Susan Farrell v. Planters Lifesavers Company Nabisco, Inc

OPINION OF THE COURT RENDELL, Circuit Judge. This appeal raises a number of employment law issues relating to the recruitment, hiring and later firing of appellant, Susan Farrell. The District Court granted summary judgment in favor of defendants Planters Lifesavers Company and Nabisco, Inc., on all of Farrell’s claims. 1 See Farrell v. Planters Lifesavers Co., 22 F.Supp.2d 372 (D.N.J.1998). Farrell appeals the dismissal of her retaliation and quid pro quo sexual harassment claims under Title VII and the dismissal of her North Carolina contract law claim. 2 Act *275 ing as Amicus Curiae, the Equal Employment Opportunity Commission supports Farrell’s appeal from the dismissal of her federal claims. We find that Farrell established a prima facie case for both federal causes of action, and we will reverse the grant of summary judgment precluding her retaliation and quid pro quo claims. We will, however, affirm the dismissal of her state contract law claim. We have jurisdiction over this appeal pursuant to 28 U.S.C. § 1291. The District Court had jurisdiction over the Title VII claims under 28 U.S.C. § 1331, and exercised jurisdiction over the state law claim under 28 U.S.C. § 1332. Viewing the record from Farrell’s perspective, the facts in this case are as follows. I. In 1992, Planters, then located in Winston-Salem, North Carolina, hired Douglas DeLong as its Director of Materials Management. One of DeLong’s central tasks involved formulating a plan to cut operating costs in Planters’ Materials Management Department. Beginning in the early 1990s and continuing through 1997, Planters sought to cut its operating costs by consolidating its workforce. DeLong quickly reorganized the Materials Management Department, bringing together the Purchasing, Packaging Services, Graphic Design and Production Planning Departments within the Materials Management Department and placing them all under his direct authority. In August 1993, DeLong wrote a memorandum to Norm Jungman, his supervisor, suggesting Planters merge the Packaging Services and Graphic Design Departments as part of its consolidation. DeLong explained that he hoped to merge the two departments by late 1994 or early 1995. In late 1993, Planters decided to discharge the Director of Packaging Services, Ronald Yonker. Almost contemporaneously, Planters approached Susan Farrell, through a recruiter, to become a Packaging Engineer in the Packaging Services Department. Based on her qualifications, and DeLong’s recommendation, Planters subsequently considered Farrell as a candidate to replace Yonker as Director of the Packaging Services Department. At the time, Farrell was a packaging engineer at McCormick & Company in Hunt Valley, Maryland. In January of 1994, Farrell traveled to Winston-Salem to interview. By mistake, Planters had Farrell interview with Yonker, who did not know of the decision to fire him and believed Farrell to be interviewing for another position. Concerned about her own job security by virtue of Planters’ treatment of Yonker, Farrell sought assurances during the recruitment process. DeLong assured her that she would only be fired for poor performance. A number of individuals told her that Yonker had been repeatedly warned about his performance before the decision was made to terminate him. Planters formally offered Farrell Yonker’s position, re-titled as Senior Manager of Packaging Services, by letter dated February 4, 1994. Planters also promised to purchase Farrell’s home in Maryland for $240,000 and pay for Farrell’s relocation back to Maryland if her employment with Planters ended within two years because of “performance concerns or position elimination.” Farrell accepted the offer on February 11, 1994, and relocated to North Carolina. She began work at Planters on March 28, 1994. Farrell’s husband remained in Maryland. In mid-November 1994, Farrell traveled . to Chicago to attend a Pack Expo, an annual packaging exposition, with a number of Planters and Nabisco, Inc. managers. While attending the show, DeLong told Farrell that his supervisor, Norm Jungmann, was about to be fired and that he would assume Jungmann’s position shortly. DeLong then praised Farrell’s work performance, told her that he felt her *276 style complemented his, and asked her if she would be interested in becoming the head of the Industrial Engineering Department in addition to her duties as manager of the Packaging Engineering Department once he replaced Jungmann. A few hours later, DeLong asked Farrell to accompany him the next day on a planned business trip to Puerto Rico to tour a Planters’ facility. DeLong instructed Farrell to book tickets on the same flight as his, with seats together. Farrell made the arrangements. DeLong and Farrell had traveled on business trips together on two prior occasions. During the flight to Puerto Rico on November 16, 1994, DeLong placed his hand just above Farrell’s knee while telling Farrell that his wife became jealous when he traveled with Farrell. He asked Farrell whether her husband became jealous when she traveled with DeLong. Farrell responded by removing his hand from her leg and firmly telling him “no, I don’t give him a reason to and I suggest you do the same.” Farrell says DeLong’s demeanor changed when she rejected his advance: he turned away, curled up and slept or pretended to sleep. 