A Tale of Two Autonomies

Despite comprehensive theoretical discussions on the nuances of autonomy, research tends to treat autonomy as a unidimensional concept. In contrast, this study of Norwegian welfare professionals presents empirical support for the multidimensional nature of autonomy, drawing on cross-sectional survey data from three datasets spanning six years. The findings show significant differences between welfare professionals’ experiences of professional and personal autonomy. An analysis of the relationship between professionals’ experience of performance demands and these two types of autonomy challenges the notion that increasing performance demands limits professional autonomy.

The struggle between professional knowledge and managerial control continues to be a topic of discussion among professionals and academics alike.Central to this is the suggestion that professional workers are losing autonomy to increasing demands and controls on production and efficiency (Evans, 2009;Harris, 2002;Lawler & Bilson, 2009), a development attributed with potentially grave consequences for the professional role (Evans, 2009;Harris, 2002;Lawler & Bilson, 2009;Lymbery, 2012).An empirical revision of the dimensions of autonomy is necessary because a chasm seems to exist between the classical theoretical understanding of autonomy and empirical research.Theory tends to present autonomy as a multidimensional concept, while empirical efforts often treat it as unidimensional, creating confusion about what we discuss, the individual dimension of autonomy or the concept as a whole?
My findings inform the general debate on autonomy by mending the chasm between theory and observation regarding the multidimensionality of autonomy.Using unique cross-sectional data on welfare professionals in the period 2004-2010, I apply theory and principal component analysis to establish experiences of autonomy as a concept with personal and professional dimensions.Specific to professionals, this article explores the influence of perceived performance demands on welfare professionals' experiences of personal and professional autonomy.This focus is particularly relevant in light of research describing a loss of professional autonomy due to increased demands on performance and managerial control.I wish to challenge this story.
 to combine theory and principal component analysis to establish measurements for empirically demonstrating that performance demands, professional autonomy, and personal autonomy are experienced as different concepts, a prerequisite step for exploring research aims two and three;  to compare welfare professionals' experiences of performance demands, professional autonomy, and personal autonomy;  to explore the relationship between performance demands and experiences of professional and personal autonomy.

Professional workers and autonomy
"Professional worker" describes someone working in a profession.Defined by Brante as "occupations conducting interventions derived from scientific knowledge of mechanisms, structures, and contexts" (2011, p. 17).Professions are considered more or less professional based on characteristics such as education, social status, prestige, and historical origin.This scale of professionalism has led to the use of subcategorizations such as "semi-" or "pre-professions" (Brante, Johnsson, Olofsson, & Svensson, 2015).
Welfare professions discussed in this study include social workers, child-welfare workers, teachers, and nurses who constitute a cohort by their connection to the social welfare system.They are treated as similar to "semi-professions" (Brante et al., 2015), but I label them welfare professionals to emphasize the nature of their work, rather than their place in a professional hierarchy.Welfare professions may have a more fragmented and disputed knowledge base compared with classical professions, yet act with considerable autonomy (Brante et al., 2015), and are in the Scandinavian context often employed in the public sector.
Autonomy is surprisingly hard to define, often presented with diverging characteristics.In an analysis of the term, Ballou (1998) finds autonomy to be about acting within a set of rules with ability, capacity, competence, decision-making, critical reflection, freedom, and self-control.Others see it as an attitude rather than a characteristic of the work situation (Hall, 1968), or as defined through occupational behavior (Schutzenhofer, 1987).Autonomy is neither good, bad, nor absolute; it is contingent on the situation, is a gradual scale rather than dichotomous, and can be both a source of strength and a hindrance for professionals (Evans & Harris, 2004).Autonomy is used to describe professional freedom at the micro-, mezzo-, and macro-level (Gross, Tabenkin, & Brammli-Greenberg, 2007).This article examines autonomy on the micro-level, looking at welfare professionals' experienced freedom of decision-making, used interchangeably with discretion.
Autonomy is of particular interest to professional workers because they enjoy a protected status in exchange for using their autonomous skills to serve society (Engel, 1970;Freidson, 2001;Parsons, 1991).This is regulated by a "professional contract" which sets the relationship between control and freedom for professionals.An arrangement where a governing body provides education and protection of the professional status in exchange for professionals using their abilities for the betterment of society (Brante et al., 2015).Although Mastekaasa (2011) suggests that autonomy is not particularly important to the professional worker, it remains a topic of interest in the discourse on the professional role and discussions on control and freedom for professional workers (Brante et al., 2015;Øverbye, 2013).

