Scholarly article on topic 'Teacher attrition the first five years – A multifaceted image'

Teacher attrition the first five years – A multifaceted image Academic research paper on "Educational sciences"

Share paper
Academic journal
Teaching and Teacher Education
OECD Field of science
{Teacher / Attrition / Career / "Longitudinal study"}

Abstract of research paper on Educational sciences, author of scientific article — Per Lindqvist, Ulla Karin Nordänger, Rickard Carlsson

Abstract Based on a longitudinal study on Swedish teachers' (N = 87) career trajectories this article presents a comparison between quantitative and qualitative data within the cohort and puts this in relation to general statistics on teacher attrition. The analysis indicates that caution is advised in interpreting and making use of general statistics. Teacher attrition is a more non-linear and complex phenomenon than what is typically proposed. In many cases drop-outs are temporary. Individuals not only leave, but also return to, the profession over time and their out-of-school experiences can in many cases be understood as individual initiatives to enhance teaching ability in the long run.

Academic research paper on topic "Teacher attrition the first five years – A multifaceted image"

Teaching and Teacher Education 40 (2014) 94-103

Contents lists available at ScienceDirect

Teaching and Teacher Education

journal homepage:


Teacher attrition the first five years - A multifaceted image

Per Lindqvist*1, Ulla Karin Nordänger1, Rickard Carlsson

Department of Educational Science, Linnaeus University, 39182 Kalmar, Sweden



> Combining qualitative data with statistics in a longitudinal study on teachers' career show that:

> Teacher attrition is a more complex and non-linear phenomenon than what is often proposed.

> Teacher drop-outs are in many cases temporary.

Teachers out-of-school experiences can in many cases be understood as enhancing teaching ability.


Article history: Based on a longitudinal study on Swedish teachers' (N = 87) career trajectories this article presents a

Revived 4 July 2013 comparison between quantitative and qualitative data within the cohort and puts this in relation to

Revived in re^ed form general statistics on teacher attrition. The analysis indicates that caution is advised in interpreting and

10 Fetauaiy 2°14 making use of general statistics. Teacher attrition is a more non-linear and complex phenomenon than

Accepted 19 February 2014 , . . ,, , , , ,,■■,,,, ,

what is typically proposed. In many cases drop-outs are temporary. Individuals not only leave, but also

return to, the profession over time and their out-of-school experiences can in many cases be understood

TeaWei^' as individual initiatives to enhance teaching ability in the long run.

Attrition ® 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

Career license (

Longitudinal study

1. Introduction

Teachers are the most important professionals in a country that wants to invest in the future (Svenska Dagbladet, 2010).

This statement from the Swedish education minister exemplifies an international trend in policy that emphasizes the relationship between the competitiveness of a state and the quality of its educational system. Trained and skilled teachers are a fundamental requirement in such reasoning and increasing efforts to provide students with teachers have become a challenging worldwide quest. The UNESCO Institute for Statistics (2009) claims that half of the world's countries need to expand their teaching forces in order to be able to enroll all primary school-age children by 2015. Countries not only in Sub-Saharan Africa have by far the greatest need for additional teachers, but also Western countries such as Ireland, Spain, Sweden and the USA are pointed out as facing

* Corresponding author. Tel.: +46 (0) 480446757.

E-mail address: (P. Lindqvist).

1 Lindqvist and Nordanger are the main authors to the article and have contributed equally.

teaching gaps, although these can be considered as moderate in comparison (ibid.). In the case of Sweden, prognoses indicate that the number of certified teachers in the compulsory school will be too low to cover the demand during the next 20 years. In 2020, the Swedish educational system will, according to national statistics, lack roughly 22,000 teachers, approximately 20% of the teaching workforce (Statistics Sweden, 2012; Swedish National Agency for Higher Education, 2012).

The most common measure to overcome such a shortage of teachers is to try to increase recruitment into the profession. Hence, a number of campaigns to attract young people to teaching has been launched during recent years. In several countries recruiting strategies that involve incentives such as loan subsidy programs, signing bonuses or higher salaries has been the policy responses to the problem (OECD, 2005). In addition, alternative routes into the profession have been put on the agenda by governments around the world. Teach for America and Teach First in England represent only two of the numerous efforts to expand the supply pool of potential teachers globally.

However, statistical findings also indicate that the major problem for schools is not a shortage of teachers coming into the system. The real problem is that, even in countries where sufficient

0742-051X/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (

numbers of teachers are trained, it appears as if many of the newly graduated choose not to go in to teaching at all (Luekens, Lyter, & Fox, 2004) or to leave after just a few years (Cooper & Alvarado, 2006). This observation has been developed in the scholarly literature, notably in the works of Ingersoll (2003, 2007) and hints at a different kind of measure to remedy the shortage of teachers. The alternative it suggests is that it may be a more efficient strategy to put in an effort to retain and support active teachers, or to attract teachers who quit or never started teaching to return to the profession. Putting it metaphorically, it is better to patch the holes in the bucket before trying to fill it up.

The image that comes to mind is of a bucket rapidly losing water

because of holes in the bottom. Pouring more water into the

bucket will not be the answer if the holes are not first patched.

(Ingersoll, 2007, p. 6)

In the case of the Swedish teaching "bucket" there were 235,878 teachers (including pre-school teachers) working in Sweden 2010 (Swedish Government, 2010). Compared to the number of graduated teachers at that time, one can logically conclude that 37,500 of the graduated (16%) were working outside the educational system. If these "missing teachers" were re-recruited to the teaching profession they would, to a large degree, fill up the future shortage of teachers, especially in certain categories.

The ambition in this article is to take a closer look at the holes in the bucket by presenting data from a longitudinal study of Swedish teachers.

In comparison with the rates of turnover from other occupations teaching has higher rates than higher-status occupations (professors), about the same as comparable semi-professions (nurses) and lower than some lower-status occupations (federal clerical workers) (Ingersoll, 2003). Statistics from Sweden shows that the annual attrition rates are about the same for teachers and nurses (Hasselhorn, Muller, & Tackenberg, 2005). Although the level of turnover could be comparable with rates in other professions the importance of teacher attrition lies in its costs for schools and effects on large number of pupils. A less stable teaching force will result in educational and organizational disturbances. Research indicates that teacher turnover has a harmful effect on student achievement, especially in poorly performing schools, and that turnover also negatively affects the students of those teachers who remain in the same school from one year to the next. Thus, even teachers outside of the redistribution — the stayers — are somehow harmed by it (Ronfelt, Loeb, & Wyckoff, 2013). Financial costs also accompany teacher attrition. In an American study Borman and Dowling (2008) claim that the total cost of replacing public school teachers who dropped out of the profession was nearly $2.2 billion in 2001. Developing knowledge about teacher attrition is thus an important issue for both policy and research. Such knowledge could, for example, help policy makers invest in initiatives to identify the teachers most at risk of quitting or most likely to return to teaching and to change the conditions that appear most crucial for the decisions to stay, leave or return.

