Maria M. AGOSTO-DE-FELICIANO, et al., Plaintiffs, Appellees, v. Awilda APONTE-ROQUE, etc., et al., Defendants, Appellants; Maria Teresa TORRES-HERNANDEZ, et al., Plaintiffs, Appellees, v. Pedro A. PADILLA, etc., et al., Defendants, Appellants; Maria Teresa TORRES-HERNANDEZ, Plaintiff, Appellant, v. Pedro A. PADILLA, etc., et al., Defendants, Appellees, 889 F.2d 1209

COFFIN, Senior Circuit Judge. Panel decisions in these two cases were withdrawn and both cases were reheard en banc. The issues common to both are whether and under what standards the protections of Elrod v. Burns, 427 U.S. 347, 96 S.Ct. 2673, 49 L.Ed.2d 547 (1976), and Branti v. Finkel, 445 U.S. 507, 100 S.Ct. 1287, 63 L.Ed.2d 574 (1980), should be available to government employees whose employment, while not terminated, was subjected to less advantageous conditions and responsibilities because of the employees’ political affiliation.

We convey our views on those legal questions here, and remand for application of the standard to the particular facts of each case. These cases represent a second wave in a tide of litigation arising out of changes in political administrations in Puerto Rico. The first wave involved outright dismissals and has produced a substantial number of opinions.

In all of these our task was to determine, in accordance with both Elrod and Branti, (1) whether the agency in which a discharged plaintiff had served was engaged in activity whose direction or pace properly could be affected by partisan political considerations, and (2) whether the particular position of the plaintiff was a policymaking or confidential one. On the basis of these criteria, we considered whether the plaintiffs job was protected against politically motivated discharge.

Here the plaintiffs have protected status, and they have not been discharged; the question at issue is what type of employer action short of dismissal will support a claim for violation of the plaintiffs’ right of free political association. I. Facts In November 1984, the Popular Democratic Party (PDP) won the gubernatorial election in Puerto Rico. Defendant Awilda Aponte Roque, a member of the PDP, was then named Secretary of the Department of Public Education. Aponte in turn named defendant Maria P. Scott as director of the Department’s Humacao Region.

Plaintiffs, all members of the New Progressive Party (NPP), are officials in the Humacao Region and have been “career” employees with the Department for more than twenty years. Plaintiffs’ action under 42 U.S.C. § 1983 alleges that political discrimination resulted in a drastic change and reduction of their duties, and that this violated the First and Fourteenth Amendments of the Constitution. The relevant testimony in their jury-waived trial may be summarized as follows. Defendant Scott assumed office on January 14,1985.

Within four days, without having met with any of the plaintiffs, including Maria Agosto de Feliciano, her immediate subordinate, and without having developed any clear reorganization plan, she issued a memorandum setting forth new “guidelines for the Department.” These removed responsibilities from plaintiffs, and required plaintiffs to obtain approval for their work from other employees who were PDP members and who previously had held positions subordinate to plaintiffs. Phone calls, mimeographing, photocopying and typing were to be similarly authorized.

According to several plaintiffs, defendant Scott explained that her actions stemmed from the recent political change. When plaintiffs wished to meet with her to discuss their problems, she insisted upon including her “confidence group” — PDP members who previously had been subordinate to plaintiffs. Plaintiffs twice complained by letter to defendant Aponte of being deprived of some of their functions, which had been reassigned to lower level personnel, and of their inability to communicate with defendant Scott. Aponte’s only response was to indicate that she was referring the matter to Scott.

We now summarize the evidence with regard to the specific changes in the plaintiffs’ responsibilities: Maria Agosto de Feliciano. Her prior job description consisted of some 26 responsibilities. Although the list appears somewhat inflated by bureaucratic jargon, it is nonetheless obvious that as the Assistant Regional Director she had varied supervisory and other responsibilities, including representation of the regional director in all activities in which the director is unable to participate. She also was in charge of the office in the regional director’s absence.

After defendant Scott’s arrival on the scene, Agosto’s duties were confined to the special education program and to signing checks. She no longer oversaw general supervisors. Virginia Diaz Diaz. She served as liaison between the Department of Public Education and private schools, coordinated the teaching practice program in the Project School without grades, and directed a re-gionwide committee on school organizations. After defendant Scott’s arrival, she continued to work with schools without grades and private schools, but no longer directed the school organizations committee.

In addition, she had to ask defendant Scott or Scott’s secretary for permission to use a telephone, to make photocopies, or to have typing done. Luz Camacho de Ortiz. Her prior list of responsibilities included some 23 items, ranging from surveying needs, developing work plans, and evaluating curricula and training, to managing vocational education and supervising student organizations.