The two autonomies
Among numerous dimensions of autonomy, this article concerns the distinction between personal and work-related autonomy (Engel, 1970).Personal autonomy is the "freedom to conduct tangential work activities in a normative manner in accordance with one's own discretion" (Engel, 1970, p. 12), while work-related autonomy is "freedom to practice his profession in accordance with his training" (Engel, 1970, p. 12).These dimensions capture two central aspects of autonomy: on the one hand, personal autonomy over tangential work activities such as organizing and scheduling, and on the other, work-related autonomy in applying professional practices and training.Although they are likely to influence each other to some degree, they are understood as inherently different dimensions of autonomy (Engel, 1970).
Research aimed at describing experiences of autonomy tend to lump these two aspects together (Ballou, 1998;Hackman & Oldham, 1975;Mastekaasa, 2011), creating hybrids such as "the degree to which the job provides substantial freedom, independence, and discretion to the employee in scheduling the work and determining the procedures to be used in carrying it out" (Hackman &Oldham, 1975, p. 162: andapplied similarly in Mastekaasa, 2011).An understanding that merges Engel's (1970) dimensions of personal autonomy (controlling tangential aspects of work) and work-related autonomy (the process of determining procedures).Brante et al. (2015) present multiple dimensions of autonomy, including ones measuring the professional's control over tangential elements such as when to work, at what tempo and order, and how to act towards clients.This understanding follows the spirit of Engel's two autonomies in separating between tangential tasks such as when to work, what tempo, and professional autonomy in how to act towards clients.It differs from Engel's in that they define professional autonomy as the freedom to perform the work and procedures with no explicit connection to training.This understanding is similar to Hackman and Oldham's (1975).I adopt this understanding because of the close connection between professional training and the decisions professionals make (Bøe, 2013), and that being a "professional" is not necessarily limited to training or education, but can also be considered contingent on the type of work performed (Løwendahl, 2016).
The data in this article comes from a battery of questions measuring "job demands" and "job-decision latitude," or the "discretion permitted the worker in deciding how to meet these demands" (Karasek Jr, 1979, p. 285)."Job-decision latitude" is similar to Engel's (1970) work-related autonomy by being concerned with professional decision-making.However, it differs in that it does not include an explicit link between decisions about the professional content of their jobs and their training.Making Karasek (1979) in this regard more similar to Hackman and Oldham (1975), Brante et al. (2015), andLøwendahl (2016) than Engel (1970).In this article, I refer to this kind of professional decision-making as "professional autonomy.""Job demands" include both workers' experiences of having demands on their performance and of being in control over their work situation such as time and workload (Karasek Jr, 1979).The former measuring demands on how hard and fast they work, the latter measuring personal autonomy; being in control of tangential elements of work such as one's work situation and schedule (Engel, 1970;Hackman & Oldham, 1975;Mastekaasa, 2011).Supported by a principal component analysis, I propose that Karasek's (1979) category of job demands should be refined into two different concepts: (1) personal autonomy or the degree of control professionals experience over their work situation excluding professional decisions, and (2) performance demands, or the professionals' experienced demands on their performance.
This refinement allows me to treat autonomy as a concept with multiple dimensions of professional and personal rather than a singularity.Allowing for a better understanding of professional workers' experiences and how the various dimensions of autonomy may be influenced by performance demands.