In relation to the importance of teacher attrition for central educational concerns, comparably little research has been carried out in this area, and the findings are often presented as progress reports or prognoses. Nevertheless such studies provide indications of what to focus on and how to demarcate further research.

The proportion of graduated teachers who drop-out often seems to correlate with the number of years in the profession. Statistical data gives an image of a U-shaped distribution of exits, in which younger and older teachers (retirement excluded) are more likely to leave (Ingersoll, 2001). Extensive quantitative studies from USA estimate that only 40—50% of the graduated teachers are still

working as teachers five years after graduation (Ingersoll, 2003). The situation seems to be similar in Great Britain (Cooper & Alvarado, 2006). Consequently, in this article we will focus on the first five years, which seems to be a particularly critical period in teachers' decision to stay in or leave the profession (see also Hammerness, 2008).

Results from research on teacher attrition are generally on a one-shot basis, drawn from a wide target population of teachers, producing general overviews of a population from a long distance at a particular point of time. In a review of teacher attrition Schaefer, Long, and Clandinin (2012) point out that prior research seems to focus on providing correct answers, quick fixes and de-contextualized data. More rarely attrition is considered as a process over time where cohorts of teachers are followed in longitudinal studies, through extensive parts of their careers, in order to identify typical patterns of development and examine individual variations (Cohen, Manion, & Morrison, 2011). "We need", Borman and Dowling (2008, p. 399) assert, "truly longitudinal data with more than two time points to capture more nuanced pictures of teachers' career trajectories".

Through a unique material consisting of correspondence between 87 teachers and their former teacher trainer, from their graduation continuing up to the present, we are facing the opportunity to follow a cohort of Swedish teachers during their first 19 years after graduation. The informants have, regardless of whether they have been sick, been on parental leave or just have quit working as teachers, continued to participate. The number of teachers in this longitudinal study is small relative to the sample sizes available in teacher-specific databases on which results — as the ones mentioned above — are based. However, our material allows analyses and comparisons that have not previously been possible. In the article we will first present "the overall picture", that is statistics of attrition within the cohort, and compare it with findings from studies on larger databases. Then we will zoom in on qualitative data to capture "close-up pictures" of the individual trajectories during the first five years. The ambition is to show the dynamics in teachers' career trajectories and to put this picture in relation to internationally widespread general statistical overviews.

2. The overview — statistics and previous findings on teacher attrition

There is a small, but growing, body of research on teacher turnover, an umbrella term including teachers who move within (migration) and/or leave from (attrition) teaching. In this article we will concentrate on presenting results concerning teacher attrition in the cohort, the "leavers". We are here primarily interested in teachers' choices to exit from or re-entry into the profession. That is not to say that the great number of teachers staying in the profession but moving to another school is a non-essential question. Luekens et al. (2004) have shown that "the movers" are about as many as those who leave school. For a school with high teacher turnover it makes no difference whether the leaving teachers change to another school or to another profession. The school is, in both cases, negatively affected and must deal with the loss of the teachers.

Attrition rates in developed countries vary. In Germany and France less than 5% of the teachers leave schools within the first five years while comparable rates from USA and UK are 30—50% (Cooper & Alvarado, 2006). In studies of American teachers the annual attrition rose by 41% from 1987 to 2008 (Ingersoll & Merrill, 2012). The same pattern is found in Sweden where an increased rate can be discerned over the last 30 years. The average frequency among Swedish teachers was doubled during the 1980s and 1990s and has continued to grow ever since. Teachers that have graduated

over a five-year period and who are still in the Swedish teacher register (a database containing yearly information about all teachers active in Swedish schools) when the period is summed up have declined sharply. In 1998, when our cohort of teachers had been working for five years, national statistics show that 82% of the recently graduated were still working as teachers after five years. In 2008 the number had decreased to 68% (Swedish Government, 2010). These increasing attrition rates are not related to fluctuations in the pupil population. In order to understand them we must turn to other explanations (discussed in the following sections).

In international research on who leaves, why they leave and what kind of schools they leave behind some broad trends can be discerned. Results from an American perspective (Borman & Dowling, 2008; Ingersoll, 2001; Luekens et al., 2004) suggest that teachers who are more likely to be "leavers" are female, white, married, working within special education, math or science. The schools they are leaving are often urban or suburban with high enrollments of poor, minority and low-achieving students. This pattern is also indicated in studies on teacher attrition and turnover in OECD (2005)-countries.

There is some evidence that pay matters in teachers' decisions to stay or leave (Johnson, Berg, & Donaldson, 2005; Kyriacou, Kunc, Stephens, & Hultgren, 2003) but it does not make the entire difference. Organizational factors within schools, such as lack of support from administrators, student discipline issues and lack of input and decision-making power seem to be playing a larger role (Borman & Dowling, 2008; Kyriacou et al., 2003). Johnson et al. (2005) report that high ranked reasons for leaving schools are for example inappropriate or unmanageable assignments, accountability pressures and paperwork. In a Swedish survey of teachers working conditions (Statistics Sweden, 2005) the respondents ranked opportunities to find other work, no prolongation of present work, poor psycho social working environment and low pay as the most common reasons for leaving.

However, all the results above have been generated out of databases or large surveys, at a one-shot basis and/or drawn from retrospective data. Information about teacher attrition as a process over time can only be found in the few longitudinal studies that have been conducted during the last decade. Wilhelm, Dewhurst-Savellis, and Parker (2000) have, by self-report measures in five-year intervals, followed 156 teachers for 15 years. The results show the same attrition patterns as earlier statistical studies, but they also found that the "leavers" often had a more negative image of the profession prior to entry than those who stayed. The authors suggest — in relation to the findings — that factors related to working conditions may be of less importance than individuals' perceptions of the profession.