3 Farrell and DeLong engaged in little or no further conversation for the rest of the flight. The next day, DeLong flew back to Winston-Salem, informing Farrell that he was leaving a day early in order to find out more about Jung-man’s termination. Farrell also says that DeLong often commented when she wore a skirt, and states that in October, DeLong told her that she was pretty calm considering she was living apart from her husband and that he would be “bitchier” if he were her. However, after the November flight, De-Long never made reference to the advance on the plane, nor made a second advance. On December 13, 1994, less than a month after the trip to Puerto Rico and less than two weeks after Planters paid for her possessions to be moved to North Carolina and purchased her home in Maryland, Gary Eckenroth, Planters’ Vice President for Human Resources, went to Farrell’s office and asked her to come up to his office. On the way, Eckenroth told her that Planters was going to eliminate her position. Once inside his office where DeLong was waiting, Farrell says Ecken-roth told her Planters would call her termination a position elimination, in order to allow Farrell to retain benefits and give her a severance package, but that she was actually being terminated because of interpersonal problems with other members of Planters’ management. ’ When Farrell asked for specifics, DeLong mentioned Suzanne Jabbour and Ed Lyons. He also made general reference to some others whom he did not name. After Farrell protested, questioning why Planters had just paid for movers and purchased her house in Maryland if they were terminating her, Eckenroth said that he did not know that Planters had just moved her. Eckenroth then asked DeLong to leave the room. Once DeLong left, Farrell says that Eckenroth told her that he had not checked DeLong’s report of complaints and he promised he would do so. The next day, Farrell spoke with Jabb-our and Lyons and they both denied making negative comments to DeLong, but confirmed that DeLong had asked them about her. Lyons also told Farrell that he had told DeLong that he felt DeLong had a personal problem with Farrell. That same day, Farrell says Eckenroth told her that Jabbour and Lyons came to him and confirmed that they had not made negative comments about her to DeLong. In fact, DeLong’s own internal memorandum, dated December 8, 1994 states “Ed Lyons *277 said she was helpful with his group.” Eck-enroth also apologized for not having investigated DeLong’s claim, but told Farrell that she would have to leave because rumors of her termination had begun to circulate. A few days later, Eckenroth talked to her about accepting the severance package in return for releasing her claims. Farrell did not agree to any terms of separation and left her position on December 28,1994. DeLong’s memorandum, dated December 8, describes the events leading up to Farrell’s termination. The memorandum refers to three conversations he had on December 7 and 8 with various Planters’ managers who came to DeLong and complained about Farrell. The memorandum then refers to a meeting between DeLong and Eckenroth, spurred by these conversations, where they “discussed the option of eliminating” Farrell’s position and merging the Packaging and Graphics Departments as DeLong had suggested in 1993. DeLong notes that he spoke with Planters’ managers, including Jabbour and Lyons, “to gather feedback about Farrell.” DeLong summarizes their comments: “The most common response received was, T don’t know what she does.’ ‘A lot of talk but no results.’ ‘Nice suit, but nothing in it.’ Ed Lyons said she was helpful with his group.” DeLong then detailed his subsequent conclusions, including: When all the issues with her peers were discussed and other feedback received discussed, it was clear that I had to deal with Susan. I could not run an area with the type of conflict that existed between Susan and the rest of my staff. [biased on these discussion and conversations with people over the last several months who found Susan very difficult to work with, i.e., Phil, Mike, Peggy and Rob, I made the decision to eliminate her job and combine Graphics and Packaging under Peggy as per earlier recommendation and hire another Packaging Engineer. Farrell alleges that her rejection of De-Long’s advance resulted in her termination, forming the basis of her federal claims. She also argues that DeLong’s assurances created an implied term in her employment contract permitting her termination for cause only. Farrell alleges that Planters breached this provision by firing her, forming the basis for her state law claim. Planters disputes many of the facts alleged by Farrell, and the inferences that might be drawn from them, and describes different reasons for Farrell’s termination. According to Planters, Farrell was terminated because upper management made the decision to consolidate the Graphics and Packaging Departments and Planters determined that Peggy Bryan, the head of the Graphics Department, would better serve Planters than Farrell in the consolidated position. According to Planters, Eckenroth met with Sandy Putnam, Vice President of Planters, to discuss cost containment and reduction measures in November of 1994 and specifically discussed the possibility of implementing DeLong’s 1993 proposal that Planters should consolidate the Packaging Services and Graphic Design Departments in late 1994 or early 1995. Planters explains that DeLong and Eck-enroth met on December 8 to discuss the consolidation. At that meeting, Eckenroth asked DeLong to make a recommendation about whether they should keep Farrell or Bryan, which he did. Planters contends that DeLong’s December 8 memorandum is materially consistent with their explanation for Farrell’s termination since it describes a discussion about consolidating departments, efforts by DeLong