Communion involves motivation to join with, love, care for, build relationships with, or even sacrifice oneself for others. These motivations are important across life contexts but may be particularly important when it comes to work in the caring professions.
Historically, agency and communion have been considered to be in tension with each other, two opposing motivations. However, recent work suggests that these two narrative themes might work together in the personal narratives of individuals with a strong feeling of obligation to the public good. These individuals seem to reconcile these two strivings by attaining feelings of personal success through helping others or by accomplishing impressive tasks only with the help and solidarity of friends, family, and colleagues (Frimer et al., 2011; Walker & Frimer, 2007). Frimer et al. (2011) call this tendency enlightened self-interest. Freedman & Appleman (2009) provide some evidence that enlightened self-interest may play a role in teacher retention. They found that committed urban teachers, in contrast to urban teachers who had left the classroom, found deeply meaningful the process of building relationships with and helping their students (communion) and also displayed hard work and persistence (agency).
McAdams model with its three layerstraits, characteristic adaptations, and personal narrativesis a useful framework to use to study teacher personality for several reasons. First, it encourages nuance and complexity in the measurement of personality. Many psychologists and laypeople construct personality in a simplistic way. They consider it to consist of temperaments, inherited before birth, that are difficult to alter. In many ways, they simply stop at McAdams first layerthe traits layer. In contrast, McAdams model gets at more abstract and changeable elements of personality that speak to the broader psychological design of a persons life. It helps researchers avoid painting a one-dimensional, essentialist portrait of a person, what McAdams calls a psychology of the stranger (1992).
Relatedly, McAdams model is a developmental one. It allows for change over time. Although the model acknowledges more or less stable traits that might predispose a person to enjoy or excel at the work of an educator, it also embraces the idea that people might be able to change how they view the events of their lives and careers. Recent work, in fact, supports the idea that teachers may be able to change their perceptions of aspects of their work environment in order to feel more satisfied (Fernet, Guay, Senécal, & Austin, 2012; Tsouloupas et al., 2010). Using McAdams model of personality to study teacher retention, then, should provide insight into ways people who have already chosen teaching can be supported so they are more likely to remain in the occupation.
The sample for this study consists of 107 graduates of a single teacher licensure program. They are classified as academically elite because they graduated from this highly selective program where the average verbal GRE score for admitted applicants is in the 90th percentile, the average quantitative GRE score in the 68th, and the average analytical writing score in the 80th. Approximately 550 individualsall of the programs graduates who had earned their teacher certification through the programwere contacted through the programs alumni email listserv about an opportunity to participate in a study requiring them to complete two online surveys about themselves and their experiences as teachers and inviting people with a range of career pathsboth those still in the classroom and those who have left for other opportunities. A sample of 125 total teachers, approximately half still in teaching and half who had left teaching, was sought, and once 125 individuals in these proportions had volunteered to participate, enrollment was cut off. Of the 125 initial volunteers, 118 completed full study participation. Eleven moved from teaching into administrative roles and are excluded from these analyses.
These individuals completed two extensive surveys aimed at gathering a rich and holistic set of personality variables. One survey consisted of multiple choice and short answer questions about demographics, backgrounds, teaching histories, and personal qualities, and the other consisted of longer, open-ended answers to narrative prompts. Because these surveys required a substantial time commitment, participating individuals were paid $50.
To begin to explore the role of teacher retention, the sample size was reduced to 61 to capture only those teachers with a very long-term commitment (N=34, commitment of at least 15 years, M=25.43 years) and those with a short-term commitment (N=27, commitment of 7 or fewer years, M=4.14 years). Doing so carries two important benefits. First, in an exploratory study with many variables and a modest sample size, looking at the ends of the teacher commitment spectrum helps to identify the personality variables that most strongly differentiate those who are committed for the very long term from those who plan to quickly leave, or have already left. These results can be used later to test whether the personality variables that differentiate very long-term committers from very short-term committers also predict occupational commitment in a continuous way across the full sample. Second, reducing the sample in this way allows for the use of discriminant function rather than regression analysis, which is a more appropriate choice for an exploratory study of personality. Discriminant function analysis can reveal a constellation of personality variables that work together to predict commitment to teaching, whereas regression analyses can only show which individual personality variables have predictive power controlling for a variety of other personality variables. Personality variables do not operate in isolation in the real world; they interact, combine, and conflict with each other. Beginning with discriminant function analysis allows for this complexity.
Teacher commitment is calculated by adding together the number of years the participants have already taught plus the number of years more they plan to teach. Defining the dependent variable in this way captures some of the nuance inherent in teacher commitmentit picks up both time teachers have already devoted to the job as well as their plans for the future. Among the long-term committers in this sample, 83% had already taught for at least 5 years, 47% for at least 7 years, 26% for at least 10 years, and 9% for at least 15. Among the short-term committers, 63% had taught for at least 2 years, 8% for at least 5, and none (by definition) longer than 7. Sixty-eight percent of the short-term committers had already left the occupation by the time they completed the surveys.