In a US project "The Next Generation of Teachers", based on longitudinal data from 50 novice teachers during their first three years in the profession (Johnson & Birkeland, 2003; Johnson et al., 2005; Peske, Liu, Johnson, Kauffman, & Kardos, 2001) the researchers tried to explore the possibility that a new generation of teachers might bring with them new conceptions of career. The findings suggest that new teachers approach teaching more tentatively or conditionally than older generations. Rinke (2013) describes a current notion of teaching as an "exploratory" career in which the teachers view teaching as one of many careers they might have. In spite of this seemingly relaxed attitude to teachers' work many of the leavers and movers expressed the same commitment and dedication as those who envision teaching as a lifelong career. This is also noted by Peske et al. (2001) who point at the potential value for this group of "leavers" and "movers" to contribute to public education. Anderson and Olsen (2006, 2007) have, from results of a one-year longitudinal study, also highlighted the potential of the leavers. The "leavers" in their study

were leaving classroom teaching but were not in fact leaving the field of education, and were still strongly committed to urban education. The authors suggest that we should term these teachers "shifters" rather than "leavers" and that we need more inclusive and multiple frames for teacher careers that acknowledge them as shifters. Quartz et al. (2008) reveal that shifting accounts for a significant proportion of teacher attrition and therefore should be added to the landscape of teacher retention research. Also, the fact that younger teachers were much more likely to shift than to leave education entirely points, according to the authors, to teaching careers as a generational matter. In line with Johnson et al. (2005) they discuss whether today's teachers may be entering the profession with career goals that differ from those of previous generations.

Although comparisons between research results — emerging out of different contexts, countries and cultures — should be handled with caution, we can conclude that teacher attrition seems to be an escalating problem — especially in some nations — and that Sweden seems to be one of these. We can see that the bucket is leaking, and by analyzing data from the cohort we aim to shed some light on: What do the holes look like? When do they occur? Is there a flow in and out? Can we detect possibilities to plug the leaks? Studying attrition as a process and presenting images of the phenomenon from different ranges seems to be a potential way of addressing these questions.

3. The present study

For almost 20 years we have been able to follow a group of 87 Swedish teachers. The first 15 years through semi-structured questionnaires, exchanged between them and their former lecturer at teacher education. It is important to note that the teacher educators main motive to gather the information was pure curiosity and a will to "keep in touch" with her former students. The questions asked are therefore not systematically formulated or theoretically informed. After her retirement we inherited the material and continued to gather data once a year through more systematic questions (although adjusted to the longitudinal items) in formal questionnaires. From 2013 we are also doing follow-up interviews with key informants.

The data is unique in that the group consists of all teachers of the entire group that graduated. In a sense then, our group is not only a sample from a population (teacher students in general) but also in fact constitute a population it itself. In that sense, we have no issues with representativeness, since we have in fact studied the entire population, rather than a sample. As such, we will in most cases not make use of inferential statistics (e.g., confidence intervals or p-values), since this is redundant when we talk about this specific population. We will, however, make use of such statistical techniques when we try to generalize our findings. For example, we will compare our group with that of the general data from the entire teaching population in Sweden (Swedish Teacher Register) in order to see if our group is representative of a larger population.

The material is furthermore unique since the percentage of answers is extremely high (Table 1). The respondents have — in most cases — continued to answer the surveys even though they have left their jobs as teachers.

Table 1

Response rates.

Year 1993 1994 1995 1996 1997 1998 2000 2008 2012

Response rate (%) 100 100 100 100 100 99 93 85 83% (95%)

The response rate for year 2012 is a bit deceiving. After receiving the responses on the survey we contacted (by phone) those who had not answered and received in this way additional data on some issues. For example if they were still working as teachers. The actual response rate for specific questions is therefore 95% for 2012. Hence, although there are some missing data in the later years, the central part of the present study (i.e., working as teachers or not) remains intact.

It is important to note that the first five measurements were made at one-year intervals, but that later measurements were made with different, and longer, intervals. Hence, the data is more detailed for the first five years. Statistically, we will interpolate our data when we present our findings. In other words, the data for the years 2001 —2007 that do not have any data points will be based on a linearly interpolated average of the data from 2000 and 2008 measurements. This is important to keep in mind, since such interpolated averages appear much smoother than they really are. For example, a decline in parental leave during 2000—2008 appears to be a smooth linear trend with equally less people being on parental leave for each year. In reality, there is likely to be some variation, or bumps, in the trend through these years, as is evident during the first five years where we have as detailed data. Because of this different level of detail in different years, we will focus most of our in-depth discussion on the first five years, which in previous studies have also been found to be the years with the highest level of movement, and use a more zoomed out approach when looking at the entire 19-year span.

Even though the questionnaires initially were not systematically formulated, the informants have — in all surveys — reported if they work as teachers, where they work and what kind of work they are doing (including non-teaching work). In survey 1—4, 6 and 9 they have also described experiences of and expectations on their work as teachers. Examples of questions asked repeatedly over the years are:

What would you prefer to work with next semester?

Which of your earlier expectations on working as a teacher have been fulfilled?

Which colleague or person has been most important for you in your work?

What have been the major problems in your work, so far?

In later questionnaires some retrospective information about their career trajectories has been gathered. Examples of such questions asked (survey 8—9) are:

In your previous answers your "teacher trajectory" appears (for some long, others short). Can you tell us something about what has influenced its appearance and if you remember any decisive turning points or critical moments?

What spontaneous comment do you have, thinking back on your first fifteen years as a teacher?

The purpose of the different questionnaires can be described as general and the type of questionnaires can be labeled as semi-structured (Cohen et al., 2011). In later surveys a few more structured and closed questionnaire items, generated by answers from previous open questions, has been included.

In order to enable a more comprehensive understanding of teacher attrition we decided to present images of the phenomenon from different ranges. With a mixed method approach we have had the possibility to combine particularity with generality, to make quantitative and qualitative data "mutually illuminating" (Cohen

et al., 2011, p. 24). The mixed design of the study is sequential (Teddlie & Tashakkori, 2006) in which qualitative and quantitative procedures run one after the other, in order to sufficiently answer the research questions. In the first stage of the analysis, parts of the mainly qualitative data have undergone basic qualitative analyses in order to be transposed into quantitative variables (examples are: working as a teacher, in what subjects and grades, movements in and between schools). These variables have been analyzed by means of STATA 12.1 and SPSS 19 and have been the basis for the creation of — what we call — "overall pictures". These can give us a survey of the whole cohort and can be compared to previous statistical findings. In stage two we have tried to illustrate the individual trajectories by plotting quantitative data into 87 diagrams. But the data also allows further analyses. At the third stage we have had the possibility to move beyond the figures and numbers and actually study how each individual, at each occasion, describe their trajectory. In addition to the overall statistical picture we can return to qualitative data and create "close-up pictures".

4. The cohort

The cohort graduated in December 1993, after 3.5 academic years at a university in a small town in the southeast of Sweden. A quarter of them stated that this was their hometown already when they started teacher education and the majority had been recruited from nearby areas, only 15 of the 87 had traveled long distances to become teachers. This is in line with the recruitment of student teachers to other teacher training programs at small colleges in Sweden. Nor do the students' social backgrounds differ significantly from comparable cohorts recruited to small colleges at this time (Bertilsson, Borjesson, & Broady, 2008). Approximately 30% of the cohorts' parents have completed their education after compulsory school (nine years), while about 30% have attended university. 24 of these are — or have been — working as teachers.