After defendant Scott’s arrival, she was deprived of her former supervisory duties and had the sole functions of working in connection with talented students and of supervising an elementary school science program for which she had inadequate background. Vincente Vasquez Castro. His prior job description included 21 items of rather broad responsibilities in the areas of vocational education and student services. Evaluation of personnel, curriculum revision, budget distribution, development of new techniques, and oversight of programs relating to school needs, transportation, and scholarship were some of them.

After defendant Scott’s arrival, he no longer supervised; such responsibilities were assigned to a former subordinate, a PDP member. Vasquez’s sole function was partial responsibility for the school transportation program. Defendants advanced several arguments to the district court. They challenged the contention that the changes in plaintiffs’ duties were politically motivated, asserting that their knowledge of plaintiffs’ political affiliation was evidenced only by plaintiffs’ self-serving testimony.

Defendant Scott testified that the jobs held by plaintiffs had always been subject to reshaping and reassignment of duties by her predecessors, that plaintiffs’ remaining responsibilities were indeed substantial ones, and that she was in the midst of a process of determining the needs of her region and, when a complete analysis had been made, she would begin “adding functions pertinent to the position^].” The district court held that plaintiffs had established a prima facie case of political discrimination and that the defendants had not submitted a “credible justification.” 631 F.Supp. 1082 (D.P.R.1986).

It granted an injunction reinstating plaintiffs to their former duties. It observed that neither defendant had raised the defense of qualified immunity, and awarded, against both defendants, compensatory damages for emotional and mental distress in the amount of $60,000 for each plaintiff, and punitive damages in the same amount for each plaintiff.

Defendants filed this appeal, making the following arguments: 1) plaintiffs have no cause of action under the First or Fourteenth Amendments; 2) defendants are entitled to qualified immunity; 3) the district court’s findings of fact are erroneous; 4) the award of damages is excessive; and 5) the injunctive relief is overly broad. II. A Legal Standard for Adjudicating Allegedly Partisan Reallocation of Duties and Changes in Working Conditions of Civil Service Employees A. The difficulties in determining whether a government employee is protected from a politically motivated discharge are considerable.

See cases listed in note 1. As the facts of Agosto de Feliciano v. Aponte Roque illustrate all too well, however, the difficulties facing courts in cases involving employer action less final and definitive than dismissal are potentially enormous. They arise from the need to sift out the chaff of minor irritants and frustrations from the wheat of truly significant adverse actions.

The major issues presented in these cases are whether and under what circumstances public employer action short of dismissal, adversely affecting a career employee, based solely on the latter’s political affiliation, should be held to violate the First Amendment right of free association. By definition, we no longer have the bright line threshold of dismissal present in Elrod and Branti. Our search for workable standards promises to be difficult.

Indeed, Justice Powell in his dissenting opinion in Elrod speculated that “[t]he difficulty of formulating standards might pose a bar to judicial review of some patronage practices not before us.” 427 U.S. at 377 n. 1, 96 S.Ct. at 2691 n. 1. Other circuits have resolved these issues in widely diverging manners. The Second and Third Circuits have adopted a rather wide-open approach. See Lieberman v.

Reisman, 857 F.2d 896, 900 (2d Cir.1988) (“Whenever under color of state law unfavorable action is taken against a person on account of that person’s political activities or affiliation, it raises First Amendment concerns.”); Bennis v. Gable, 823 F.2d 723, 731 (3d Cir.1987) (“the constitutional violation is not in the harshness of the sanction applied, but in the imposition of any disciplinary action for the exercise of permissible free speech”).

See also Note, First Amendment Limitations on Patronage Employment Practices, 49 U.Chi.L.Rev. 181 (1982) (arguing that all politically inspired adverse personnel actions, whether or not equivalent in severity to dismissal, are unconstitutional). The Seventh and Fourth Circuits have been more restrictive. See Rutan v. Republican Party of Illinois, 868 F.2d 943, 949-51 (7th Cir.1989) (en banc), cert. granted, — U.S. —, 110 S.Ct. 48, 107 L.Ed.2d 17 (1989) (to be actionable, adverse personnel moves must amount to the “substantial equivalent [of] a dismissal”); Delong v.

United States, 621 F.2d 618, 623-24 (4th Cir.1980) (similar). But see Pieczynski v. Duffy, 875 F.2d 1331, 1333 (7th Cir.1989) (“Harassment of a public employee for his political beliefs violates the First Amendment unless the harassment is so trivial that a person of ordinary firmness would not be deterred from holding or expressing those beliefs.”) In any event, our starting point must be Elrod and Branti. We discern several teachings, either explicit or implicit, in those decisions. First, the holdings were explicitly confined to patronage dismissal cases.

Justice Stewart, his concurrence necessary to form a majority in Elrod, made it clear that “the broad contours of the so-called patronage system, with all its variations and permutations,” were not being considered. 427 U.S. at 374, 96 S.Ct. at 2690. We therefore realize that, in dealing with something less definitive than dismissals, we enter a field yet unploughed by the Court. Second, notwithstanding the limited scope of the Court’s holdings, they imply applicability of the First Amendment protection to some patronage employment decisions short of dismissal.