to canvass opinion about Farrell, and the decision to retain Bryan. Furthermore, in his certification and deposition testimony, DeLong denied ever having suggested to Farrell the possibility of a promotion or making the advance on *278 the plane. Eckenroth and DeLong also denied that they told Farrell that the decision was a position elimination in name only during their meeting with Farrell. In their certifications and deposition testimony, DeLong and Eckenroth also support the choice to retain Bryan by stating that they received complaints about Farrell’s availability, ability and attitude throughout her employment. Planters points out that in the summer of 1994, a peer group gave Farrell the mock award title of “phantom leader,” and in October, DeLong wrote a memorandum to Farrell telling her to increase her visibility at Planters. After her termination, Farrell filed suit. The District Court granted Planters summary judgment on all of Farrell’s claims. Farrell appeals and argues that the District Court erred in dismissing her Title VII quid pro quo sexual harassment and retaliation claims as well as her North Carolina contract law claims. We discuss each in turn, beginning with Farrell’s Title VII claims. We exercise plenary review over summary judgment and we apply the same standard that the lower court should have applied. See Armbruster v. Unisys Corp., 32 F.3d 768, 777 (3d Cir.1994). A federal court should grant summary judgment “if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” See Fed.R.Civ.P. 56(c). In making this determination, “a court must view the facts in the light most favorable to the nonmov-ing party and draw all inferences in that party’s favor.” Armbruster, 32 F.3d at 777. II. Farrell claims that Planters and Nabisco violated Title VII in two separate ways and presents her discrimination claim as two different causes of action. She alleges that her termination was both an impermissible act of retaliation and an act of quid pro quo sexual harassment. In this case, it is clear that both of Farrell’s claims rely upon the same essential facts: DeLong’s sexual advance, her rejection of the advance and her subsequent termination. The District Court determined that Farrell failed to produce evidence showing a causal link between the rejection and her termination, as is required to establish a prima facie case for each claim. Considering Farrell’s retaliation claim, the District Court first determined that the three to four week period between the rejection and the termination was insufficient alone to establish the causal link. The court then searched the record for evidence of a “pattern of antagonism” or a “retaliatory animus.” Finding none, the District Court granted the defendant’s motion as to Farrell’s retaliation claim. The District Court dismissed Farrell’s quid pro quo sexual discrimination claim in a similar manner. Relying heavily upon Lynch v. New Deal Delivery Serv., Inc., 974 F.Supp. 441 (D.N.J.1997), the Court dismissed Farrell’s claim because there was no evidence that DeLong either acted hostilely towards Farrell after she rejected his advance or pressed the issue again. For the reasons described below, we conclude that the District Court erred by requiring that the causal connection for both claims be supported by a pattern of antagonism, retaliation or hostility and, thereby, engaged in too narrow a review of the plaintiffs evidence. Considering the record before us, we find ample evidence from which to infer a causal connection between Farrell’s rejection of DeLong’s advance and her subsequent termination that enables Farrell to make out a prima facie case for both her claim of retaliation and her claim of quid pro quo sexual harassment. Since we find that Farrell makes out a prima facie case on all the evidence before us, we need not decide whether the three to four week period between the advance and termination *279 would be sufficient, if considered alone. We note that the District Court did not reach the issue of whether Planters and Nabisco proffered a legitimate non-discriminatory reason for Farrell’s termination or whether Farrell could illustrate that the reason was pretextual, so we will reverse and remand to the District Court for consideration of those issues. We will discuss the requirements for Farrell’s pri-ma facie case of retaliation and quid pro quo sexual harassment in that order. A. Under our precedent, to advance a prima facie case of retaliation, a plaintiff must show that: (1) the employee engaged in a protected employee activity; (2) the employer took an adverse employment action after or contemporaneous with the employee’s protected activity; and (3) a causal link exists between the employee’s protected activity and the employer’s adverse action. See, e.g., Kachmar v. Sungard Data Systems, Inc., 109 F.3d 173, 177 (3d Cir.1997); Jalil v. Avdel Corp., 873 F.2d 701, 708 (3d Cir.1989); see also Krouse v. American Sterilizer Company, 126 F.3d 494, 500 (3d Cir.1997) (describing the third requirement as a “causal connection”). The District Court determined that Farrell failed to establish the third element of the prima facie case. 4 It reasoned that “a temporal proximity of three to four weeks may support an inference of retaliation,” but held “[a]bsent evidence of intervening antagonism or retaliatory animus ... Farrell has failed to establish a causal link between her rejection of DeLong’s advance and her termination. Thus, Farrell has not established a prima facie case of retaliation.” Farrell, 22 F.Supp.2d at 393. The District Court limited its inquiry into whether Farrell offered any non-temporal