The 87 participants consisted of 63 women and 24 men, a slightly higher number of men than in comparable national statistics. At the time of their last year of studies, they were between 22 and 47 years old, with a mean age of 22.6 and a median age of 24. These figures differ from national statistics at that time, showing an average value for beginner teacher students at 27.3 years.

In summary, the cohort can be seen as representative of students in teacher education in Sweden in the late 1900s. The only difference that can be detected statistically is that they were younger than average teacher students when they started their education. However, from an international perspective there are reasons to believe that the characteristics of Swedish teacher students differ slightly from comparable groups. The median age of Swedish students is the highest in Europe (Statistics Sweden, 2013) and the gender distribution appears to be somewhat more even than in, for example, the U.S. (Bureau of Labor Statistics, 2013).

At the time when they attended pre-service teacher education the Swedish version had recently been reformed and was now divided into two main tracks: "the program for early ages" (grades 1—7) and "the program for older ages" (grades 4—9). All of our respondents attended the program for early ages and were, after graduation, certificated to teach in primary and secondary school. But not in all subjects. The program was divided into two possible areas, language/social sciences or math/natural sciences, which restricted their certification. 78% of the cohort had a study orientation towards social sciences, whereas the rest of them had towards natural sciences. However, this was not equally distributed for women and men, with 84% of the women having an orientation towards language/social science and 63% of the men. This difference was significant, c2 = (1, n = 87) = 4.762, p < .05.

5. The context

The cohort entered a school in considerable transformation. The Swedish educational system was, as in many other western countries, increasingly decentralized and deregulated during the 1990s. In 1991, the central government handed over the responsibility for compulsory, upper secondary and adult education to municipalities and local authorities were given more autonomy in the resource allocation decisions.

The possibility to start up (and for parents to choose) independent schools was introduced during the 1990s. The former "neighborhood principle", in which allocation to a school was fixed depending on the pupil's place of residence, was 1992 replaced by a reform that manifested freedom of school choice. Schools now became exposed to competition and economical dependent on how many pupils they could attract (Carlgren & Klette, 2008; Parding, Abrahamsson, & Berg-Jansson, 2012).

A new employment and labor-law agreement traded the former salary system of collective bargaining and years of working against an individual salary system. Furthermore, the former centrally governed hours of actual teaching was now replaced by an obligation for all full-time teachers to be present at school 35 h per week. Teacher collaboration in cross-disciplinary teams was imposed. The new teachers, each equipped with relatively narrow competencies (language/social or math/natural sciences) from their teacher education, were supposed to be a part of such teams. In reality they entered school organized according to more traditional principles where junior-level teachers taught all subjects in their "own" classes, ages 7—9, intermediate-level teachers did the same in ages 10—12 while senior-level teachers were teaching their "own" subjects in ages 13—15. In our data we can see that 80% of the cohort ended up teaching at junior and intermediate level while 20% were working at the senior level. It appears to be a weak relationship between type of education and what you work with as a teacher. Indeed, most ended up teaching in both language/social science and math/natural science.

6. Results

6.1. The overall picture

Firstly, we look at whether this group of students actually came to work as teachers during this five-year follow-up period. As can be seen in the figure below, 85% had started to work as teachers already during their first year after graduation. This increases to 94% after two years. However, after this we find a negative trend and at year five only 72% of the students in the cohort report that they are active teachers. When we look at this for men and women separately, we see that men tend to work as teachers more often then women in these first five years.

We suspected that both these effects might, in fact, be due to parental leave. (In Sweden you can receive compensation to stay home from work to take care of your child for a total of480 days per child, 390 of these with 80% of your income. Hence, parental leave in Sweden often implies leaving school for an entire school year, and is thus always noted as attrition in yearly statistical measurements. The parental benefit can be received until your child finishes the first year of compulsory school) This would mean that it looks as if the participants become less active as teachers over time because they are more likely to have children over the course of these five years. This might also explain the difference between men and women. Knowing this, Fig. 1 has to be updated. If we separate teachers who are on parental leave from those who are not working as teachers of other reasons the situation looks like this (Fig. 2).


100 90 80 70 60 50 40 30 20 10

'Missing data

Not working as teacher ■Working as teacher

1996 YEAR

Fig. 1. The proportion working as teachers, not working as teachers, or missing, during the first five years.

Indeed, taking parental leave into account gives us a quite different picture, and there is no longer any statistical difference in attrition rates between men and women. Parental leave could of course be described as a form of attrition, but it could also be described as a sort of "further education". Many of the respondents comment on the parental leave as an occasion of truly learning something about being a teacher. Instead of being considered as a drop-out of the bucket, parental leave could, at least in a Scandinavian context, be described as an individual way of enhancing teaching ability (Bjeren & Elgqvist-Saltzman, 1994). Furthermore, in our cohort, parental leave does not seem to be a trigger for leaving the profession permanently. No such correlations can be found. Also, Swedish statistics show that 95% of Swedish women are back in employment after three years of maternal leave (Statistics Sweden, 2007). This number differs from international studies on returning teachers. In the United States, for example, only about 35—40% of the women who leave the workforce after the birth of a child return to teaching (Vera, 2013). This pattern, the author concludes, confirm that the main driving force of American female teachers' decisions to leave for good is not better job opportunities outside the teaching profession but family formation reasons (see also Stinebrickner, 2002). In his study of various reasons for teacher turnover Ingersoll (2003) shows that family reasons are mentioned twice as often as for example job dissatisfaction.

Returning to our cohort we can see that 87% of the cohort remains active at year five, if we include teachers on parental leave in


100 90 80 70 60 50 40 30 20 10

Missing data

■ Not working as teacher

■ Parental leave

■ Working as teacher

1996 YEAR

Fig. 2. The proportion working as teachers, being on parental leave, not working as teachers, or missing, during the first five years.

the number of working teachers. The inclusion of teachers on parental leave in rates of retention seems reasonable, in relation to the Swedish context. The figures differ slightly from national statistics from the same year (1998) that show a retention rate of 82% after five years, but the result differs considerably from U.S. statistics (e.g., Ingersoll, 2003) that estimates retention rates as low as 54% after five years. One-sample t-tests, confirmed that in both cases, our sample is significantly different from these populations, t(84) = 2.3954, p = .018, and t(84) = -2.0840, p = 0.040, respectively.

The attrition in our cohort the first five years cannot be statistically explained by any of the following predictors: age, gender, subject or whether one's parent(s) worked as teacher(s). In fact, when parental leave is considered, the attrition becomes too small to be meaningful to subject to regression analysis, and we have thus not included a regression model or any table of the regression coefficients.