We say this because of the Court’s references to “political belief and association” as being “the core of those activities protected by the First Amendment.” See, e.g., Elrod, 427 U.S. at 356, 96 S.Ct. at 2681. Expanding on this theme, the plurality in Elrod quotes, with the endorsement of Justice Stewart, the sweeping language of Perry v.

Sindermann, 408 U.S. 593, 597, 92 S.Ct. 2694, 2697, 33 L.Ed.2d 570 (1972): “ ‘[I]f the government could deny a benefit to a person because of his constitutionally protected speech or associations, his exercise of those freedoms would in effect be penalized and inhibited.’ ” 427 U.S. at 359, 375, 96 S.Ct. at 2682, 2690. This language also found favor with the majority in Branti. 445 U.S. at 515, 100 S.Ct. at 1293.

In addition, we attach some significance to the fact that the Court in Branti saw fit to quote the district court’s explanation, accompanying its injunction prohibiting termination of plaintiff’s employment, that “[m]ere payment of plaintiffs’ salary will not constitute full compliance ...,” 445 U.S. at 509 n. 3, 100 S.Ct. at 1290 n. 3. Third, our resolution of the basic questions in these cases must recognize that there are First Amendment interests on both sides.

As the Court in Elrod stated, “It is apparent that at bottom we are required to engage in the resolution of conflicting interests under the First Amendment.” 427 U.S. at 371, 96 S.Ct. at 2688.

The First Amendment interest at stake on the government’s side that seemed to weigh most heavily in the Court’s analysis was “that representative government not be undercut by tactics obstructing the implementation of policies of the new administration, policies presumably sanctioned by the electorate,” id. at 367, 96 S.Ct. at 2686, or alternatively, the “need to insure that policies which the electorate has sanctioned are effectively implemented,” id. at 372, 96 S.Ct. at 2689. Although this interest is not sufficient “to validate patronage wholesale,” it is “not without force.” Id. at 367, 96 S.Ct. at 2686. B.

What we draw from these teachings is that our task is to devise a standard for evaluating employee burdens that protects against at least some politically motivated actions short of discharge, but that also gives due breadth to the government’s interest in the effective implementation of its policies. In searching for such a standard, we first consider what level of generality is most fitting. The choices, broadly speaking, are whether to apply categorical

definitions as in Elrod and Branti, or to engage in very fact-specific and ad hoc analysis of the interests of an employee, the extent to which those interests have been adversely affected by a government decision, and the particular needs of government. The latter approach has been adopted by the Supreme Court in cases where employees have been dismissed in retaliation for speech activity. See, e.g., Rankin v. McPherson, 483 U.S. 378, 107 S.Ct. 2891, 97 L.Ed.2d 315 (1987); Connick v. Myers, 461 U.S. 138, 103 S.Ct. 1684, 75 L.Ed.2d 708 (1983); Pickering v.

Board of Education, 391 U.S. 563, 88 S.Ct. 1731, 20 L.Ed.2d 811 (1968). This approach is well adapted to cases involving expression by public employees, for the form, content, manner, time, and place of expression can be infinitely varied, as can be the circumstances bearing on the employer’s legitimate needs. See Note, Politics and the Non-Civil Service Public Employee: A Categorical Approach to First Amendment Protection, 85 Colum.L.Rev. 558, 567 (1985). We believe a categorical approach is appropriate here, although it is not so obvious a choice as in Elrod and Branti.

In those cases, there were fewer variables; the employer action at issue always was a discharge. In the cases we now consider, however, we face a virtually infinite number of possible employer actions that will fall along a wide spectrum of severity. The employer's interest in taking politically motivated action also will vary significantly because the jobs at issue will range from those falling just outside the core political positions excepted from the Elrod-Branti protection to those with far more remote links to the political process.

We could, therefore, attempt to weigh the impact on the employee against the government’s interest in each case, constructing a kind of sliding scale of protection from political harassment. We decline to do so not only because such a procedure seems overwhelming in light of the many possible combinations of factors, but also because we believe there are strong reasons for adopting a categorical cutoff point of severity of harm below which actions simply should not be considered a constitutionally significant burden on the employee’s political association right.

First, if the threshold for an actionable constitutional violation were low, we believe employees would too often resort to litigation when the employer’s action was actually apolitical in nature. All employees, regardless of politics, may face a variety of lesser aggravations and inconveniences in the workplace, and a common catalogue of such non-political grievances — such as feelings of insufficient autonomy, complaints about unpleasant new duties, or restricted access to the telephone — could all too easily, and incorrectly, be ascribed to partisan political motivation.

Although the government employer could be expected to prevail in cases where there was no political motivation for the actions, we should discourage employees from drawing the courts into the day-to-day operations of government. Second, unlike in Rankin, Connick, and Pickering, where the public employer’s asserted interest was in efficiency, the government’s competing interest in a political patronage case also is based in the First Amendment.