proof of causation to evidence of a “pattern of antagonism” or “retaliatory animus” and,finding none, dismissed her claim. See id. (citing Kachmar, 109 F.3d at 177 and Krouse, 126 F.3d at 503-04). We think this analysis viewed too narrowly the scope and nature of conduct and circumstances that could support the inference of causation. The District Court seemed to have been requiring more than one retaliatory act, or one closer in temporal proximity, or some demonstrative activity, to the exclusion of all other facts or events potentially probative of causation. In doing so, it committed error. We have spoken often of the probative value of temporal proximity in retaliation cases. Recently in Robinson v. City of Pittsburgh, 120 F.3d 1286 (3d Cir.1997), and Krouse v. American Sterilizer Co., we remarked that our case law is “seemingly split” as to whether temporal proximity between the protected activity and the alleged retaliatory act can be sufficient in itself to create an inference of a causal connection for the purposes of a prima facie case of retaliation. See Robinson v. City of Pittsburgh, 120 F.3d at 1302; see also Krouse, 126 F.3d at 503 (finding period of nineteen months insufficient on summary judgment). However, we caution that this “split” is not an inconsistency in our analysis but is essentially fact-based. Rather, we have ruled differently on this issue in our case law, depending, of course, on how proximate the events actually were, and the context in which the issue came before us. 5 *280 For example, in Jalil v. Avdel Corporation, 873 F.2d 701 (3d Cir.1989), we reversed the grant of summary judgment in favor of the defendant because the plaintiff had established causation for the purposes of his prima facie case merely by showing that his discharge occurred only two days after his employer had received notice of Jalil’s EEOC claim. See Jalil, 873 F.2d at 708 (“He demonstrated the causal link between the two by the circumstance that the discharge followed rapidly, only two days later, upon Avdel’s receipt of notice of Jalil’s EEOC claim.”). However, in Krouse, also a case appealing the grant of summary judgment, we explained that temporal proximity alone will be insufficient to establish the necessary causal connection when the temporal relationship is not “unusually suggestive,” and determined that nineteen months was too attenuated to create a genuine issue of fact. See Krouse, 126 F.3d at 503 (citing Robinson, 120 F.3d at 1302). In this case, Farrell, supported by the EEOC, asks that we find that the relatively short period in question sufficient to establish the inference of causation required at this stage. 6 However, because we find that she establishes the inference on the entire record before us, including the suggestive timing, we need not decide whether the timing alone would be sufficient. In the case before us, the District Court required additional evidence of “intervening antagonism or retaliatory animus,” after it opined that the timing of three to four weeks “may” support an inference of causation. In essence, it held that if temporal proximity is not clearly suggestive standing alone, a “time plus” other intervening retaliatory acts will be required. We part ways with the District Court, because while we agree that timing plus other evidence may be an appropriate test where the temporal proximity is not so close as to be “unduly suggestive,” we disagree as to the character of the “other” evidence that will suffice to create the causal link for purposes of the prima facie case. The District Court drew its reasoning from our statement in Kachmar that “[wjhere there is a lack of temporal proximity, circumstantial evidence of a ‘pattern of antagonism’ following the protected conduct can also give rise to the inference.” See Farrell, 22 F.Supp.2d at 393 (quoting Kachmar, 109 F.3d at 177). However, it faked to note the succeeding sentence, which is all-important: “These are not the exclusive ways to show causation, as the proffered evidence, looked at as a whole, may suffice to raise the inference.” Kach-mar, 109 F.3d at 177. Although timing and ongoing antagonism have often been the basis for the causal link, our case law clearly has allowed a plaintiff to substantiate a causal connection for purposes of the prima facie case through other types of circumstantial *281 evidence that support the inference. For example, a plaintiff may establish the connection by showing that the employer gave inconsistent reasons for terminating the employee. See Waddell v. Small Tube Products, Inc., 799 F.2d 69, 73 (3d Cir.1986); see also EEOC v. L.B. Foster Co., 123 F.3d 746, 753-54 (3d Cir.1997), cert. denied, 522 U.S. 1147, 118 S.Ct. 1163, 140 L.Ed.2d 174 (1998). In Waddell, we noted that the District Court could “appropriately” have taken inconsistent explanations into account in finding the causation necessary to satisfy the prima facie case. See Waddell, 799 F.2d at 73 (“The district court noted the inconsistency in Small Tube’s explanations of its refusal to rehire and could appropriately have taken that into account.”). In L.B. Foster Co., we noted that the plaintiff had established a prima facie case of retaliation based on temporal proximity between the events plus inconsistencies in the defendant’s testimony, certain conduct towards others, and refusals to provide a reference for the plaintiff. See L.B. Foster, 123 F.3d at 753-55 (citing Waddell, 799 F.2d at 73). In Woodson v. Scott Paper Co., 109 F.3d 913, 921 (3d Cir.1997), we limited our inquiry to evidence of a pattern of antagonism, but only because that evidence was sufficient on its