Adding the 19-year perspective on retention and attrition in the cohort the overall picture looks like this (Fig. 3).

As we can see, the overall picture shows that 15 years later (19 years after graduation) 67% of the cohort is still working as teachers (95 % response rate). That is, 58 of our original 87 teachers confirm that they are active as teachers. These figures match results from the few earlier studies on teacher retention over longer periods (Statistics Sweden, 2001). For obvious reasons, none of them are now on parental leave, and the rate of teacher attrition over the whole period appears to be 29%. Looking at the overall picture there is a small, gradual and linear decline in teacher retention (after the peak in year two) over the years — especially the first five. The image that comes to mind is that teachers, one after the other, drop-out of the leaking bucket and stay out. The individual trajectories, however, show a more mixed up picture. When we look at the image from a closer range we can see that the "drop-out" in many cases is temporary. Individuals are not only leaving from, but also returning to, the profession over time.

6.2. Zooming in — the close-up picture

To look at these movements in more detail, we plotted, in 87 individual line charts, whether each individual worked as a teacher or not for each of the first five years. In every diagram a line illustrated a single individual career path. Comparing them, we can see that the most common career trajectory is to start working as a teacher the first year and continuing to do so for each of the five years thereafter (the line is in the upper part of the diagram). This is true for 51 of the participants. Another interesting result is that,

from a closer range, we can see that all 87 did indeed work as teachers at some point during the first five years. This finding differs from other studies (for example Luekens et al., 2004) and indicates that "the bucket" may leak at some points but that some of the lost drops actually find their way back.

Ignoring the ones that are on parental leave within the first five years, 21 individuals in the cohort are — at some point —reporting that they are not working as teachers (the line is zig-zag shaped). What are they actually doing instead?

Let us take a closer look at year one. 13 of the 87 that graduated are not working at all as teachers the first year. One of them had a baby just before graduation and is now on maternal leave. What about remaining 12? Four of them continue to study. In all of the cases they study languages, two of them abroad. The motive is related to the lack of English courses in their teacher training. To study language is a way to meet the demands of their future work. "I think it is necessary to complement my skills with English", one of them says. This seems like a reasonable strategy. When looking at the work assignments the newly graduated got over the first years we can conclude that more than half of them had to teach English although they were not formally qualified for it. Many of them give comments on this situation in their letters. The teacher education that they have just finished does not match the organization in schools. Instead of ending up in teacher teams where their fairly narrow competences, directed towards a small number of subjects, would be complemented by colleagues — they end up alone in front of a whole class and with the responsibility to teach all subjects. What, in the general picture, appears to be a loss of competence could, in these cases, instead be understood as individual initiatives to enhance teacher ability. Three of the four that studies their first year later return to teaching and remain active after 19 years, the fourth later becomes a pre-school teacher. They are all "recycled" into the education system.

Of the eight remaining "leavers" the first year, four individuals try in different ways to find a teaching job but do not succeed. Three of them remain unemployed and the fourth moves to a bigger city combining an office job with consistent efforts to obtain a position as a teacher. Later on they all find a way into the profession and are all still active as teachers after 19 years.

The remaining four continue to work in employments they have had earlier or that have been running parallel to their studies. They do not express explicitly that this is due to difficulties in obtaining teaching jobs. Instead, they seem to cling on to alternative possibilities pondering on whether teaching is the right occupational path. Looking back on his career after 15 years one of


1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Fig. 3. The proportion working as teachers, being on parental leave, not working as teachers, or missing, during the entire range of years studied (1994—2012). The data was collected on nine occasions, see Table 1. Missing years are interpolated.

them reflects: "I was too young. I clearly did not know what I wanted in life and should have had more work experience before I started college". Some of them are looking for teaching jobs, although not very persistently. "I cannot say I've strained myself too much", one of them writes. Interestingly, we find that all of them enter into the profession within the first five years. Two of them are still working in schools after 19 years, one as a teacher and one as a principal.

To summarize attrition the first year, we can conclude that of the 13 individuals that do not work as teachers the first year, only two are lost in the long run.

In Table 2 we have tried to complement the description of attrition the first year with information on what each individual actually did when they were not working as teachers in the first five years. Here we also inserted "part-time" teaching and working as a substitute, that is individuals that had not reported being teachers that year (answered no to the question: Do you work as a teacher?), but still — in answers concerning other items in the questionnaire — described that they were, or had been, working partially in schools. The empty cells indicate that the individual worked as a full-time teacher that specific year. We have also tried to relate the first five years to a "quick scan" over the years 6—19, what happens to the individual trajectories in that period? How much of the attrition over the first five years is retrieved? How many are "leavers", "stayers" and "shifters" at the end of year 19?

It is noteworthy that, as mentioned before, all of the 21 individuals work as teachers at some point during the first five years and that the majority stays within the educational system in the 19-year perspective. But still, in this subset of the cohort the attrition rate over the whole period (the years 1—19) is 42% compared to the attrition rate in the whole cohort which is 29%. Nine of the 21 leave the profession permanently in the 19-year perspective. The majority of them (six) can be described as "early leavers". They drop-

out of the bucket within the first five years and they stay out. Who are they?

6.3. The early leavers

This category includes all the individuals that leave the teaching profession within the first five years and that do not return to teaching (as far as we know after 19 years). They are the seemingly lost ones. In our cohort this comprises six individuals, 8% of the entire cohort. These figures are somewhat lower than comparable statistics in Sweden at that time — 18% — (Swedish Government, 2010) but differ a great deal from Ingersoll's (2003) estimation of as many as 46% leaving the profession within five years (although he acknowledge that the estimation do not account for those who later re-enter teaching).

What are the distinguishing features of our group? One of the early leavers differs from the others. He is far older than the rest when he starts teacher education, gets sick year four, acquires a disability pension and leaves the profession for good. The majority of the early leavers is notably younger, however. Three of them seem "predestinated" to leave. They express no inclination to seriously enter into the profession. Their choice of education seems to be a gamble. When asked at the end of their education "What are your expectations on your future work as a teacher?" their answers differ from the rest of the group. They write: "I really can't say I have any expectations" or "I can't come up with any at this point".

The three who seem predestinated to leave furthermore express that it really does not matter what kind of placement they get after graduation or that they really do not want class teaching positions. In line with the findings of Wilhelm et al. (2000), who suggest that the leavers have a more negative image of the profession prior to entry, one of our leavers write: "Right now I'm pretty tired of everything associated with education and schools". Two of them

Table 2

Individual trajectories for those who did not work as teachers during all of the five first years.