See Elrod, 427 U.S. at 371, 96 S.Ct. at 2688 (recognizing administration’s interest in the effective implementation of political policies that presumably were sanctioned by the electorate). This interest would be undervalued if administrations could be forced to defend in court every decision to adopt new procedures, to create new organizational structures, and to reallocate functions.

Thus, again, while the government could be expected to prevail when it takes actions in furtherance of its interest, and should lose when the only motive is to reward or punish political friends or foes, we should seek a method for resolving valid complaints that will avoid excessive court intervention by giving some breathing room to legitimate efforts to make good on promises to the electorate. Third, we believe that even some politically motivated interactions between high-level policymakers and protected civil service employees should not trigger a constitutional cause of action.

These less serious occurrences — social preference or pressure, nuances affecting status and prestige, and articulated or indirectly manifested slurs and gibes — may be an all too real by-product of our long-standing organization of political life into two or more parties. See Bart v. Telford, 677 F.2d 622, 625 (7th Cir.1982). We believe it appropriate to require the citizen who believes and affiliates with a political party to develop a suitably thick skin to withstand the rigors of our societal, often highly politicized, life.

In short, these factors suggest that insubstantial changes in an employee’s work conditions and responsibilities, even when politically motivated, either would not reasonably chill the employee’s exercise of the right to free political association, or would cause a level of burden that is almost certainly outweighed by the government’s need to protect its own interest in implementing new policies.

We therefore believe our task is to develop a general standard that would at the same time protect the rights of political belief and association against real and substantial assaults, and yet also protect a government from a legal battle over every move it makes to implement its policies. We believe this balance is reached, and a cause of action for violation of the employee’s free association right therefore stated, only when the government’s actions are sufficiently severe to cause reasonably hardy individuals to compromise their political beliefs and associations in favor of the prevailing party. C.

We thus turn to our primary task, the creation of a workable standard for recognizing a violation of a civil servant’s freedom of association. We have pondered and labored at length in an effort to define this constitutional violation in a way that will neither unduly strain governments with harassing lawsuits nor unduly discourage employees with formidable complaints from asserting their rights. We have found that no simple catch phrase suffices to meet these dual goals.

In particular, we have rejected as unhelpful the “constructive discharge” or “tantamount to dismissal” formula adopted by at least two other circuits. See Rutan, 868 F.2d at 949-51; Delong, 621 F.2d at 623-24. If we used such a standard, we would be holding that an employee’s First Amendment right is significantly burdened only when he or she has, in effect, been fired for the exercise of that right. But comparing what happened to an employee in this context with the repercussions of discharge risks undervaluing the employee’s right.

Pension rights, seniority, fringe benefits, and the security represented by government employment all may be retained in these cases — as they were in the instant one — because the plaintiffs nominally may continue to hold the same jobs they previously had held. Moreover, many of the employees involved in such cases would be likely to endure severe hardship before contemplating resignation. In the case of a senior civil servant, for example, mobility is likely to be extremely limited and only rather cataclysmic changes could reasonably induce one to quit.

To link a constitutional violation to a finding of “constructive discharge,” or to conditions “tantamount to dismissal,” may therefore allow an imper-missibly high burden on the free exercise of the political association right. Thus, we think it ill-advised to adopt this short-hand method of describing the burden that we believe is sufficient to state a cause of action for violation of an employee’s free association right.

We also have rejected as insufficiently revealing the tag “constructive demotion,” although we note that the results of the standard we are about to describe would, in many respects, mirror those of a constructive demotion analysis. We nevertheless decline to adopt the short-hand phrase because when a bureaucrat is denied responsibilities and perquisites appropriate to the position occupied, it does not necessarily mean that he or she has been accorded those appropriate to a lower level.

In our view, a “constructive demotion” inquiry would simply obfuscate matters by raising the distracting question of the level to which the plaintiff has descended. We have, however, found instructive a precedent of our own that is repeatedly cited for its discussion of “constructive discharge,” Alicea Rosado v. Garcia Santiago, 562 F.2d 114 (1977).

That case has guided us not by its “tantamount to dismissal” standard but by two aspects of its methodology: first, the technique of canvassing specific injuries to evaluate the severity of the change wrought in the plaintiff’s employment position, and second, judging the change against a standard that asks whether the new work situation is “unreasonably inferior to his prior position.” In Alicea Rosado, we found no constructive discharge because the duties of the transferred plaintiff remained appropriate for his rank, even though he was no longer in charge of an office.

Although this meant a “limited blow to [his] pride,” 562 F.2d at 119, and the transfer would have increased his unreimbursed commuting costs, we concluded that “[a] more drastic reduction in the quality of working conditions [would be] needed,” id. at 120, to constitute a constructive discharge. We conclude that a similar technique, and a similar standard of severity of harm, are appropriate here, where our goal is to determine whether the new work conditions would place substantial pressure on even one of thick skin to conform to the prevailing political view.