own to link the complaints of discrimination and Woodson’s discharge. See id. at 921. We specifically stated that we “need not consider whether other types of evidence might also support a causal link finding in the absence of temporal proximity.” Id. at 921 n. 3. Similarly, in Robinson v. SEPTA, 982 F.2d 892 (3d Cir.1993), the intervening pattern of antagonism was so strong that it overcame the lack of temporal proximity and, alone, proved the causal link. See id. at 895. There was no need to look beyond this pattern for other, circumstantial, evidence. Moreover, we have been willing to explore the record in search of evidence, and our caselaw has set forth no limits on what we have been willing to consider. In Krouse, we affirmed the District Court’s grant of summary judgment after concluding that the plaintiff “has not proffered any evidence establishing a causal connection” other than the nineteen month gap between filing an EEOC complaint and the alleged retaliatory act. See Krouse, 126 F.3d at 503. We noted there that “[w]hen temporal proximity between protected activity and allegedly retaliatory conduct is missing, courts may look to the intervening period for other evidence of retaliatory animus.” Id. Accordingly, we conclude that the District Court employed too restrictive a view of the type of evidence that can be considered probative of the causal link. It is not limited to timing and demonstrative proof, such as actual antagonistic conduct or animus. Rather, it can be other evidence gleaned from the record as a whole from which causation can be inferred. As we explained in Kachmar, “[i]t is important to emphasize that it is causation, not temporal proximity [or evidence of antagonism], that is an element of plaintiffs prima facie case, and temporal proximity [or antagonism] merely provides an evidentiary basis from which an inference can be drawn.” Kachmar, 109 F.3d at 178. Before viewing the record with this wider lens, however, we will focus briefly on the District Court’s decision that Farrell had not satisfied the same element — namely causation — with respect to the prima facie case for her claim of quid pro quo sexual harassment. B. Farrell’s companion claim of quid pro quo sexual harassment contains a similar requirement of cause and effect. See Robinson v. City of Pittsburgh, 120 F.3d 1286, 1296 (3d Cir.1997). In Robinson— one of the few cases in which we have shed light on this issue — we stated that a plaintiff may prove a claim of quid pro quo sexual harassment by showing that “his or her response to unwelcome advances was subsequently used as a basis for a decision about compensation, [terms, conditions, or *282 privileges or employment].” Id. at 1297. We further explained that “the plaintiff need not show that the submission was linked to compensation, etc., at or before the time when the advances occurred. But the employee must show that his or her response was in fact used thereafter as a basis for a decision affecting his or her compensation, etc.” Id. The District Court granted summary judgment to Planters on Farrell’s quid pro quo claim finding that she failed to establish the causal connection required by Robinson. The District Court used the test for satisfying the prima facie stage that had first been set forth by the Court of Appeals for the Eleventh Circuit in Henson v. City of Dundee, 682 F.2d 897 (11th Cir.1982). See Farrell, 22 F.Supp.2d at 386(“(4) her submission to the unwelcome advances was an express or implied condition for receiving job benefits or her refusal to submit resulted in tangible job detriment”) (emphasis added). Curiously, however, the District Court relied almost exclusively on Lynch v. New Deal Delivery Serv. Inc., 974 F.Supp. 441, 452 (D.N.J.1997)- a case based on diversity jurisdiction focusing exclusively on New Jersey state law — to conclude that Farrell’s causation evidence fell short because, like the plaintiff in Lynch, Farrell did not produce any evidence that DeLong “pressured her” or “acted hostile toward Farrell after she rejected his advance.” See Farrell, 22 F.Supp.2d at 887-88. Farrell contends that the District Court erred in finding that she had not adduced evidence sufficient to establish the causal relationship required by Robinson. We agree and will reverse because we believe that — by relying upon Lynch — the District Court departed from our decision in Robinson and — much as it did with respect to its analysis of the retaliation claim — incorrectly narrowed the scope and nature of its assessment of the causation evidence. As we will explain below, we view the evidence as sufficient to establish this causal link as well. In Lynch, the plaintiff raised a number of discrimination claims, including quid pro quo sexual harassment, under the New Jersey Law Against Discrimination. The district court granted summary judgment to the defendant because it found that since the plaintiff admitted that the defendant did not act hostilely towards her or attempt to coerce her after she declined his invitations to dinner, “no reasonable factfinder could conclude [the defendant’s] attentions ever involved an implicit or explicit threat” of retaliation as required under NJLAD. See Lynch, 974 F.Supp. at 452 (citing Lehmann v. Toys R Us, Inc., 132 N.J. 587, 626 A.2d 445 (1993)). The District Court, noting that it found the reasoning in Lynch “persuasive,” dismissed Farrell’s quid pro quo claim because there was no further pressure or hostility after she rejected DeLong’s advance. See Farrell, 22 F.Supp.2d at 387-88. However, our law contains no requirement that the plaintiff show that the employer implicitly or explicitly threatened retaliation when making the advance. We explained in Robinson that “the plaintiff must show that his or her response to unwelcome advances was subsequently used as a basis for a decision about compensation, etc. Thus, the plaintiff need not show that the submission was linked to compensation, etc., at or before the time when the advances occurred.” Robinson, 120 F.3d at 1297 (emphasis added). While evidence of hostility or repeated demands for sexual favors would strengthen any plaintiffs case, the lack of such evidence does not render it fatally flawed. By following the reasoning of Lynch, the District Court narrowed its analysis of Farrell’s evidence of causation by effectively en-grafting an element onto the cause of action that is not required under our jurisprudence. 