Individual Year 1 Year 2 Year 3 Year 4 Year 5 Year 6—19

1 Unemployed Stays (still teaching after 19 years)

2 Studying Stays (still teaching after 19 years)

3 Studying Stays (still teaching after 19 years)

4 Unemployed Stays (still teaching after 19 years)

5 Substitute/cashier Stays (still teaching after 19 years)

6 Unemployed Stays (still teaching after 19 years)

7 Nurse-assistant Teaching / shifts (becomes principal)

8 Financial assistant Substitute/financial assistant Studying Studying / stays (still teaching after 19 years)

9 Waitress Waitress/studying Waitress/studying Studying Principal / stays

10 Part-time Part-time Part-time Cashier Teaching / studying / leaves

teacher/cashier teacher/cashier teacher/cashier (becomes a financial assistant)

11 Substitute/financial assistant Substitute/financial assistant Financial assistant Financial assistant Financial assistant Leaves (continues as financial assistant)

12 Switchboard operator Substitute/studying Scriptwriter Webmaster Webmaster Leaves (continues as webmaster)

13 Studying Studying Studying Studying Teaching / studying / teaching / leaves (becomes a social worker)

14 Substitute/au pair Substitute/cashier Working as social worker Social worker Studying / leaves (becomes a lawyer)

15 Traveling/salesman Salesman/studying Waitress/travels Teaching / studying / leaves (becomes a psychologist)

16 Reported sick Disability pension Leaves (disability pension)

17 Studying Studying Studying / shifts (becomes a pre-school teacher)

18 Studying Studying Stays (still teaching after 19 years)

19 Telemarketing/teaching Studying Stays (still teaching after 19 years)

20 Studying Studying Studying / leaves (becomes an opera singer)

21 Working at the airport Airport crew Leaves (continues as airport manager)

express an explicit wish to be employed as substitutes, which is very unusual (only four out of the 87 pronounce a wish to be substitutes when asked before graduation). Their search for teaching positions seems to be laid-back and at the same time they are also looking for, and are open to, alternative routes. Later they enter quite different but highly respected professional occupations. One of them becomes an opera singer, one a lawyer and then "Anna" who becomes a webmaster.

Anna is 24 years old when she graduates from teacher education. Both of her parents are teachers. She thinks that being a teacher might be a future prospect but has no major expectations on the work. She comments on this three years later: ".as a teacher student you should be able to give a good justification to your choice. if not — perhaps you should reconsider. that's what I should have done". Before graduation she expresses a wish to work as a substitute. "I want to move around as much as possible and try as much as I possibly can". She applies for some positions the first year but not in a very dedicated manner and continues to work at a job she has had alongside her education while "thinking about what I should do with my life". After year one she moves to the capital of Sweden. Her main reason is that she wants to attend evening courses in drama at a specific institute. During this time she occasionally works as a substitute at different schools and combines that with studies in journalism. Retrospectively, when asked after 19 years, she remembers that her commitment as a substitute was low at the time: ". I completely lost hope when they commented negatively around me staying with the pupils at the recesses — following an exciting drama — instead of hanging around in the teachers' staff room".

In the autumn 1995 she finds a job as a scriptwriter. "This is great fun" she writes "I want to write!" After that she has been working as a webmaster for the entire period, responsible for four Nordic sites targeted at children. She claims having great use of her teacher education in her work, which consists of writing pedagogic and informative articles for the website but also of updating, creating and developing social media. When trying to express why she left teaching she answers "I want to be happy, because then I think I can bring happiness to others. Unfortunately I cannot do that as a teacher". In her mind a happy teacher is someone that can focus on teaching. She does not want to have the "responsibility over each student as an individual", just over the subject matter she wants to teach. When asked year four after graduation she outlines the perfect work as a teacher: "Then I would work over a fairly large area consisting of many schools, but I would just like to teach the subjects that I'm really interested in. Do you think that is egoistic?"

The critical moments occurred long before teacher education, she writes. She never really felt that she could be satisfied being a teacher. But still, she expresses a great deal of dedication in teaching (cf. Peske et al., 2001; Rinke, 2013).

I still feel passionate about the educational community. If I have had the luck to end up at a school with committed teachers, where I could have developed my ideas instead of being dejected and pushed down in a sofa with my cup of coffee.

In a follow-up interview in 2013, after 20 years, she says:

The great difficulties I experienced in my short career as a teacher were that it was conservative and hard to get other teachers to listen to new ideas. I also had some problems with gaining respect from the children. I guess that I thought that everyone should respect me instantly and felt very disappointed when they did not.

When asked if she ever could reconsider returning to teaching she answers: "Yes, if the conditions were right regarding the position, the payment and the whole package".

The remaining two early leavers, a man and a woman have two features in common; they both express a strong will to become teachers before graduation and they both have alternative work alongside their short career as teachers (one of them in a family business, the other at the local airport). Eventually these occupational paths appear as more attractive than teaching. When asked after 20 years if they would reconsider their choice to leave, one of them answers positively. What could bring him back to school then? He answers:

Interesting question! The payment is actually not priority one! The thing that could convince me to return is if focus in teachers' work was more directed towards the actual teaching, the lessons. More time spent there and less on administration — then I would consider working as a teacher again.

We can conclude that — when we look at the leaking bucket from a close-up range — perhaps two out of six holes would have been possible to patch.

7. Discussion

So what happens when we supplement general overviews on teacher attrition with close-up images? First of all, the results from the study can be transformed into a recommendation that we should be careful when we interpret and make use of general statistics. These figures are necessary and useful and they are all "true" in one sense. But how we should understand and explain them must vary. Are they on a one-shot basis? Do they include parental leave or individuals studying? Do they account for the "re-turners", those who leave but later re-enter teaching? There are reasons to believe that the large differences in outcomes of research on teacher attrition are due to such circumstances. In our cohort the attrition rate varies from 29% (according to general overview) to 8% (if we look at the individual and actual outcome in the 19-year perspective). Depending on how we set the focus, the image change. There are also reasons to believe that we should be cautious when comparing statistics between different contexts, countries, cultures and time-periods. Parental leave is, for example, a trigger for attrition in some countries, in others it is not. Consequently, using randomly selected international statistics in order to create "crisis scenarios" or to address national problems with teacher attrition seem to be a bad idea.

Bringing qualitative data from a longitudinal study in relation to statistics certainly gives us a more nuanced understanding of the dynamics in teachers' career trajectories. Using Blumer's (1969) words this kind of data has an "obdurate" character; it "talks back" to the overviews and assertions that statistical data conjures. In doing that, it also stands in the way of the search for causal relations. As we can see the early leavers consist of a small and heterogenous group of individuals. When comparing their motives for entering the teacher education with the rest of the cohort, we cannot find any major differences and there is little evidence that a feeble will to enter teaching expressed just before graduation has any effect on short time attrition in our cohort (cf. Hammerness, 2008, Wilhelm et al., 2000). It is true that three of the early leavers have vague expectations on their choice of work, but that is also the case among some of the stayers.