This level of burden is reached, we believe, when the employer’s challenged actions result in a work situation “unreasonably inferior” to the norm for the position. To determine whether such a reduction has occurred — in other words, to evaluate whether the changes were sufficiently severe to warrant the “unreasonably inferior” description — the factfinder should canvass the specific ways in which the plaintiff’s job has changed. Because we recognize that the “unreasonably inferior” standard is not self-defining, and could be subject to widely varying

interpretations, we offer below some guidelines for judging severity. These are, of course, not prescriptive, but are designed to demonstrate the severity of change necessary to constitute a violation. We are well aware of the impossibility of devising detailed prescriptions that would cover future cases. We therefore caution that the suggested results in the following illustrations might well be altered by even slight factual variations. 1.

An employee who has lost merely the “perks” of his position — for example, the best office or secretary in the agency, unlimited telephone access or unusually minimal oversight — would not meet the “unreasonably inferior” standard. 2. An employee whose job has been substantially narrowed — perhaps reduced from manager of five departments to two — but who retains supervisory authority over matters of comparable significance to those taken away would not meet the “unreasonably inferior” standard. 3.

An employee who previously had predominantly exciting and responsible work and who was left with only a few routine and technical assignments could be found to meet the standard so long as his prior duties reflected the usual nature of his position rather than his prior high status as a member of the then-prevailing party.

For example, if an assistant to the director of an agency typically has no supervisory authority, but the potential plaintiff who is in that job had supervisory status under the last director because he was a political ally of that director or because he had some other idiosyncratic nonpolitical relationship (say, friendship or “golfing buddy” status) with his superior, a reduction of the employee’s role to the norm would not give rise to a violation.

On the other hand, if the assistant to the director traditionally has run the office in the director’s absence, and the new director now assigns this task — and others of responsibility — to someone who is the plaintiff’s subordinate, a factfinder would be entitled to term the new job conditions “unreasonably inferior.” 4. An employee whose job functions remain largely the same, but who is excluded from policymaking sessions that he or she once attended would not meet the “reasonably inferior” standard.

Such a person, whose duties are not politically sensitive enough to allow a patronage dismissal, may nevertheless as a party confidante have been invited to contribute ideas and advice to those in poli-cymaking positions. The new administration need not extend similar invitations to those who do not share its political outlook. 5. An employee who had been a supervisor and had worked independently in designing the projects necessary to carry out her duties and who lost only the freedom to work independently — instead being required to report regularly on her work — would not meet the standard.

Similarly, if the employee lost her supervisory role but continued to work independently on projects of significance, she would be unlikely to meet the standard unless the supervisory part of her job had been its defining characteristic. If, however, that same employee lost both her supervisory status and her independence, in all likelihood the factfinder would be entitled to conclude that she met the standard.

The decision would turn, however, on such factors as whether the supervisory role had been a primary part of her job, whether the duties she retained were challenging and significant, and whether new and inferior working conditions accompanied the change in duties (e.g., lost access to telephone and photocopier, poorer office accoutrements, worse hours). 6. An employee who is given one or two short-term assignments that are below his rank probably would not meet the standard of severity.

In general, an employee must show a permanent, or at least sustained, worsening of conditions to reach the threshhold of constitutional injury. If, however, a temporary change in duties is so inappropriate as to be demeaning and persists for longer than a week or two, the severity threshhold might be met. 7.

An employee who retains her job duties but who is criticized harshly and on a nearly daily basis, never provided guidance on how to improve, transferred to an undesirable office space (e.g., without ventilation or in a distant location from the remainder of the office staff), and whose requests for assistance are spurned would meet the severity level.

Our standard incorporates not only the substantive “unreasonably inferior” criterion as we have attempted to define and illustrate it, but also the procedural requirement that a plaintiff establish a change in conditions sufficiently severe to meet this high standard by clear and convincing evidence. We do not subject proof of an employer’s political motivation to this more rigorous burden requirement because we feel that political animus too often would be difficult to identify by proof beyond a preponderance.

Setting a more demanding requirement for the political motive requirement would, we feel, impermis-sibly chill this kind of First Amendment right of an individual. But proof of changed conditions is not so subject to camouflage, and requiring that it be established beyond a slight tilt in the believed testimony seems to us consistent with the First Amendment interest of the governmental employer. See Addington v. Texas, 441 U.S. 418, 424, 99 S.Ct. 1804, 1808, 60 L.Ed.2d 323 (1979) (“clear and convincing” standard has been used to protect notably important interests in civil cases). See also Anderson v.