Managerialism and autonomy
Welfare professionals' experiences of autonomy and performance demands are particularly relevant due to the popular notion that new public management (NPM) reforms-part of broader neoliberal governance also in Scandinavia (Peck, 2010)have weakened professional autonomy through increased control and demands on performance (Brante et al., 2015;Harris, 2002).This "managerialism," a prevalent aspect of NPM seen in social work in the United Kingdom, represents a trend towards increased managerial control on production (Evans, 2009, p. 146).Research from the Scandinavian welfare context points to similar issues of balancing organizational control and professional autonomy among social workers and nurses (Braend, 2014;Dalsgaard & Jørgensen, 2016;Hildebrandt, 2016;Shanks, 2016).Such developments have also been found across the spectrum of welfare professions in Denmark (Willig, 2016).
Managerialism is characterized by an increased business orientation among welfare professional organizations, with greater emphasis on finances, budgeting, and control (Harris, 1998).In Scandinavia, this means an increased use of performance control measures, seen in conjunction with a loss of trust in professionals and an organizational need for control to compensate for this (Øverbye, 2013).Some research has pointed out that a skewing of the balance between management and professional practices could lead to restrictions on professional autonomy and the prioritizing of managerial control (Harris, 2002;Lawler & Bilson, 2009).For example, Danish and Norwegian nurses have experienced a disconnect between performing tasks they are trained to do as caregivers and meeting specific production measures such as turnover rates in hospitals (Braend, 2014;Thomassen & Larsen, 2016).Managerialism and NPM are broad, general descriptions of trends not easily measurable; however, both theoretical and empirical contributions suggest that an emphasis on performance and ensuring production are central elements.Therefore, I treat welfare professionals' experiences of performance demands as an indicator of managerialism.
As for the consequences of managerialism on the professional role, Evans ( 2009) argue that increased managerialism can complement rather than limit it.This because social work managers in the United Kingdom report adhering to professional values despite being tasked increasingly with production and performance goals.This then becomes an additional part of the manager role, working alongside and not against professional autonomy.Added responsibilities, such as budgeting, allows local human service managers to operate flexibly within the system and providing maneuverability in gray areas (Evans, 2009, p. 153-156).With the result that professionals report that their managers give them the discretion to make sound professional decisions despite increased managerialism (p.157).
These complementary functions are possible because managers see themselves as performing independent professional and managerial roles (Lawler & Hearn, 1997).Derber's (1983) literature review found that professionals are controlled differently in how they work and what they produce, suggesting that professional and personal autonomy are two separate dimensions which may co-exist (p.335).A literature review by Kirkpatrick (2006) shows that increased control over social work practices in the United Kingdom manifests itself as an increase of rules, procedures, and standardization, without these changes influencing professional values and practice to the degree one might suspect.This is also found in the Scandinavian welfare system as Shanks ( 2016) suggests that Swedish welfare service managers keep a strong professional identity regardless of increased focus on cost effectiveness.The ability to maintain their professional identities in the midst of neoliberal reforms suggests that increasing demands may not necessarily limit the "controller's" professional role, nor the professional autonomy among the "controlled."As professionalism and control coexist in "professional bureaucracies," professionals may lose some control over the means of their work (personal autonomy) but retain power over the ends (professional autonomy) (Engel, 1970;Harris, 1998).These theories explain how professional and personal autonomy can coexist and support each other as separate dimensions of autonomy; individuals need not possess one or the other exclusively but can possess both to different degrees.

Hypotheses
Drawing on research on autonomy and managerialism, I present two hypothesis supporting my second research aim: to compare experiences of performance demands, professional autonomy, and personal autonomy among welfare professionals three years removed from graduation in the period 2004-2010.
The first hypothesis tests whether the two proposed dimensions of autonomy are experienced as different dimensions by the professionals.I use data from three samples to test if these are recurring differences.The second hypothesis tests how experiences of performance demands developed over the period for welfare professionals three years removed from graduation as a cohort.
1. Welfare professionals experienced personal autonomy and professional autonomy as different dimensions of autonomy.2. Welfare professionals as a cohort experienced increasing performance demands in the period.From the disagreement on the relationship between managerialism and professional identity, I present two hypotheses to support my third research aim: to explore the relationship between performance demands and experiences of professional and personal autonomy.These hypotheses test to what degree managerialism limits professional and personal autonomy.
1. Experiences of performance demands have a negative correlation with experiences of professional autonomy among welfare professionals.2. Experiences of performance demands have a negative correlation with experiences of personal autonomy among welfare professionals.