The same goes for the correlation between organizational factors within schools and the act of leaving teaching early. If we take a closer look at the six early leavers, in the eyes of a policy maker searching for casual relations, one can conclude that perhaps two of

the six "holes in the bucket" could have been patched if the individuals had encountered a more positive work environment or have had better working conditions. The other four seem to have better alternatives or made themselves open to alternative routes. Surprisingly, the considerable changes in educational policy during the 1990s (see pp. 11—12) do not seem to have great impact on the decision to leave. It is well known that the transmission of reforms from the policy arena to the educational context is lagging behind (Lundgren, 1974). The real impact of the 1990 reforms in teachers' working life does not seem to appear until the beginning of the 21st century. In our survey of 2012, 45% of the "stayers" in the cohort claim that they at least once seriously considered leaving the profession. When asked about their motives the answers now reflect the impact of the reforms. Work overload, increased documentation and the notion of altered professional objectives are now often mentioned as triggers for considering leaving. These arguments are also clearly reflected in the data from the late leavers. Such alarming indicators do of course deserve attention (and further studies) but we must also keep in mind that they, in this case, are retrospective and therefore constructed in a rational logic.

The analysis of the longitudinal non-retrospective data from the early leavers makes us wonder, however, if career decisions are as rational as we often seem to suppose. Rationality does not recognize that people create and re-create goals in the course of time or that individuals do not have one stable goal but often operate with a variety of possible outcomes. Having a feasible way out, a parallel occupation, may of course influence the decision to leave teaching. Nor does rational explanations take into account the fact that people's social relations are part of the decision-making process and do not only constitute the conditions and means for an autonomous subject to play "own" choices against. One of the early leavers is, already before she enters teacher education, involved in the family businesses. The social commitment that this entails seems hard to avoid. Earning a degree that entitles you to work as a teacher is perhaps not enough to break the social bonds. We must also keep in mind that leaving or staying is not always a decision dependent on the individual will or her social relations. Economic conditions do of course play a significant role. The generous economic terms in the Swedish insurance system (regarding parental leave, etc.) and the fact that leave of absence does not (officially) lead to any form of "penalties" when/if someone returns to work could be one of the underlying contextual variables that affects decisions to stay or leave.

The heterogeneity in the group of leavers and the multifaceted image of attrition during the first five years provide reasons to discuss whether the idea of rational action ignores the role of serendipity in people's career decision-making (Hallqvist, 2012). When we look for information in relation to the close-up pictures we can see that there, to some degree, seem to be unpredictable chance factors that contribute to vocational choices. But is it just a matter of random chance? Perhaps some of the leaving teachers have certain dispositions that give them an exploratory attitude (Rinke, 2013). They are not just passively relying on luck; instead they seem to remain open to new and unexpected opportunities. They seem to have an attitude of "planned happenstance" (Mitchell, Levin, & Krumboltz, 1999). Hodkinson and Sparkes (1997) argue that career decisions can only be understood in terms of life histories of those who make them. Non-planned decisions responding to happenstance are always made within individuals' horizons for action and depending on the habitus of the individual. When Anna points out that the critical moments which influenced her career took place way before she entered teacher education, she gives us a hint that what seems to be a career by chance might as well be understood as influenced by an "identity generated through interaction with significant others within the culture the subject has

lived and is living" (Hodkinson & Sparkes, 1997, p. 33). So far these kinds of arguments can only be seen as speculations with loose couplings to existing data. The next step in the project will be to gather and analyze the life stories of certain key informants in the cohort, in order to obtain accurate data for such analyses.

Finally, is there reason to believe that we should abandon the image of teaching as a long-term career, and look upon it as a temporary profession? Are we, in a sense, returning to Lorties' (1977) classical description of the teaching profession as a low paid, temporary job for young women prior to their real career (of child rearing)? Some researchers question if the term "career trajectory" is relevant in the study of teachers' work lives and that we need to re-conceptualize the term "trajectory" in such a way that it reflects the complexity of contemporary career patterns. We agree that the very concept of trajectory signals certain linearity, a pathway of a unified moment in one particular direction. But in the light of our data we must argue that there are reasons for maintaining "old" concepts. 51 of our teachers have straight-line trajectories. They started to work as teachers and are still, 19 years later, working full time. However, there seems to be a division among the teachers of those who view and live teaching as a long-term profession and those who see and live it in a more exploratory manner.

The majority of the teachers in our cohort belong to Generation X, born 40—45 years ago. Perhaps the view of a career as a form of "professional exploration" (Rinke, 2013) will be a more common feature among teachers in the next generation, Generation Y? Results from studies indicate that newer generations view their working lives as series of multidimensional careers instead of one linear and continuous career (Dwyer & Wyn, 2001; Stone-Johnson, 2011). We need results from studies on new generations of teachers and by starting a new longitudinal cohort of newly graduated teachers the upcoming new-year, we hope to contribute.

An unavoidable question for research on teacher attrition is whether it is possible, or even desirable, to "patch all the holes in the bucket"? May be it is the "right" people that are leaving? Perhaps a certain amount of attrition can be seen as healthy and potentially beneficial for the profession? To a certain degree this could be true, but some research points in another direction. There is evidence that schools tend to lose the more able than their less able teachers (Ronfelt et al., 2013) especially in poorly performing schools. Regardless of all objections above we must not be distracted from the larger, important point: teacher attrition and turnover are serious problems that can be productively addressed. This quest is "much more important than quibbling over the precise national rate and how it was calculated" (Di Carlo, 2011, p. 3).


Anderson, L., & Olsen, B. (2006). Investigating early career urban teachers' perspectives on and experiences in professional development. Journal of Teacher Education, 57, 359—377. Anderson, L., & Olsen, B. (2007). Courses of action: a qualitative investigation into urban teacher retention and career development. Urban Education, 42, 5—29.

Bertilsson, E., Borjesson, M., & Broady, D. (2008). Lararstudenter, utbildningsmeriter och social bakgrund, 1977—2007 [[Teacher students, educational qualifications and social background, 1997—2007]]. Sociology of Education and Culture Research Reports, No. 45. Uppsala University. Bjeren, G., & Elgqvist-Saltzman, I. (1994). Gender and education in a life perspective.

Lessons from Scandinavia. Avebury. Blumer, H. (1969). Symbolic interactionism: Perspective and method. Englewood

Cliffs, NJ: Prentice Hall International. 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), 367—409.

Bureau of Labor Statistics. (2013). Current Population Study 2011. Carlgren, I., & Klette, K. (2008). Reconstructions of Nordic teachers: reform policies and teachers' work during the 1990s. Scandinavian Journal of Educational Research, 52(2), 117—133.