The sample of long- and short-term committers is 73% female and 26% male (some percentages do not add up to 100 due to rounding), reflecting the makeup of all graduates of this teacher preparation program (74% female, 26% male) and national trends in the gender breakdown of classroom teachers. They range in age from 2457: 48% in their twenties, 35% in their thirties, and 17% in their forties and fifties. In this sample, 78% of the participants identify as white, 10% as Asian/Pacific Islander, 3% as Black, 2% as Latino, and 3% as bi- or multiracial with 4% not reporting. This racial/ethnic background reflects the overall makeup of the backgrounds of students who graduated from this teacher preparation program (78% white, 10% Asian/Pacific Islander, 1% Black, 2% Latino, and 1% bi- or multiracial with 8% not reporting). This is also an affluent group. Eighty-seven percent of participants list the income level of their family of origin as being middle class, upper middle class, or very affluent. And nearly all of the participants in this study had, by selection, attained the same level of educationa masters degreealthough 6% had also earned a terminal degree like a JD or PhD.
This study explores a rich set of personality variables at all three layers of McAdams model in order to identify those that best distinguish academically elite teachers with a long-term commitment to the occupation from those with a short-term commitment. There are, of course, hundreds of personality variables that could potentially be included. Therefore, the focus here is on (1) those variables research and theory most strongly suggest might be associated with job retention in general and retention in teaching specifically, and (2) measures that are widely used among personality psychologists and that have been shown to be valid in many research studies across a wide range of cultural contexts. The included variables at each of McAdams layers of personality are listed below.
For this personality layer, the Big Five personality traits of extraversion, agreeableness, neuroticism, openness to experience, and conscientiousness were used, measured by the NEO-FFI-3 (Costa & McCrae, 1992).
To measure participants most strongly held values, the Portrait Values Questionnaire (Schwartz, 1994, 2009; Schwartz et al., 2001) was used. For this questionnaire, participants respond to a series of items designed to measure which of 10 basic values (universalism, benevolence, conformity, tradition, security, power, achievement, hedonism, stimulation, self-direction) are most important to them.
Participants were also asked to list their goals according to a procedure developed by Emmons (1986). In it, respondents list five goals toward which they are currently working and then rank the goals on a 5-point scale to indicate how difficult the goal is. This measure was used to produce two different variables: goal difficulty and goal prosociality. To measure difficulty in this study, the difficulty rankings of the five goals were added together. This provides a general sense of the level of challenge to which each participant aspires across life domains. For prosociality, the goals were coded for the presence or absence of a concern for the good of others or society in general. Two coders who had attained interrater reliability (Kappa=.905), blind to identifying information about the participants, completed this coding. The scores for all five goals were added together as a total measure of goal prosociality.
Psychological strengths, in particular the qualities of hope, self-efficacy, optimism, and resilience, combined to represent a higher-order construct called psychological capital, were measured using the PCQ-24 scale (Luthans, Youssef, & Avolio, 2007).
Participants were also asked to provide detailed narrative accounts of a life high point and teaching high point, and these accounts were coded for agency, communion, and the integration of agency and communion, or enlightened self-interest (Frimer et al., 2011).
For agency and communion, two coders, blind to identifying information about the participants, coded the life high point and teaching high point according to a procedure developed by McAdams (2001). Each account was coded for the total number of agency and communion codes (possible range of 04) represented in the narrative. Instances of agency included moments in which participants expressed pride at a personal or professional achievement, such as winning an award, increasing student achievement on standardized tests, or receiving praise from a superior. Communion was usually manifested as concern for students, friendship with personal or professional contacts, or, in peoples personal lives, a commitment to a romantic relationship. The two coders worked together on several preliminary sets of narratives to achieve interrater reliability (for agency: ICC=.807, two-way consistency model, single measures; for communion: ICC=.835, two-way consistency model, single measures).
To measure enlightened self-interest, two members of the study team (the author and an assistant) worked together to develop a holistic coding scheme for the integration of agency and communion. The coders began by reading and discussing a series of life and teaching high points. Enlightened self-interest was manifested in a variety of ways. In many accounts, participants would recall working very hard and feeling personally successful because they cared for and wanted to help their students (agency either stemming from or leading to feelings of communion). For example, one teacher describes the many positive emotions she felt when the debate team she sponsored won an important competition:
One fantastic moment in my life is my debate team coming in first place for the city championships. It was our very first year in the debate tournament and we had all been working hard to make a name for ourselves. I was amazingly excited for the kids and I felt fantastic as it was a validating experience for all the work that we had done to be able to not only place one team in the championships, but have two teams competing in the top four teams for places. Being our first year in debate, this moment felt fantastic because it proved how hard we had work as a group to get there and that we were truly a force to reckon with. I think this moment conveys my competitive streak and my commitment to my team. I did not win much personal notoriety from it, but it really made me team feel good about themselves and the work that they had done. I love to win, though I try not to let my kids see that part of it.