Cryovac, Inc., 862 F.2d 910, 925-26 (1st Cir.1988) (adopting “clear and convincing” standard for proof that wrongfully concealed evidence was inconsequential). While recognizing the risk of too narrowly describing the task at hand, we suggest that the factfinder’s responsibility, in brief, is to determine whether the employee has retained duties, perquisites and a working environment appropriate for his or her rank and title.

In some instances, the factfinder will be able to look primarily at the history of the job at issue to determine whether the employee’s new role is unreasonably inferior to what the job is supposed to be. In other cases, however, there may be no stable prior job history to serve as a standard, and the factfinder will be able to examine only the change in the particular employee’s working conditions to determine whether the “new” job is unreasonably inferior to the one she previously had. In every case, the burden will be on the employee to establish this fact through clear and convincing evidence. D.

The case is not over, of course, even when a factfinder determines that the harm suffered by an employee is sufficiently severe to meet our standard. The employee still must persuade the factfinder by a preponderance of the evidence that the diminution in duties was motivated by discrimination on the basis of political affiliation. Under the formula used in political discharge cases, an employer then may seek to establish by a preponderance of the evidence that the changes would have been made regardless of political affiliation. See Mt. Healthy City Board of Education v.

Doyle, 429 U.S. 274, 287, 97 S.Ct. 568, 576, 50 L.Ed.2d 471 (1977); Cordero v. DeJesus Mendez, 867 F.2d 1, 5 (1st Cir.1989); Kercado Melendez v. Aponte Roque, 829 F.2d 255, 264 (1st Cir.1987).

Examples of such defenses would be a showing that plaintiff’s performance had been so unsatisfactory that assignment to less demanding duties was indicated, a showing that elimination of an employee’s most important projects resulted primarily from unavoidable budget cuts, or a showing that most employees within the relevant unit, regardless of political affiliation, experienced a reduction in duties as part of a personnel expansion and reallocation of responsibilities. One defense available to the government deserves more sustained attention. This is what we shall call the “changeover” defense.

It refers to a new administration’s claim that challenged actions taken reasonably soon after the “changeover” in power were designed to advance its First Amendment-related interest in implementing its policies. This defense has two aspects. The first recognizes the obvious fact of life that many changes in government can be expected early in a new administration’s tenure.

In order to best accomplish its policy goals, a new administration may substantially reorganize the structure of one or more departments, it may readjust departmental priorities, or it may initiate entirely new procedures for carrying out its functions. The changeover in government is thus likely to produce substantial alterations in certain employee’s jobs not because those employees are members of the outgoing party but because the incoming party, as a matter of policy, does not view those jobs to be important.

We believe an incoming administration must be given ample room to effect this sort of change without the fear of triggering a multitude of lawsuits by employees whose jobs have changed. Therefore, we believe the factfinder should give some deference to a new government’s explanation of how changes made shortly after it assumed power fit into its overall policy objectives. The government would retain the burden of proof on this issue, although its burden would be lightened by the deference given to it.

The second aspect of the “changeover defense” is more sensitive because it requires a modification of the traditional Mt. Healthy formula. Rather than requiring the government to state non-political reasons for its actions, this prong assumes that the new administration may, at times, feel the need to assign duties deemed espe-dally critical to its political philosophy to employees who share that philosophy. Such employees may be members of the administration’s own party.

This shift in duties based on political affiliation could be proper where the employer can show a reasonable basis for believing that such an employee would likely be more helpful in implementing the new policies than an employee who is a member of an opposing political party.

Although in such cases we are assuming that the cumulative duties of the employee who lost responsibility do not fit the Elrod-Branti exceptions, and that the employee therefore would be protected from political discharge, it may be that he is not protected from losing certain job duties that are politically sensitive within the new administration’s framework for government. On this prong, as well, the factfinder should give some deference to the government’s explanation of its needs.

Deference to the government’s explanation of its actions does not mean abdicating the responsibility to determine whether the new administration’s changes were in fact made to further the effective implementation of its policies. A plaintiff always must be given the opportunity to demonstrate that the government’s asserted justification is simply a pretext.

In evaluating the changeover defense, the factfinder should take into account, inter alia, whether the actions occurred precipitately or after some opportunity for appraisal, whether they seem connected with previously announced goals, and whether they flowed from an organizational or procedural study. In cases implicating the second prong of the “changeover defense,” the factfinder must look specifically at the duties in question to determine whether they concern politically sensitive matters.

Thus, drastic changes in job duties and conditions occurring early in the life of an administration may not be made entirely free of potential constitutional liability. Our policy of deference to the government’s explanations, however, should allow officials to take legitimate actions toward implementing their policies without undue fear of liability. So long as the government is able to provide a substantive reason for its decision that is supported by the facts in the record, the government will be able to justify its actions with the changeover defense. E.