Methodology
These hypotheses were tested with data from StudData, a Norwegian database of students and former students from professional programs at several universities and colleges in Norway.The samples include former students who were three years beyond graduation in 2004, 2006, and 2010, having completed welfare professional programs at the institutions.This design provided information about how professionals in the same situation, but at three different times experienced performance demands and personal and professional autonomy.Data from three different graduating classes allows for tracking the stability of these experiences with three samples, and the development of the cohort over time.Participation was voluntary, and respondents could refuse to continue participation at any time.The Norwegian Data Inspectorate approved the survey.The response rate of the database in 2004, 2006, and 2010 was 56%, 58%, and 37% respectively.Respondents not currently employed were removed from the sample.Some questions were missing answers for some respondents.The number of welfare professionals (nurses, teachers, preschool teachers, social workers, and child-welfare workers) in the responding cohort varied between 850 and 1025.At three years after graduation, respondents in the selection had a mean age of 30.4,30.9, and 30.1 respectively.Eighty-five percent of respondents were women in 2004, 86% in 2006, and 78% in 2010, which corresponds to the higher number of women admitted to welfare professional programs.

Variables
I combined eight self-reported statements concerning work characteristics into three merged variables measuring welfare professionals' self-reported experiences of performance demands and personal and professional autonomy.I devised the variables from Karasek's (1979) model for measuring job demands and job-decision latitude, which asks, "To what degree would you agree with these statements about your current job?" Karasek's merged variable "job demands" consists of five statements.This variable is refined to measure two merged variables for this study.Two statements concerning how professionals experience demands on how quickly and hard they work are used to measure the variable "performance demands," while the three statements concerning professionals' experience of being able to control their own work situations are used to measure the variable "personal autonomy."The principal component analysis, shown in Table 1, collaborates this separation, providing empirical evidence that Karasek's variable "job demands" actually contains two different variables.Three statements measuring "job-decision latitude" are then used to measure the variable "professional autonomy." "Performance demands" measures how professionals experience demands on how hard fast and hard they work, is the mean of the following two questions: 1) "My work demands that I work very fast" and 2) "My work demands that I work very hard."A high score indicates a high level of experience of performance demands, and relates closely to demands on performance and production, central features of managerialism.
"Personal autonomy" measures how professionals experience their ability to control their own work situations, is the mean of the following questions: 3) "I am not asked to do disproportionate amounts of work"; 4) "I have enough time to get the work done"; and 5) "I am not exposed to conflicting demands from others."A high score indicates a high level of experience of personal autonomy.
"Professional autonomy" measures the freedom professionals experience in making decisions about the content of their work and their discretion to determine procedures (Brante et al., 2015;Hackman & Oldham, 1975;Løwendahl, 2016).As a measure of experienced autonomy in decision-making, this variable does not differentiate between decisions made based on individual preference and those based on professional knowledge.Nevertheless, it is considered professional autonomy because it concerns the say that professionals have in determining the measures to use in their work.Professional training is thus treated as an implicit element in the decisions professionals make.Professional autonomy is measured as the mean of the following questions: 6) "My job makes it possible to make a lot of decisions on my own;" 7) "In my job, I have very little freedom to decide how I do my work;" and 8) "I have a lot of say about what happens in my job."(The answers to question 7 were reversed to create a unified scale.)A high score indicates a high degree of experienced professional autonomy.
I utilized the educational code registered to each participant to determine their professional backgrounds.Previous studies have used different praxes when grouping professionals.For instance, Mastekaasa (2011) included physicians, nurses, teachers, and social workers.Inspired by Brante et al. (2015), and the availability of professional groups in the dataset, I included the following welfare professions in my analysis: nurses, teachers, preschool teachers, social workers, and child-welfare workers.Multiple regression analysis with ordinary least squares (OLS) to control for the effects of professional group, age, and gender showed a non-substantial effect on the relationship between performance demands and the two autonomies.This does not mean that every profession experienced the three variables similarly, but that the relative relationship between the variables is not substantially different between the included professions.Based on this I treated all included professions as one cohort of welfare professions.Subsequently, I present the findings as correlation coefficients for easier interpretation.