Here, both agentic and communal motivations are apparentboth a competitive streak and a commitment to [the] team. The teacher obviously cares deeply about her students and their achievement. She also focuses heavily on how hard she and the team worked and how proud they were to achieve together.
Another common type of enlightened self-interest found in these narrative accounts involves working in solidarity with another adulta parent, teacher, or group of teachersin order to accomplish a difficult goal. For example, one teacher wrote:
A real teaching high must be when I team-taught a poetry class with a beloved friend and mentor. She and I planned the class, implemented the class, and each day dove into the class together. It was an awesome experience because we learned from each other. We taught each other new things about poetry and poets. We taught each other new pedagogical methods. We listened to each others' ideas and built off each others' strengths. The students loved it because they heard two (sometimes disagreeing) voices in the room about poetry. Throughout the experience, I felt lucky, too. I felt lucky that we were given permission to do this, that we had taken the necessary time to plan it, and that each day we worked intuitively for the students and their needs. She is a master teacher, so I learned how to better relate to students, how to think more deeply about my teaching practice, and how to strengthen my relationship with my own writing.
The teacher focuses here on how much she learned and improved her practice, growing into a better, more competent teacher because of this experience. She is able to do this because of communion, because of a meaningful connection with a beloved friend and mentor who models excellent teaching for her.
Once both coders had a clear idea of what enlightened self-interest looked like in the narrative accounts, they coded each life and teaching high point, blind to identifying information about the participants, for the presence (+1) or absence (+0) of enlightened self-interest. The two coders worked together in preliminary rounds of coding to attain interrater reliability (Kappa=.918). All disagreements were settled by discussion.
Background and School Environment Variables
Participants also gave demographic information and listed the name of the most recent school where they taughteither their current school or the school they left when they exited the occupation. This information was used to collect objective data from the National Center for Education Statistics Public School Search and Private School Search as well as from school websites, such as information on the racial breakdown of students in the school and the schools urbanity.
The variables included in this analysis, then, along with their associated descriptive statistics for long-term committers and short-term committers, are reported in Table 1.
Table 1. Descriptive Statistics, Long-Term vs. Short-Term Committers
Before conducting discriminant function analyses, it is important to explore correlations between study variables in order to understand how certain traits, characteristic adaptations, and personal narratives might overlap and interact. All correlations between psychological variables are reported in Table 2.
Table 2. Correlations Between All Psychological Variables
The many variables in this table behave as expected. At the trait level, for example, extraversion, a measure not only of someones outgoingness but also positive energy and excitability, is negatively correlated with neuroticism (r=-.432, p<.01), a measure of ones tendency to feel negative emotions. Likewise, participants values tend to be highly correlated with related values (e.g., conformity and tradition; r=.520, p<.01) and negatively correlated with opposing values (e.g., universalism and power; r=-.368, p<.01). The variables created to measure the prosociality and difficulty of individuals goals are not correlated with any of the other variables except communal themes in personal narratives (r=.181, p<.05) in the case of goal prosociality and neuroticism in the case of goal difficulty (r=.206; p<.05). Finally, the narrative variables show a few correlations to note. Agency themes are correlated with openness to experience (r=.198, p<.05), psychological capital (r=.185, p<.05), and, negatively, with conformity (r=-.200, p<.05). Communal themes are correlated with extraversion (r=.205, p<.05), agreeableness (r=.340, p<.01), neuroticism (r=-.242, p<.01), goal prosociality (r=.191, p<.05), benevolence (r=.221, p<.05), and enlightened self-interest (r=.397; p<.01). Finally themes of enlightened self-interest are correlated with extraversion (r=.279; p<.01), agreeableness (r=.243, p<.01), and benevolence (r=.217; p<.05).
Next, discriminant function analyses were used to explore which personality variables were most predictive of being categorized as a long-term committer to teaching. Two sets of discriminant function analyses were conducted. The first does not take into consideration school advantage. The second does: the percent of white students at the teachers current or most recent school (usually the exit school) is entered as a variable. Although a relatively crude measure of school advantage, the percentage of white students at a school is used over other options, such as the percent of students who receive free or reduced priced lunch, because it is available for a greater number of schools. Furthermore, because of racial inequalities in the U.S. education system, school racial breakdown is highly correlated with student advantage. In this studys sample, for example, school percent white students and school percent of students on free/reduced lunch shows a very strong relationship (r =.864, p<.001).
Including this measure of school environment allows for an exploration of whether there are personality variables that predict commitment over and above the work environments in which these teachers have found themselves. Age is included in both analyses since commitment is calculated in part by how many years a person has already taught.
Because it is well-suited for a large number of predictor variables, stepwise discriminant function analysis was used. This type of analysis combs through the included variables one at a time and generates a function that includes an independent variable only if it significantly differentiates long-term committers from short-term committers. Once the next most discriminating variable added into the model, combined with the impact of the previously entered variables, no longer categorizes people into the given groups at a significantly higher rate than chance, the process stops and the function is complete. In these analyses, variables are included based on how well they minimize the Wilks lambda, a measure of the amount of variance between short-term and long-term committers not explained by the discriminant function.