We do not believe our standard will open the door to a flood of new lawsuits. The severity of the harm necessary to state a cause of action and the defenses available to government defendants should discourage those with petty complaints. Moreover, many cases may be ready for decision at the

summary judgment stage since, in our view, conversations and exhibitions of attitude are far less relevant than evidence as to working conditions and responsibilities. Judges should encourage the submission of sufficient documentation to enable such a resolution to be made as early and as often as possible. And to the extent that litigants or their lawyers make conclusory claims without “reasonable inquiry” to assure that claims are “well grounded in fact” and “warranted by ... law,” Fed.R.Civ.P. 11, the district courts have ample power to forestall abuses. See id. See generally Muthig v.

Brant Point Nantucket, Inc., 838 F.2d 600 (1st Cir.1988). Finally, although district court findings of fact will stand unless clearly erroneous, their measurement against the general standards we have set forth will be reviewed by us as a question of law: whether the politically motivated change in an employee’s work conditions imposes a substantial burden on the employee’s right to free association that is not outweighed by the government’s interest in implementing effectively its policies.

Our own effort will be directed at striking a balance which will be sensitive to both the rights of civil servants to their political beliefs and associations and to the rights of administrations in power to effectuate their perceived mandates from the electorate. We shall also remain keenly sensitive to the danger that courts be seized upon as arbiters of petty grievances. We recognize the existence of a vast mine field and the possibility that subsequent clearer perception of the hazards may well cause us to refine or change our standards. III.

Application of Standard We conclude that the district court should have the opportunity in the first instance to evaluate the plaintiffs’ complaints in light of the standard we announce today. That court may feel able to make judgments based on the evidence it has heard already, or it may choose to hear additional evidence on factors the parties did not previously deem relevant in determining whether an individual has been subjected to unreasonably inferior working conditions.

For example, the parties may not have developed evidence concerning the duties and perquisites appropriate for a given position, instead directing their energies toward showing the differences between an employee’s old job duties and the new ones. These two inquiries may not lead to the same result. We therefore remand these cases so that the district court may reconsider its judgments in light of the standard we have developed for identifying a violation of a government employee’s constitutional right of free political association.

Our disposition makes it unnecessary to consider at this time the other issues raised by the parties. The judgments of the district court are vacated, and the cases are remanded for further proceedings consistent with this opinion. . See, e.g., Figueroa Rodriguez v. Lopez Rivera, 878 F.2d 1478 (en banc) (1st Cir.1989); Caro v. Aponte Roque, 878 F.2d 1 (1st Cir.1989); Corde-ro v. DeJesus-Mendez, 867 F.2d 1 (1st Cir.1989); Rodriguez Burgos v. Electric Energy Authority, 853 F.2d 31 (1st Cir.1988); Estrada Izquierdo v. Aponte Roque, 850 F.2d 10 (1st Cir.1988); Hernandez-Tirado v.

Artau, 835 F.2d 377 (1st Cir.1987); Morales Morales v. Arias, 834 F.2d 255 (1st Cir.1987); Nunez v. Izquierdo-Mora, 834 F.2d 19 (1st Cir.1987); Zayas-Rodriguez v. Hernandez, 830 F.2d 1 (1st Cir.1987); Vazquez Rios v. Hernandez Colon, 819 F.2d 319 (1st Cir.1987); Perez Quintana v. Gracia Anselmi, 817 F.2d 891 (1st Cir.1987); Raffucci Alvarado v. Zayas, 816 F.2d 818 (1st Cir.1987); Rosado v. Zayas, 813 F.2d 1263 (1st Cir.1987); Mendez-Palou v. Rohena-Betancourt, 813 F.2d 1255 (1st Cir.1987); Monge-Vazquez v. Rohena-Betancourt, 813 F.2d 22 (1st Cir.1987); Cheveras Pacheco v.

Rivera Gonzalez, 809 F.2d 125 (1st Cir.1987); Rodriguez Rodriguez v. Munoz Munoz, 808 F.2d 138 (1st Cir.1986); Jimenez Fuentes v. Torres Gaztam-bide, 807 F.2d 236 (1st Cir.1986) (en banc); DeChoudens v. Government Development Bank of Puerto Rico, 801 F.2d 5 (1st Cir.1986) (en banc); De Abadia v. Izquierdo Mora, 792 F.2d 1187 (1st Cir.1986). . We detail here the facts in Agosto de Feliciano v. Aponte Roque, No. 86-1300, 1987 WL 39747, to provide context for the legal issues we decide herein. We see no need to present as well the factual background of Torres Hernandez v.

Padilla, Nos. 86-1643 and 86-1644, 1987 WL 39748, which involves only one additional plaintiff. . A career employee under Puerto Rico’s civil service personnel system is selected strictly on merit and can be removed only for cause. Puerto Rico Public Service Personnel Act of 1975, 3 L.P.R.A. §§ 1301, 1331-1338. In contrast, confidential employees under the Act are of “free selection and removal," and are those who "intervene or collaborate substantially in the formulation of the public policy, who advise directly or render direct services to the head of the agency_” Id. at §§ 1350 and 1350(4). .