Variables' validity
The category of performance demands differs from personal autonomy in that it concerns the experience of demands regarding time and effort at work rather than the experienced control over the work situation.Performance demands increase when professionals are asked or expected to produce at a certain level, while personal autonomy decreases when professionals lack control over their work situation, such as being unable to prioritize tasks or facing conflicting expectations of what do to.Therefore, professionals may experience the expectation to work hard and fast to reach a goal or quota (high performance demands) while also experiencing a high degree of control concerning what to do (high professional autonomy) and when to do it (high personal autonomy), such as when social workers are expected to follow up a certain number of clients, but may choose which clients to follow up when and with what measures.
The validity of "personal autonomy" poses a challenge for the analysis and results.As Table 1 indicates, the statements describing one's ability to control the tangible elements of the work situation are distinct from those describing professional autonomy and performance demands, which raises the question of what to call this variable.My decision to name it "personal autonomy" rather than alternatives such as "workload" was based on established theories about professionals' ability to control their tangential work situation beyond the professional content (Derber, 1983;Engel, 1970;Hackman & Oldham, 1975;Karasek Jr, 1979).The crucial argument for this decision was the strong relationship between the workload elements of question 3 and 4 ("I am not asked to do disproportionate amounts of work" and "I have enough time to get the work done") and autonomy in question 5 ("I am not exposed to conflicting demands from others.")The relationship between these three questions is clearly stronger than with those in performance demands and professional autonomy, which suggests that my variable "personal autonomy" is related to experiences of both workload and autonomy, and not one or the other (Table 1).Therefore, I am confident that it measures participants' experience of being in control of their own workday and such is a fitting measure for personal autonomy.

Analytical approach and methods
The internal consistency of each merged variable was measured by their Cronbach's Alpha score.The principal component analysis shows that each scale was unidimensional (Table 1).A high Cronbach's Alpha score (zero to one) indicates that the merged variables measure the same underlying concept.The values for "personal autonomy" for each of the three years (2004, 2006, and 2010) are 0.65, 0.64, and 0.69 respectively.Values for "professional autonomy" are 0.62, 0.64, and 0.63, and for "performance demands," 0.71, 0.77, and 0.74.One potential weakness of Cronbach's Alpha is that few items, as in this study, tend to score lower than many (Tavakol & Dennick, 2011).As there is no absolute threshold for internal consistency (Cortina, 1993), care must be taken not to use Cronbach's Alpha as the only proof of a merged variable being internally consistent.Although the scores in this study are not very high, the combination of unidimensional variables demonstrated by principal component analysis (Table 1) and the Cronbach's Alpha scores, give me confidence that the three merged variables measured different, unidimensional concepts to a satisfying degree.
I confirmed the refinement of Karasek's "job demands" into "personal autonomy" and "performance demands" with a principal component analysis and a rotated components matrix using Varimax rotation with Kaiser Normalization.Principal components analysis indicates how statements load into different variables, signifying which statements measured the same underlying concept.Statements loading in the same variable are understood to measure the same underlying concept.I performed a preliminary analysis with Direct Oblimin, a rotation-assuming correlation, to confirm the different variables before presenting with Varimax rotation to present the variables clearly.Rotation with Varimax illustrates the differences between variables more clearly by assuming that they are independent of one another.
I used Spearman's rank order to determine the correlation between performance demands and professional and personal autonomy.Results that point to statistical significance are indicated with no stars (p>0.05),two stars ** (p<0.05), and three stars *** (p=0.00).All statistical analyses were performed using IBM SPSS Statistics v.23 for Windows.