In Model A, the model run without the school advantage variable, four variablesage plus three of the personality measuressignificantly discriminate between long-term and short-term committers. The Wilks lambda for the discriminant function is .557, with approximate χ2 (4, N = 60) = 33.343, canonical correlation = .665, p<.001. The Wilks’ lambda indicates that about 56% of the variance between the groups is not explained by the discriminant function. The canonical correlation coefficient—.665—is the association between the discriminant function and the dependent variablebeing a long-term committer. The significant chi-square result indicates the created discriminant function does better than chance at predicting who will fall into the long-term and short-term committer categories.
The four variables that, together, maximally differentiate the long-term committers from short-term committers are listed in Table 3. Also listed in Table 3 is the step at which each variable is added, the Wilks lambda at each step, standardized canonical coefficients, and the p value for the function at each step. Participant age is added at Step 1; in other words, age minimizes Wilks lambda the most. Age is followed by goal prosociality, then enlightened self-interest, then goal difficulty. The other variables are discarded because they do not significantly distinguish between groups. This function classifies 75.7% of the participants into the correct grouplong- or short-term committers.
Table 3. Stepwise Discriminant Function Analysis Results, Model A
Next, a second stepwise discriminant function analysis was runModel Bthat includes the percentage of white students enrolled at the most recent school where the teachers worked. This measure of school advantage does not significantly discriminate between groups. Five other variables do: age, goal prosociality, goal difficulty, agency/communion integration and, in this model, power. The Wilks’ lambda for the discriminant function is.397, with approximate χ2 (5, N = 48) = 41.111, canonical correlation = .777, p<.001. This indicates that about 40% of the variance between long- and short-term committers is not explained by the created function, and that the function does discriminate between these groups better than chance. Table 4 shows results for this second stepwise discriminant function analysis.
Table 4. Stepwise Discriminant Function Analysis Results, Model B
School percent white students is not chosen as one of the variables that best discriminates between groups. However, including this variable in the analysis bumps valuing power up in importance so it makes a significant enough difference to include in the discriminant function. This likely occurs because school percent white students was missing for some participants, and as the sample size decreases, it changes slightly. In this case, the coefficient is negative; teachers with a long-term commitment value power less than teachers with a short-term commitment. This function categorizes 81.4% of the participants correctly into the groups: long-term committers or short-term committers.
One final set of analyses was run to test whether these personality variables predict commitment to teaching in a continuous way in the sample as a whole. Two composite variable were created by calculating the mean of z-scores for (1) goal prosociality, goal difficulty, and enlightened self-interest and (2) goal prosociality, goal difficulty, enlightened self-interest, and valuing power (valuing power is negatively associated with teacher occupational commitment, so it was subtracted, rather than added, to the other variables z-scores here). Two regressions were run, each testing a composite variableone without the control for school advantage and one with. Results are reported in Table 5.
Table 5. Linear Regressions Predicting Commitment to Teaching in Full Sample
Results suggest that the composite variables do predict retention in the sample as a whole. Composite Variable 1 (the variable that does not include valuing power) predicts occupational commitment at a significant level when school advantage is not controlled for (β=.342, p=.000) and at the trend level when school advantage is controlled for (β=.203, p=.052). Composite Variable 2 (the variable that does include valuing power) predicts commitment with (β=.322, p=.003) and without (β=.413, p=.000) the school advantage control.
The most important personality variables that distinguish long-term and short-term committers are goal prosociality, goal difficulty, and enlightened self-interest. These results point to a particular personality profile that manifests itself across two layers of McAdams' model. First, at the characteristic adaptations layer, long-term committers set goals that are ambitious. Their mean total goal difficulty score is 18.57, or on average a ranking of 3.71 per goal, a score that indicates their aims fall between a ranking of moderately difficult and difficult, closer to the difficult end of the spectrum (vs. a 3.36 for short-term committersa score closest to a ranking of "moderately difficult"). Furthermore, these goals tend to be prosocial, or focused on benefiting other people or society as a whole. The long-term committers listed, on average, 1.8 prosocial goals while the short-term committers listed 1.1.
The motivation to pursue aims that are both difficult and prosocial is reflected in long-term teachers personal narratives. When they recount life and teaching high points, these participants weave together a desire to build up the self with a commitment to building up others. The teachers with a long-term commitment to the occupation included 1.87 instances of enlightened self-interest in their life and teaching high points, versus 1.69 instances for short-term committers. Furthermore, the regression analyses suggest that these findings hold beyond the longest- and shortest-term committers, predicting occupational commitment in a continuous way in the sample as a whole.