Cohen, L., Manion, L., & Morrison, K. (2011). Research methods in education. New York: Routledge.

Cooper, J. M., & Alvarado, A. (2006). Preparation, recruitment and retention of teachers. In HEP education policy series No. 5. UNESCO.

Di Carlo, M. (2011). Do half of new teachers leave the profession within five years? Shanker Blog.

Dwyer, P., & Wyn, J. (2001). Youth, education and risk: Facing the future. New York: Routledge Falmer.

Hallqvist, A. (2012). Work transitions as biographical learning: Exploring the dynamics of job loss. Diss. Linköping (1st ed.). Linköping: Linköpings universitet.

Hammerness, K. (2008). "If you don't know where you are going, any path will do": the role of teachers' visions in teacher' career paths. The New Educator, 4(1), 1 —22.

Hasselhorn, H. M., Muller, B. H., & Tackenberg, P. (2005). NEXT Scientific Report, July 2005. The European NEXT-study. University of Wuppertal.

Hodkinson, P., & Sparkes, A. C. (1997). Careership: a sociological theory of career decision making. British Journal of Sociology of Education, 18(1), 29—44.

Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: an organizational analysis. American Educational Research Journal, 38(3), 499—534.

Ingersoll, R. M. (2003). Is there really a teacher shortage? Center for the Study of Teaching and Policy, University of Washington.

Ingersoll, R. M. (2007). Misdiagnosing the teacher quality problem. (CPRE Policy Briefs No. RB-49), Consortium for policy research in education. University of Pennsylvania.

Ingersoll, R. M., & Merrill, L. (2012). Seven trends. The transformation of the teaching force. In Paper presented at the annual meeting of the American Educational Research Association. Vancouver, Canada.

Johnson, S. M., & Birkeland, S. E. (2003). Pursuing a "Sense of Success": new teachers explain their career decisions. American Educational Research Journal, 40(3), 581—617.

Johnson, S. M., Berg, J. H., & Donaldson, M. (2005). Who stays in teaching and why: A review of the literature on teacher retention. Washington, DC: NRTA.

Kyriacou, C., Kunc, R., Stephens, P., & Hultgren, Ä. (2003). Student teachers' expectations of teaching as a career in England and Norway. Educational Review, 55(3), 255—263.

Lortie, D. C. (1977). Schoolteacher: A sociological study. Chicago: Univ. of Chicago Press.

Luekens, M. T., Lyter, D. M., & Fox, E. E. (2004). Teacher attrition and mobility: Results from the teacher follow-up survey, 2000—2001. Washington, DC: U.S. Department of Education, National Center for Education Statistics.

Lundgren, U. P. (1974). Frame factors and the teaching process. In Göteborg studies in educational sciences (Vol. 8). Stockholm: Almqvist & Wiksell.

Mitchell, K. E., Levin, A. S., & Krumboltz, J. D. (1999). Planned happenstance: constructing unexpected career opportunities. Journal of Counseling and Development, 77,115—124.

OECD. (2005). Teachers matter: Attracting, developing and retaining effective teachers. Paris: OECD Publishing.

Parding, C., Abrahamsson, L., & Berg-Jansson, A. (2012). New conditions for identities, cultures and governance of welfare sector professionals: the teaching profession. Theory and Politics in Organization, 12(3), 294—308.

Peske, H. G., Liu, E., Johnson, S. M., Kauffman, D., & Kardos, S. M. (2001). The next generation of teachers: changing conceptions of a career in teaching. The Phi Delta Kappan, 83(4), 304—311.

Quartz, K. H., Thomas, A., Anderson, L., Masyn, K., Lyons, K. B., & Olsen, B. (2008). Careers in motion: a longitudinal retention study of role changing among early-career urban educators. Teachers College Record, 110, 218—250.

Rinke, C. R. (April 2013). The career pathways of teachers who leave: exploring the transition out of education. In Paper presented at the annual meeting of the American Educational Research Association. San Francisco, CA.

Ronfelt, M., Loeb, S., & Wyckoff, J. (2013). How teacher turnover harms student achievement. American Educational Research Journal, 50(1), 4—36.

Schaefer, L., Long, J. K., & Clandinin, D. J. (2012). Questioning the research on early career teacher attrition and retention. Alberta Journal of Educational Research, 58(1), 106—121.

Statistics Sweden. (2001). Avhopp fran lararyrket och framtida pensionsavgang [[Dropout from teacher profession and future retirement]] (Report 2001:2). Statistics Sweden.

Statistics Sweden. (2005). Lararforsorjningen 2004 (Statistiska meddelanden UF 26 SM 0501). In [The provision of teachers 2004]. Statistics Sweden.

Statistics Sweden. (2007). Foraldraledighet och arbetslivskarriar. En studie av mam-mors olika vagar i arbetslivet [[Parental leave and work life careers. A study on mothers trajectories in work life]]. Statistics Sweden.

Statistics Sweden. (2012). Trender och Prognoser 2011 [[Trends and forecasts 2011]]. Statistics Sweden.

Statistics Sweden. (2013). Statistik och utvardering, grundskolan [[Statistics and evaluation, compulsory school]]. Statistics Sweden.

Stinebrickner, T. R. (2002). An analysis of occupational change and departure from the laborforce: evidence of the reasons that teachers leave. Journal of Human Resources, 37(1), 192—216.

Stone-Johnson, C. (2011). Talkin' bout my generation: boomers, Xers and educational change. Journal of Educational Change, 12, 221—239.

Svenska Dagbladet. (2010, February 11). Tuffare krav for lararutbildning [[Tougher demands on teacher education]]. Svenska Dagbladet. Retrieved from http://

Swedish Government. (2010). Tillgangen pa behoriga larare [[Supply of certified teachers]] (Report 2010:7). Utredningstjansten.

Swedish National Agency for Higher Education. (2012). Hogskoleutbildningarna och arbetsmarknaden [[Higher education and the labor market]] (Report 2012:22R). Swedish National Agency for Higher Education.

Teddlie, C., & Tashakkori. (2006). A general typology of research designs featuring mixed methods. Research in the Schools, 13(1), 12—28.

UNESCO Institute for Statistics. (2009). Projecting the global demand for teachers: Meeting the goal of universal primary education by 2015. Technical Paper No. 3. UNESCO Institute for Statistics.

Vera, C. P. (2013). Career mobility patterns of public school teachers. Munich Personal RePEc Archive.

Wilhelm, K., Dewhurst-Savellis, J., & Parker, G. (2000). Teacher stress? An analysis of why teachers leave and why they stay. Teachers and Teaching: Theory and Practice, 6(3), 291—304.