Results
The results support that the five statements constituting Karasek's (1979) one variable "job demands" should be separated into the two distinct variables I call "performance demands" and "personal autonomy" (Table 1).The eight statements in Karasek's measurement of "job demands" and "job-decision latitude" indicate that they feed into three separate variables with an Eigenvalue over 0.9, explaining at least 63% of the variance in the eight statements that I call "performance demands," "personal autonomy," and "professional autonomy" (Table 1).Principal component loadings from all three datasets show similar patterns.Even though some elements, such as question four, may appear weakly related to questions one and two, they are substantially different enough to be considered as three distinct concepts in the next level of analysis.Bold numbers indicate statements loading to the presented variable. 1 1 The presented order of the loadings was changed from the output for uniform presentation.

Experiences of autonomy and performance demands
The average scores of welfare professionals' reported experiences in years 2004, 2006, and 2010 show a slight increase for performance demands, and that personal and professional autonomy are experienced as different dimensions in all three samples (Table 2).Note: Responses on a scale from low (1) to high ( 5) experiences of each concept with N and standard deviation.
Concerning hypothesis 1 the results indicates that "personal autonomy" and "professional autonomy" are experienced as different dimensions of autonomy.Comparing the experiences of personal and professional autonomy with paired samples Ttest show that these differences are statistically significant (p=0.00) in all three samples.The recurring nature of these differences suggests that this is a stable relationship.The results also support hypothesis 2 in that experiences of performance demands increased somewhat in the period, specifically: 1.There was a recurring statistically significant difference in the experiences of professional and personal autonomy among welfare professionals.2. There was a slight increase in experienced performance demands for welfare professionals as a cohort during the period.

Relationship between autonomies and performance demands
While the means indicate the stability and development of how these concepts are experienced, correlation analysis is necessary to test hypotheses 3 and 4 regarding how the experiences of one concept influence the experience of the others.Correlation analysis indicates that welfare professionals who experienced increasing performance demands experienced less personal autonomy (Table 3).Experiences of increasing performance demands were not related to experiences of professional autonomy.Results show a moderate negative correlation between experiencing performance demands and experiencing personal autonomy in the study years (-0.341***,-0.414***, and -0.444*** respectively).The correlation analysis indicated no statistically significant relationship between experiences of increasing performance demands and professional autonomy (p>0.05).
The results demonstrate that increasing performance demands influence these dimensions of autonomy differently.While correlation is not causality, the data indicates that experiences of personal autonomy are moderately related to experiences of performance demands, while experiences of professional autonomy are unrelated to experiences of performance demands.
My interpretation of this is that the professional actors experience that they are able to keep their professional decision-making ability despite being of increasing performance demands.This does not reject that increasing performance demands may limit experiences of autonomy as a whole, but it suggests there are substantial differences in the influence of increasing performance demands on the distinct dimensions of autonomy.The difference shown here indicates that the relationship is stronger between performance demands and personal autonomy than performance demands and professional autonomy.These results reject hypothesis 3, and support hypothesis 4, specifically: 3. Experiencing increasing performance demands did not limit experiences of professional autonomy among welfare professionals.4. Experiencing increasing performance demands limited experiences of personal autonomy among welfare professionals Personal autonomy and professional autonomy appear as weak dependent concepts with a statistically significant correlation in all three years (0.137***, 0.112**, and 0.103**), supporting hypothesis 1 suggesting that they are experienced as distinct dimensions.

Main findings
The main findings of this study are as follows: 1. Principal component analysis supports the theoretical separation of Karasek's (1979) two variables into the three variables: performance demands, professional autonomy, and personal autonomy, enabling the empirical testing of this study's hypotheses.2. Welfare professionals three years removed from graduation experienced professional autonomy and personal autonomy as different dimensions.Experiences of performance demands for the cohort slightly increased in the period.3. On an individual level the increasing performance demands seem to have a negative impact on experiences of personal autonomy, but not on professional autonomy.This relationship was recurring over the period.These findings present two major theoretical implications that warrant further discussion: First, autonomy is measurable as a multidimensional concepts, and second, experiencing increasing performance demands might not serve to limit professional autonomy to the degree that some have suggested (Evans, 2009;Harris, 2002;Lawler & Bilson, 2009).