There are, of course, several important limitations to discuss. First, despite the fact that their standardized test scores suggest the individuals in this sample might excel as educators, we simply do not know if they are or were especially effective teachers. They are deemed academically elite because they gained admission to and graduated from a highly selective teacher preparation program. They are not to be confused with individuals who have documented success as teachers or who meet some external standard such as being highly qualified under the No Child Left Behind law. Understanding what personal qualities keep academically elite teachers, as defined, in the classroom is, in general, useful information, as faculty stability is beneficial for students (Ronfeldt et al., 2012). However, if the individuals in this sample who left teaching were not effective teachers, this turnover may serve an important purpose. Future studies should explore the relationship between effectiveness and retention in more detail; qualitative work like that of Cochran-Smith et al. (2012) and Barnatt et al. (2016) represents a good starting point for understanding this interplay in rich ways.
On a related note, individuals who moved out of teaching and into administrative positions were excluded from the study. The role personality plays in the career trajectories of these individuals, who may be especially ambitious educators and who have excelled enough as teachers to be entrusted with school leadership positions, might be a fruitful topic for future studies.
Second, a large number of personality variables were explored using a relatively small sample of teachers and former teachers. Stepwise discriminant function analysis works well in such situations, but two caveats are important to mention. First, the variables that were chosen for the discriminant function correlated with several other study variables that were not chosen. Both goal prosociality and enlightened self-interest were positively correlated with communal themes in narratives (r=.191, p<.05; r=.397, p<.05), goal difficulty was associated with neuroticism (r=.206, p<.05), and enlightened self-interest was also associated with extraversion (r=.279, p<.01), agreeableness (r=.243 p<.01), and benevolence (r=.217 p<.05). There is a chance, then, that these correlated variables not chosen for the discriminant function might significantly predict commitment were they tested alone but hide behind variables that predict it just a bit better in the discriminant function analysis. To ensure that communion, neuroticism, extraversion, agreeableness, and benevolence do not in fact predict teacher commitment, ANCOVAs (with age as a covariate) were run testing for differences between long- and short-term committers, and regressions were run (again with age as a covariate) testing the value of these variables in predicting teacher commitment in the sample as a whole. No significant differences were found.
A second disadvantage of running analyses on large number of variables using a small sample is the fact that all of the tests run were rather blunt. There was only enough statistical power, for example, to look at long-term committers versus short-term committers, not at smaller slices of the sample, such as teachers of different genders, races, social classes, or generations. There might be very real and very interesting personality differences among these subgroups. In addition, using discriminant function analysis necessarily entailed looking first only at the most highly committed and the most highly uncommitted individuals in this sample. An analytical approach that begins by looking at the ends of the commitment spectrum and leaves out the middle has its disadvantages. The fact that the variables that significantly differentiated long-term and short-term committers in the discriminant function analyses also significantly predicted commitment across the entire sample in the regression analyses is reassuring, However, an important goal for future work will be to test hypotheses about the role of personality in teacher commitment across the full spectrum of a range of samples of teachers and former teachers.
Finally, although the personality measures administered to participants were chosen with care, there is always a chance in exploratory studies like this one that some important personality variable that distinguished the groups was not considered. Much remains to learn about the role of personality in teacher retention, and future studies should consider the role of other personality variables, include measures of teacher effectiveness in their data collection, and recruit larger samples so more finely cut analyses can be completed.
This study sought to move our understanding of teacher retention forward in several ways. In a policy climate where factors external to educators, such as salary scales and tenure laws, are the focus of conversation, it explores the internal, person-level attributes that may act as insulating factors (Hamman et al., 2010), motivating some teachers to stay in the occupation much longer than others. It also explores these variables using a complex and holistic model of personality. When personality is considered in studies of job satisfaction and retention, it is too often operationalized in a simplistic way. Including measures at all three layers of McAdams model of personality allows a nuanced portrait of individuals with a long-term commitment to classroom teaching to emerge. Finally, the study focused on academically elite teachers. These individuals represent a talented group of people with a wide variety of career options available to them who may be more likely than the average person to excel as an educator.
The results are important for several reasons. First, this study suggests that teacher personality does play a role in the retention of academically elite teachers. This is an important finding in and of itselfit suggests a fresh angle from which future researchers should approach the issue of retention. It will still be useful, of course, to discuss policies that deal with the circumstances around teachers, such as the advantages of different salary schedules or what might be an ideal class size, but personality should also be an important part of the conversation.
Second, the findings show that the personality variables that best distinguish long- and short-term committers all occur at the dynamic, malleable layers of McAdams model of personality. From a practical perspective, this is exciting. It suggests that it might be fruitful to consider interventions that could bolster retention-predicting qualities in young teachers, encouraging them to stay longer in the occupation or in a particular school and thereby build a more stable teacher labor force. For example, if a tendency toward enlightened self-interest is truly important in encouraging a long-term commitment to the classroom, teacher educators should consider whether it might be conceivable to teach such a perspective. To give another possibility, it seems that academically elite teachers might sustain their commitment to the occupation by pushing themselves to achieve ambitious goals within a work context that does not provide much public recognition of accomplishments. New, driven teachers, then, might consider adopting this as a coping strategy and make a habit of privately challenging themselves to attain difficult aims.