7 *283 Robinson, however, as the District Court noted, does require the plaintiff to “show that his or her response was in fact used thereafter as a basis for a decision affecting his or her compensation, etc.” Id.; Farrell, 22 F.Supp.2d at 387. As with Farrell’s retaliation claim, the question becomes what evidence may the court consider in deciding whether that nexus is sufficiently proven to establish a prima facie case. While we recognize that a retaliation claim under Title VII and an adverse job discrimination claim are separately codified, compare 42 U.S.C. § 2000e-3(a), tvith 42 U.S.C. § 2000e-2(a), we see no reason to conclude that Farrell’s burden should be higher, or the scope of evidence permissibly considered narrower, in this cause of action for quid pro quo sexual harassment than in a retaliation claim. Our decision in Robinson, where we reversed the grant of judgment as a matter of law in favor of the defendants, leads us to conclude that the analysis can be, in fact, quite broad. See Robinson, 120 F.3d at 1298-99. 8 In Robinson, the plaintiff alleged that her supervisor blocked her transfer because she rejected his advance. See id. at 1298. At trial, the plaintiff testified that her supervisor had repeatedly promised her that he would recommend her for transfer but that after a party where he pulled her into a compromising position for a picture, he told her that after talking to others in the department, they said she had a bad attitude. See id. The plaintiff also testified that a co-worker confirmed that her supervisor blocked her transfer, had made negative comments about her to others, and had a romantic interest in her. See id. We conclude that the inquiry into whether a plaintiff has shown that a rejection of certain conduct was “used as the basis for employment decisions affecting such individual” should not be constrained; rather, the court can consider circumstantial evidence and draw inferences in favor of the non-moving party in reaching this determination on summary judgment. C. Thus, in cases where a plaintiff must illustrate a “causal link” for purposes of establishing retaliation, or show that certain conduct was “used” as a basis for *284 employment decisions, a plaintiff may rely upon a broad array of evidence to do so. We will now review Farrell’s evidence, which we find, when considered as a whole, and reviewed in the light most favorable to Farrell, adequately establishes the necessary connection to substantiate both her prima facie of retaliation and her prima facie case of quid pro quo sexual harassment. 9 As a preliminary matter, we disagree with the District Court’s rejection of certain evidence regarding DeLong’s reaction to Farrell’s rejection of his advance on the flight to Puerto Rico. 10 In her certification in opposition to Planters’ motion for summary judgment, Farrell explained that after she rejected DeLong’s advance: DeLong’s demeanor changed, he turned to face away from me, curled up and either slept or pretended to sleep. We had little or no further conversation on the flight, and I worked alone for the rest of the flight. The following day DeLong abruptly left Puerto Rico and returned to Winston-Salem without me. The District Court, citing Martin v. Merrell Dow Pharmaceuticals, Inc., 851 F.2d 703 (3d Cir.1988), refused to permit Farrell to rely on this proof that DeLong’s demeanor changed, because the Court found that it impermissibly contradicted her deposition testimony in which she made no reference to this conduct. See Farrell, 22 F.Supp.2d at 381 n. 22. We disagree with the District Court’s rejection of this evidence because Farrell’s deposition testimony and her certification are not contradictory in the sense prohibited by Martin. In Martin, the plaintiffs certification included a different answer to the same question asked directly in her deposition. See Martin, 851 F.2d at 704-06. In this case, in Farrell’s deposition, counsel for Planters specifically asked Farrell to only describe the content of conversations Farrell and DeLong had on the airplane. Farrell was never asked how De-Long responded to her rejection, what happened after she rebuffed him or how many conversations they had thereafter. The District Court concluded, and Planters argues here, that Farrell implied in her deposition that she and DeLong had conversations after his advance and therefore the substance of the two statements are contradictory. See Farrell, 22 F.Supp.2d at 381 n. 22 (concluding that Farrell’s deposition said that she and DeLong continued to talk but her certification said “he turned away from her and refused to talk to her”). However, Farrell’s certification says that they “had little or no further conversation” after the advance. When asked to relate the conversations in her deposition, Farrell first explained that she