Discussion, implications, and future research
This study's principal component analysis indicates substantial differences between experiences of professional and personal autonomy, suggesting that future research should account for this multidimensionality.The empirical backing for the multidimensional nature of autonomy supports Engel's (1970) theories about there being at least two distinct dimensions of autonomy.It contributes to previous empirical research by empirically demonstrating autonomy as consisting of multiple dimensions rather than a singularity.The multidimensional view of autonomy provides a more accurate description of the influence that different work characteristics, such as performance demands, has on professionals and can increase the precision of empirical research and discussions.

Managerialism and autonomy
The multidimensional aspects of autonomy can explain why there are two schools of thought regarding the effects of managerialism and performance demands and how both can be right in their own way.Managerialism may limit autonomy as a whole, and yet affect the dimensions of personal and professional autonomy differently.
Although these results support the notion that experiences of performance demands tended to increase among welfare professionals in the 2000s, the correlation coefficients indicate that increasing managerialism is not necessarily an obstacle to experiencing professional autonomy.Supplementing previous research, my findings suggest that managerialism may not be as limiting to professional autonomy as has been suggested, but that it may limit welfare professionals' experience of personal autonomy.The takeaway being that professional autonomy-and through it, the professional role-may not be under the same stress as professionals' ability to organize their own work situation.
As welfare professionals experienced having to perform better, their experience of personal autonomy decreased, while professional autonomy remained unchanged.This duality may be explained by theories on the separation between control and professionalism in professional bureaucracies (Engel, 1970;Harris, 1998); that is, professionals may treat freedom to plan and freedom to perform as independent dimensions (Lawler & Hearn, 1997), or that they may be governed differently in the means and ends of their work (Derber, 1983).
An alternative explanation is that limitations on professional autonomy may be experienced as indirect or entangled restrictions.For example, too little time and too many tasks could indirectly hamper professionals' ability to perform their jobs with the desired professional autonomy.However, the weak correlation between professional and personal autonomy suggests that they are largely independent dimensions and that indirect restrictions cannot explain the results alone.Nevertheless, the direct and indirect effects of managerialism on professional autonomy through personal autonomy warrants further research considerations.
Some of this effect could potentially be explained by differing expectations coming into their professional career.Professional workers who enter initial job positions may be conscious of the tension between professional autonomy and managerial control, but have fewer or no expectations of limitations on how to control their workday.Coming recently from an educational situation with a high level of personal autonomy, new professionals may experience a "shock" of having less control over their own workday as employees.The effect of this should be limited by the study measuring experiences three years after graduation.

The professional contract
The "professional contract" regulating the relationship between control and freedom for professionals is one in which a governing body provides education and protection of professional status and in exchange professionals' use their abilities for the betterment of society (Brante et al., 2015).This study's findings suggest that any limitations from this contract are more poignant for professionals' ability to organize their workday, than for their application of professional expertise.Which indicates that the professional contract work as intended: Welfare professionals experience an ability to perform their job as trained professionals, even with increased demands on performance.Rather than losing professional influence over their work, professionals lose control over the organizing of their work.This is compatible with the intention of the professional contract, which describes a trade-off in which professional workers exchange something for their status (Brante et al., 2015;Derber, 1983;Engel, 1970).The professional contract may still be viable under increasing managerialism as welfare professionals experience continued control over their decisionmaking, even as they lose control over the tangible aspects of their workday.A possible explanation for this is that personal autonomy may lack the explicit and historical protection of professional autonomy in the professional contract, making it a more likely casualty of increasing demands for control and production.