Finally, the findings suggest that ambition plays a large role in the long-term retention of the academically elite teachers in this sample. This result is particularly interesting given the popular cultural narrative in the United States that K12 teachers are markedly unambitious. Americans tend to believe teachers are either noble and self-sacrificingexhibiting a notable lack of self-interest because they care so much about the students in their chargeor risk-averse, sometimes perniciously so, and mostly interested in set salary schedules and job security (Gabriel & Lester, 2012; Lortie, 1975; Shannon & Crawford, 1998). The teachers in this academically elite sample do not conform to that stereotype. The more ambitious the individual, as measured by the difficulty of the goals they pursue and the agency (coupled with communion) they include in their personal narratives, the more highly committed they are to the classroom. These results are reminiscent of Johnson & Birkelands (2003) finding that teachers who found a sense of success in their school contexts were more likely to stay in the occupation and in their school. They also reflect Maier's (2012) characterization of the top college graduates who join Teach for America. He argues that they find it especially appealing that admission to the program means they are "doing good and doing well," or pursuing work that is both moral and prestigious.
These results, then, have implications for administrators and policymakers as well. They should consider ways they can reward teachers and ensure they have the autonomy to pursue the kinds of goals that are important to them. For example, insisting too strongly on highly standardized, scripted curricula might chip away at promising long-term teachers desire for autonomy, achievement, and recognition. On the other hand, putting into place policies like certain kinds of like merit pay plans designed to pit individual teachers against each other may garner resistance because of long-term teachers desire to work in solidarity with their colleagues. In general, considering the evidence that the long-term teachers in this sample set goals that are both ambitious and prosocial and demonstrate enlightened self-interest in their narratives, administrators, policymakers, and the public should encourage a work environment in which teachers can make meaningful connections with others as they strive to attain the ambitious goals they have set for themselves.
Bacalod, M. P. (2007). Do alternative opportunities matter? The role of female labor markets in the decline of teacher quality. Review of Economics and Statistics, 89(4), 737751.
Barnatt, J., Terrell, D. G., DSouza, L. A., Jong, C., Cochran-Smith, M., Viesca, K. M., Shakman, K. (2016). Interpreting early career trajectories. Educational Policy. https://doi.org/10.1177/0895904815625286.
Barnes, G., Crowe, E., & Schaeffer, B. (2007). The cost of teacher turnover in five school districts: A pilot study. Washington, DC: National Commission on Teaching and Americas Future. Retrieved from http://nctaf.org/wp-content/uploads/2012/01/NCTAF-Cost-of-Teacher-Turnover-2007-full-report.pdf
Belgard, M., Rosenshine, B., & Gage, N. L. (1968, December). The teacher's effectiveness in explaining: Evidence on its generality and correlation with pupils' ratings and attention scores. In N. L. Gage, M. Belgard, D. Dell, J. E. Hiller, B. Rosenshine, & W. R. Unruh, Explorations of the teacher's effectiveness in explaining (Technical Report No. 4, pp. 920). Stanford, CA: Stanford Center for Research and Development in Teaching.
Beltman, S., Mansfield, C., & Price, A. (2011). Thriving not just surviving: A review of research on teacher resilience. Educational Research Review, 6, 185207.
Benner, A. D. (2000). The cost of teacher turnover. Austin: Texas Center for Educational Research.
Borman, G. D., & Dowling, N. M. (2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Review of Educational Research, 78(3), 367409.
Brunetti, G. J. (2001). Why do they teach?: A study of job satisfaction among long-term high school teachers. Teacher Education Quarterly, 28(3), 4974.
Cano-García, F. J., Padilla-Muñoz, E. M., & Carrasco-Ortiz, M. A. (2005). Personality and contextual variables in teacher burnout. Personality and Individual Differences, 38, 929940.
Chetty, R., Friedman, J. N., & Rockoff, J. E. (2014). Measuring the impacts of teachers II: Teacher value-added and student outcomes in adulthood. American Economic Review, 104(9), 26332679. doi:10.1257/aer.104.9.2633
Clotfelter, C., Ladd, H., Vigdor, J., & Diaz, R. (2004). Do school accountability systems make it more difficult for low-performing schools to attract and retain high-quality teachers? Journal of Public Policy Analysis and Management, 23(2), 251271.
Cochran-Smith, M., McQuillan, P., Mitchell, K., Terrell, D. G., Barnatt, J., DSouza, L., Gleeson, A. M. (2012). A longitudinal study of teaching practice and early career decisions: A cautionary tale. American Educational Research Journal, 49(5), 844880.
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources.
Costigan, A. T. (2005). Choosing to stay, choosing to leave: New York City teaching fellows after two years. Teacher Education Quarterly, 32(2), 125142.
Day, C., & Gu, Q. (2010). The new lives of teachers. New York, NY: Routledge.
Dohmen, T., & Falk, A. (2010). You get what you pay for: Incentives and selection in the education system. Economic Journal, 120(546), F256F271.