was unable to relate the conversations in chronological order, agreed that they were merged together in her mind, and then tried to relate them as best as she could. Since Farrell was not asked about De-Long’s attitude or demeanor, and the District Court’s conclusion rests upon an inference drawn from a narrow reading of the deposition and a broad reading of her certification, we disagree with the District Court and conclude that Farrell’s later certification should be considered as evidence. The District Court also refused to place any significance on DeLong’s decision to return from Puerto Rico a day early when considering whether Farrell had established evidence of hostility. The District Court reasoned that Farrell had presented no evidence that his decision was related to any issue other than Jungmann’s possible termination. See id. at 388 (reviewing Farrell’s quid pro quo *285 claim). However, Farrell faces no separate burden to substantiate the inference she draws from this decision; rather, the district court is to draw inferences in her favor at this procedural stage. See Iadimarco v. Runyon, 190 F.3d 151, 164 (Sd Cir.1999) (applying traditional summary judgment standards to evidence supporting the plaintiffs prima facie case of reverse discrimination). Farrell is entitled to rely upon this event as circumstantial evidence. Farrell, obviously, places great significance upon the relatively close timing between her rejection of DeLong’s advance and her termination. The District Court found the timing to be suggestive in its analysis of Farrell’s retaliation claim but did not find it sufficient on its own. See Farrell, 22 F.Supp.2d at 393 (finding the timing not to be “unusually suggestive”). We view the timing of Farrell’s termination as suggestive for both of Farrell’s claims. The timing evidence is also enhanced by the occurrence of two other events. Although DeLong states that he recommended Bryan over Farrell based upon meetings with other members of Planters’ management, and Planters points to complaints raised about Farrell’s performance during her employment, his decision to terminate her came only three or four weeks after DeLong praised Farrell and asked her about her interest in a promotion. Further, although Planters justifies Farrell's termination in part because of economic concerns and management discussions that took place in November 1994, her termination occurred less than two-weeks after Planters purchased her house in Maryland and moved all of her possessions to Winston-Salem. Farrell also augments evidence of De-Long’s changed demeanor and the suggestive timing of her termination with evidence in the record revealing inconsistencies in Planters’ explanation for terminating her. Farrell challenges both Planters’ explanation that the decision to consolidate the departments was determined by management and its conclusion that the choice to retain Bryan was based upon interpersonal reasons. Planters states that discussions about consolidating the two departments began in November of 1994 and suggests that the decision was made by upper management. However, DeLong’s December 8 memorandum states that complaints he received about Farrell on December 6 and 7 motivated DeLong to see Ecken-roth where they “discussed the option of eliminating the position of Senior Manager Packaging Development and combining Graphics and Packaging as per proposal dated 8/20/93.” Portions of De-Long’s December 8 memorandum also place the decision to terminate Farrell solely in DeLong’s hands and focus entirely on the alleged interpersonal conflicts as the reason for Farrell’s dismissal. For example, DeLong wrote: ‘When all the issues with her peers were discussed and other feedback received discussed, it was clear that I had to deal with Susan” and explained that “Based on the discussions and conversations with people over the last several months who found Susan very difficult to work with ... I made the decision to eliminate her job and combine Graphics and Packaging under Peggy as per earlier recommendation and hire another Packaging Engineer.” Farrell also alleges that on the day Eckenroth and DeLong informed her of the decision to terminate her, Eckenroth told her that Planters would call the decision a position elimination for her benefit but that she was actually being terminated for interpersonal reasons. Farrell also adduces evidence surrounding the choice of Bryan over her to accompany the inference that DeLong made the decision on the basis of impermissible reasons. Although DeLong named both Lyons and Jabbour at the December 13 meeting as managers who made negative comments about her, Farrell alleges that neither Lyons nor Jabbour made such *286 comments and supports her argument with reference to DeLong’s memorandum that expressly states Lyons found Farrell to be helpful. Furthermore, Farrell states that Lyons indicated to her that he told De-Long that DeLong seemed to have a personal problem with Farrell from the manner in which he solicited feedback about Farrell. Planters argues that none of these points help to raise the required inference. It urges that we should draw no inference from the timing of DeLong’s suggestion of a promotion because Farrell was not terminated because she was incompetent, but because of a required consolidation and the determination that Bryan was a better choice than Farrell. Further, Planters points out that DeLong explained in his deposition that he discounted complaints about Farrell until late November and December because Farrell was new. Planters argues that nothing should be read into the trip to Puerto Rico because