Consequences for managers
According to Noordegraaf (2011), conflicts between professional and managerial logics are not solvable by adding more professionalism to "rescue" professional work from control, nor by "moving beyond" professionalism through more restrictions on the work.Rather, Noordegraaf (2011Noordegraaf ( , p. 1362) ) argues that professionals must realize that they need to be managed and thus adapt to changing contexts.My findings suggest that practitioners and managers in welfare professions are adapting to these changes by compartmentalizing personal autonomy and professional autonomy; treating them as two separate dimensions rather than one, allowing for increased performance demands without it limiting professional autonomy.
The mechanism behind this has been previously described among nurses who take individual responsibility for providing the care they deem necessary, independent from organizational goals focusing on performance demands (Thomassen & Larsen, 2016).Serving as an example of how professionals are able to compartmentalize the decisions about what is professionally necessary, and what is organizational preferable.This poses a potential challenge for managers as professionals appear to favor their professional training over managers' instructions.Creating a chasm between organizational needs, defined in their routines and enforced by control, and what the professionals prioritize.When creating measures and instructions concerned with ensuring performance, managers should be aware of how these measures might be individually adapted by the professional to fit their professional understanding, rather than being uniformly applied.This is important in order to avoid goal displacement which risks that the provision of care becomes an individualized effort rather than an organizational goal (Braend, 2014;Thomassen & Larsen, 2016).One way of avoiding this is to involve the professionals in creating performance measures aligning professional considerations with organizational goals and need for control.
Managers should be aware that efforts to increase performance influences experiences of professional and personal autonomy differently, the non-relationship between performance demands and professional autonomy is no argument for managers to demand performance under the veil that professional autonomy appears to be unaffected.Managers must consider that a decrease in experiences of personal autonomy may have negative impact on organizational and personal performance limiting the sense of empowerment and trust necessary for work performance (Dirks & Ferrin, 2002;Harter, Schmidt, & Hayes, 2002) and for work satisfaction (Barak, Travis, Pyun, & Xie, 2009), both of which are important elements in job retention (Barak, Nissly, & Levin, 2001).Although autonomy is neither good, bad, nor absolute (Evans & Harris, 2004), managers and professionals must work together to ensure that their blend of demands and autonomy suit their work situation.It should also be considered that limits on personal autonomy are not necessarily negative, a straightjacket for the professional, in some instances it may benefit professionals who are freed to spend time on professional decisions rather than administrative ones.

Limitations and future research
One limitation of this study is that the data constitute an average from a variety of individual welfare professionals, which loses the nuances of each professional group and individual.Increased individualization (Braend, 2014;Thomassen & Larsen, 2016) could reveal greater variations between welfare professionals, within and across professions.Controlling for the relative effect of different professions reduces the risk for this regarding the different professions.Future case studies are necessary to better gauge these variations on an individual level.Some respondents may have been working in different fields than their educational code indicated.Inherent biases always exist among respondents as they choose to answer surveys; however, both risks are minimized by the three independent datasets, which indicated stable trends in the correlation between performance demands and autonomy.The situation in Norway may be different than in other countries with respect to experiences of performance demands and the corresponding effects on experiences of autonomy.Similar studies should be conducted in countries with a stronger emphasis on performance demands.
Another potential limitation is my redefinition of "job demands" into "performance demands" and "personal autonomy."Although I have argued extensively for this choice, using theory backed by principal component analysis, this redefinition should be developed further in future research to better nuance the differences, preferably by developing question batteries specifically for this purpose.Controlling for professions, age, and gender does not rule out other elements acting as a spurious effect or codependent factor.Finally, Lin (2014) suggests that organizational size might be relevant to experienced autonomy.This study lacked data to control for organizational size; future research should examine the effect of organizational size on experiences of different dimensions of autonomy.

Conclusion
This study empirically establishes the multidimensional nature of experiences of autonomy among professional welfare workers, addressing the chasm between theory and empirical data in this research area.The different ways in which professionals experience personal and professional autonomy are partly explained by increased performance demands which appear to have a greater effect on experiences of personal autonomy.The strong relationship between performance demands and personal autonomy and the simultaneously weak relationship between performance demands and professional autonomy suggests that increasing performance demands might not serve to limit professional autonomy.