Emmons, R. A. (1986). Personal strivings: An approach to personality and subjective well-being. Journal of Personality and Social Psychology, 51(5), 10581068.
Fernet, C., Guay, F., Senécal, C., Austin, S. (2012). Predicting intraindividual changes in teacher burnout: The role of perceived school environment and motivational factors. Teaching and Teacher Education, 28, 514525.
Fontana, D., & Abouserie, R. (1993). Stress levels, gender, and personality factors in teachers. British Journal of Educational Psychology, 63, 261270.
Freedman, S. W., & Appleman, D. (2009). How teacher education can contribute to teacher retention in high poverty, urban schools. Journal of Teacher Education, 60(3), 323337.
Friedman, I. A. (1995). Student behavior patterns contributing to teacher burnout. Journal of Educational Research, 88(5), 281289.
Frimer, J. A., Walker, L. J., Dunlop, W. L., Lee, B. H., & Riches, A. (2011). The integration of agency and communion in moral personality: Evidence of enlightened self-interest. Journal of Personality and Social Psychology, 101(1), 149163.
Gabriel, R., & Lester, J. (2012, April). Constructions of value-added measurement and teacher effectiveness in the Los Angeles Times: A discourse analysis of the talk of surrounding measures of teacher effectiveness. Paper presented at the American Educational Research Association Conference, Vancouver, BC.
Getzels, J. W., & Jackson, P. W. (1963). The teachers personality and characteristics. In N. L. Gage (Ed.), Handbook of research on teaching (pp. 506582). Chicago, IL: Rand McNally.
Goldhaber, D. (2002). The mystery of good teaching. Education Next, 2(1), 5055.
Griffith, J., Steptoe, A., & Cropley, M. (1999). An investigation of coping strategies associated with job stress in teachers. British Journal of Educational Psychology, 69, 517531.
Guarino, C. M., Santibañez, L., & Daley, G. A. (2006). Teacher recruitment and retention: A review of recent empirical literature. Review of Educational Research, 76(2), 173208.
Guin, K. (2004). Chronic teacher turnover in urban elementary schools. Education Policy Analysis Archives, 12(42), 130.
Hakanen, J. J., Bakker, A. B., & Schaufeli, W. B. (2006). Burnout and work engagement among teachers. Journal of School Psychology, 43, 495513.
Hamman, D., Gosselin, K., Romano, J., & Bunuan, R. (2010). Using possible-selves theory to understand the identity development of new teachers. Teaching and Teacher Education, 26, 13491361.
Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (1999, April). Do higher salaries buy better teachers? (Working Paper No. 7082). Cambridge, MA: National Bureau of Economic Research.
Hargreaves, A., & Tucker, E. (1991). Teaching and guilt: Exploring the feelings of teaching. Teaching and Teacher Education, 7(56), 491505.
Ingersoll, R. M. (2003, September). Is there really a teacher shortage? (Center for the Study of Teaching and Policy Document R-03-4). Retrieved from http://repository.upenn.edu/cgi/viewcontent.cgi?article=1133&context=gse_pubs
Ingersoll, R. M., & May, H. (2012). The magnitude, destinations, and determinants of mathematics and science teacher turnover. Educational Evaluation and Policy Analysis, 34(4), 435464.
Iverson, R. D., Olekalns, M., & Erwin, P. J. (1998). Affectivity, organizational stressors, and absenteeism: A causal model of burnout and its consequences. Journal of Vocational Behavior, 52, 123.
Jacob, B. A. (2007). The challenges of staffing urban schools with effective teachers. Future of Children, 17(1), 129153.
Johnson, S. M., & Birkeland, S. E. (2003). Pursuing a "sense of success": New teachers explain their career decisions. American Educational Research Journal, 40(3), 581617.
Johnson, S. M., Kraft, M. A., & Papay, J. P. (2012). How context matters in high-needs schools: The effects of teachers working conditions on their professional satisfaction and their students achievement. Teachers College Records, 114, 239.
Judge, T. A., Heller, D., & Mount, M. K. (2002). Five-factor model of personality and job satisfaction: A meta-analysis. Journal of Applied Psychology, 87(3), 530541.
Kelly, S., & Northrop, L. (2015). Early career outcomes for the best and the brightest: Selectivity, satisfaction, and attrition in the beginning teacher longitudinal survey. American Educational Research Journal, 52(4), 624656.
Kokkinos, C. M. (2007). Job stressors, personality and burnout in primary school teachers. British Journal of Educational Psychology, 77, 229243.
Kyriacou, C. (2001). Teacher stress: Directions for future research. Educational Review, 53(1), 2735.
Loeb, S., Darling-Hammond, L., & Luczak, J. (2005). How teaching conditions predict teacher turnover in California schools. Peabody Journal of Education, 80(3), 4470.
Lortie, D. C. (1975). Schoolteacher (2nd ed.). Chicago, IL: University of Chicago Press.