The fifth plaintiff, Vega Diaz, was found by the district court not to have established in sufficient detail the tasks assigned to him before and after the change of government and, therefore, "failed to prove that there had been substantial changes in his tasks ... or that the same were changed to tasks of lesser category.” No appeal was taken from this holding. . Categorical analysis involves development of a general standard separating protected from unprotected activity that can be applied to every case.

In Elrod and Branti, the Court established that employees whose jobs were categorized as policymaking, confidential, or communicative were not protected from political discharge, so long as party affiliation is an appropriate requirement for the position. See Elrod, 427 U.S. at 367-68, 375, 96 S.Ct. at 2686-87, 2690; Branti, 445 U.S. at 517-18, 100 S.Ct. at 1294-95; Jimenez Fuentes, 807 F.2d at 242.

The Court also assumed that discharges of such individuals would be justified in light of the new administration's interests, and did not require ad hoc scrutiny of the weight of the government’s interest in any particular case. . As did the Supreme Court in Elrod and Branti, we assume here that the employee’s interest in free political association and the administration's interest in the effective implementation of the electorate’s preferred political policies are substantial.

Elrod and Branti tell us, however, that when the administration’s interest is at its highest point, it outweighs the individual's right to be free from political pressure. Thus, an employee in a policymaking position may be discharged solely on the basis of political affiliation. .Our reference to a "categorical” approach does not mean that the trial court would be freed from a detailed inquiry into what happened to the employee and why the government took the actions it did.

Rather, once the court has found the facts, it will look at whether the employee has met the standard we set today for proving an actionable First Amendment violation. This is categorical in the sense that the factfinder will be required to apply the standard and will not be free simply to balance the interests at stake in the individual case. . We have treated outright demotions, which involve reductions in pay and official rank, as equivalent to discharges. See, e.g., Gierbolini Colon v. Aponte Roque, 848 F.2d 331, 333 & n. 2 (1st Cir.1988). .

The question of political motivation could be resolved before the issue of the severity of the harm. For example, in instances in which discovery shows that the defendants’ actions obviously were motivated by lawful concerns, it is likely that a district court faced with a defense motion for

summary judgment would choose to dispose of the case on that ground rather than to delve into the facts concerning the change in duties. . We add this important caveat: an administration ought not to be deprived of a “reorganization” defense simply because it took place at a relatively late point in its term of office. Some reforms and reorganizations may be much more complex than others and require much more investigation, analysis and preparatory drafting. And some ideas for restructuring may simply have surfaced later in good faith.

We see no need to give the government unusual deference when reviewing late-blooming changes. It should be easy enough for a new administration to defend legitimate reorganizations that come late in its term since such changes are more likely to be documented, and less likely to be politically motivated, than changes made when a new administration first comes to power. . A government defense based on policy needs presents another opportunity for ad hoc balancing.

We could decide on a case-by-case basis whether the government’s need to take an employee action based on political affiliation is outweighed by the burden on the employee. We again choose a formulaic approach: if the government demonstrates that there is some policy justification for internal reshuffling of duties, in view of its mandate as a democratically elected administration, we will refrain from any further balancing. This approach is consistent both with Elrod and Branti and with our view that governments must have some breathing room to make good on promises to the electorate.

The employee, of course, must be given the opportunity to refute the government’s assertion that policy concerns motivated the disputed action. . For example, changes made within days of a new administration’s ascent to power ordinarily would be more likely to reflect an improper political housecleaning than would changes made months later, after the new officials have had a chance to evaluate how to reorganize their departments to best meet their policy goals. .

In suggesting some deference to a new administration’s explanation of how its program was aided by the changes it made shortly after assuming power, we do not mean to set out a fixed criterion for every case. What we face in the instant case and its companion is a restructuring of important jobs and reassignment of duties to political confidants, the kind of action frequently sought to be justified on policy grounds, particularly right after a new government assumes power.

We do not consider here decisions with a more individualized impact— decisions in which the policy justification is more difficult to perceive because the adverse action against a political opponent takes place with no discernible parallel change in favor of a political colleague. Such actions could include a transfer to a distant state, Delong v. United States, 621 F.2d 618 (4th Cir.1980) or a denial of a scheduled promotion or salary increase, Robb v. City of Philadelphia, 733 F.2d 286 (3d Cir.1984); Allaire v. Rogers, 658 F.2d 1055 (5th Cir.1981); Bickel v. Burkhart, 632 F.2d 1251 (5th Cir.1980).

Such decisions arguably present such a divergence from normal administrative practice that assertions of policy justification warrant particular skepticism rather than deference. We express no views on such cases at this time. . We reject defendants’ urging that we review the facts de novo as in cases involving free speech. See Figueroa v. Aponte Roque, 864 F.2d 947, 949 n. 3 (1st Cir.1989).