Luthans, F., Youssef, C. M., & Avolio, B. J. (2007). Psychological capital: Developing the human competitive edge (pp. 237238). New York, NY: Oxford University Press.
Macdonald, D. (1999). Teacher attrition: A review of literature. Teaching and Teacher Education, 15(8), 835848.
Maier, A. (2012). Doing good and doing well: Credentialism and Teach for America. Journal of Teacher Education, 63(1), 1022. doi:10.1177/0022487111422071
McAdams, D. P. (1992). The five-factor model in personality: A critical appraisal. Journal of Personality, 60(2), 329361.
McAdams, D. P. (2001). Coding autobiographical episodes for themes of agency and communion. Retrieved from http://www.sesp.northwestern.edu/foley/instruments/agency/
McAdams, D. P., & Olson, B. D. (2010). Personality development: Continuity and change over the life course. Annual Review of Psychology, 61, 517542.
McAdams, D. P., & Pals, J. L. (2006). A new big five: Fundamental principles for an integrative science of personality. American Psychologist, 61(3), 204217.
McCarthy, C. J., Lambert, R. G., O'Donnell, M., & Melendres, L. T. (2009). The relation of elementary teachers' experience, stress, and coping resources to burnout symptoms. Elementary School Journal, 109(3), 282300.
Milanowski, A. T., & Odden, A. R. (2007). A new approach to the cost of teacher turnover (Working Paper No. 13). Seattle, WA: Center on Reinventing Public Education.
Mills, L. B., & Huebner, E. S. (1998). A prospective study of personality characteristics, occupational stressors, and burnout among school psychology practitioners. Journal of School Psychology, 36(1), 103120.
Murnane, R. J., Singer, J. D., Willett, J. B., Kemple, J. J., & Olsen, R. J. (1991). Who will teach? Policies that matter. Cambridge, MA: Harvard University Press.
Olsen, B., & Anderson, L. (2007). Courses of action: A qualitative investigation into urban teacher retention and career development. Urban Education, 42(5), 529.
Parker, P. D., & Martin, A. J. (2009). Coping and buoyancy in the workplace: Understanding their effects on teachers' work-related well-being and engagement. Teaching and Teacher Education, 25, 6875.
Podgursky, M., Monroe, R., & Watson, D. (2004). The academic quality of public school teachers: An analysis of entry and exit behavior. Economics of Education Review, 23, 507518.
Rimm-Kaufman, S. E., & Hamre, B. K. (2010). The role of psychological and developmental science in efforts to improve teacher quality. Teachers College Record, 112(12), 29883023.
Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (2005). Teachers, schools, and academic achievement. Econometrica, 73(2), 417458.
Roberts, B. W., Kuncel, N. R., Shiner, R., Caspi, A., & Goldberg, L. R. (2007). The power of personality: The comparative validity of personality traits, socioeconomic status, and cognitive ability for predicting important life outcomes. Perspectives on Psychological Science, 2(4), 313345.
Robertson-Craft, C., & Duckworth, A. (2014). True grit: Trait level perseverance and passion for long-term goals predicts effectiveness and retention among novice teachers. Teachers College Record, 116(3).
Rockoff, J. E. (2004). The impact of individual teachers on student achievement: Evidence from panel data. American Economic Review, 94(2), 247252.
Ronfeldt, M., Loeb, S., & Wyckoff, J. (2012, January). How teacher turnover harms student achievement (Working Paper No. 70). Washington, DC: National Center for Analysis of Longitudinal Data in Education Research.
Schwartz, S. H. (1994). Are there universal aspects in the structure and contents of human values? Journal of Social Issues, 50(4), 1945.
Schwartz, S. H. (2009). Basic human values: Theory, methods, and applications. Retrieved from http://segr-did2.fmag.unict.it/Allegati/convegno%207-8-10-05/Schwartzpaper.pdf
Schwartz, S. H., Melech, G., Lehmann, A., Burgess, S., Harris, M., & Owens, V. (2001). Extending the cross-cultural validity of the theory of basic human values with a different method of measurement. Journal of Cross-Cultural Psychology, 32, 519542.
Shannon, P., & Crawford, P. (1998). Summers off: Representations of teachers work and other discontents. Language Arts, 75(4), 255264.
Simon, N. S., & Johnson, S. M. (2015). Teacher turnover in high poverty schools: What we know and can do. Teachers College Record, 177(3), 136.
Tsouloupas, C. N., Carson, R. L., Matthews, R., Grawitch, M. J., & Barber, L. K. (2010). Exploring the association between teachers perceived student misbehavior and emotional exhaustion: The importance of teacher efficacy beliefs and emotion regulation. Educational Psychology, 30(2), 173189.
Walker, L. J., & Frimer, J. A. (2007). Moral personality of brave and caring exemplars. Journal of Personality and Social Psychology, 93(5), 845860.
Zellars, K. L., Perrewé, P. L., & Hochwarter, W. A. (2000). Burnout in health care: The role of the five factors of personality. Journal of Applied Social Psychology, 30(8), 15701598.