the trip was planned for DeLong before he asked her to join him and Farrell had been required to tour other facilities with De-Long. Planters disagrees that Farrell establishes any inconsistencies, arguing that the memorandum supports the view that DeLong responded to his meeting with Eckenroth by interviewing other managers and by subsequently choosing Bryan. We recognize that different inferences might be drawn from the evidence presented in the record. On summary judgment, however, when viewing the sufficiency of the prima facie case, our role is not to act as factfinder. Instead, we must consider the evidence taken in the light most favorable to the non-movant and determine whether Farrell can show the eausatioh required for a prima facie case of retaliation and quid pro quo harassment. We believe that, taken as a whole, the behavior of DeLong, the timing of Farrell’s termination and the inconsistencies she raised in Planters’ explanation for her termination are sufficient to create the required inference. We recognize that by acknowledging that evidence in the causal chain can include more than demonstrative acts of antagonism or acts actually reflecting animus, we may possibly conflate the test for causation under the prima facie case with that for pretext. But perhaps that is inherent in the nature of the two questions being asked — which are quite similar. The question: “Did her firing result from her rejection of his advance?” is not easily distinguishable from the question: “Was the explanation given for her firing the real reason?” Both should permit permissible inferences to be drawn in order to be answered. As our cases have recognized, almost in passing, evidence supporting the prima facie case is often helpful in the pretext stage and nothing about the McDonnell Douglas formula requires us to ration the evidence between one stage or the other. See Iadimarco, 190 F.3d at 166 (explicitly referring to the evidence of the prima facie case in finding evidence supporting pretext); Jalil, 873 F.2d at 709 n. 6 (“Although this fact is important in establishing plaintiff ‘s prima facie case, there is nothing preventing it from also being used to rebut the defendant’s proffered explanation. As we have observed before, the McDonnell Douglas formula does not compartmentalize the evidence so as to limits its use only one phase of the case.”) (internal quotation omitted) (citing and quoting Dillon v. Coles, 746 F.2d 998, 1003 (3d Cir.1984)). It is enough to note that we will not limit the kinds of evidence that can be probative of a causal link any more than the courts have limited the type of evidence that can be used to demonstrate pretext. 11 *287 III. Farrell also appeals the District Court's grant of summary judgment to Planters on her North Carolina state contract law claim. Farrell argues that her move to North Carolina was consideration for promises that she would be fired only for cause and, thus, these promises became an implied term of her employment which Planters allegedly breached by firing her for retaliatory or discriminatory reasons. Farrell acknowledges that the North Carolina Supreme Court recently rejected the precise theory on which she is proceeding. See Kurtzman v. Applied Analytical Indust., Inc., 347 N.C. 329, 493 S.E.2d 420, 423-24 (1997) (holding that there is not a “moving residences” exception to the general rule of at-will employment). However, Farrell argues that Kurtzman should not be applied to her case because her implied contract right vested, and the agreement was breached, before Kurtzman was decided. Farrell explains that the protection of contract rights based upon prior law is a compelling reason under North Carolina law counseling against the retrospective application of Kurtzman to bar* her claim. The District Court had difficulty accepting this argument, as do we. We will affirm. In Kurtzman, the North Carolina Supreme Court held “that plaintiff-employee’s change of residence in the wake of defendant-employer’s statements here does not constitute additional consideration making what is otherwise an at-will employment relationship one that can be terminated by the employer only for cause.” Id. at 423-24. The Court found assurances of job security such as “ ‘If you do your job, you’ll have a job’; ‘This is a long-term growth opportunity for you’; ‘This is a secure position’; and “We’re offering you a career position’ ” insufficient to alter the at-will nature of the employment although the assurances preceded the plaintiffs decision to move from Massachusetts to North Carolina. See id. at 421. In reaching its holding, the Court rejected precedent of the North Carolina’s intermediary court establishing a “moving residence” exception to the at-will doctrine. See id. at 423 (rejecting Sides v. Duke University, 74 N.C.App. 331, 328 S.E.2d 818 (App.), disc. rev. denied, 333 S.E.2d 490 (N.C.1985) and Burkhimer v. Gealy, 39 N.C.App. 450, 250 S.E.2d 678 (1979), disc. rev. denied, 297 N.C. 298, 254 S.E.2d 918 (1979)). The court also dismissed as “background discussion” language in Harris v. Duke Power Co., 319 N.C. 627, 356 S.E.2d 357 (1987), mentioning Sides and the “moving residences” exception. See id. at 423. The Court concluded that the argument that the exception was well es *288 tablished in the court’s jurisprudence was incorrect. See id. Applying its holding to the case before it, where the plaintiff had won a jury verdict, the Court remanded to the trial court with direction to enter a judgment notwithstanding the verdict. See id. at 424. We agree with the District Court that Kurtzman is dispositive despite the fact that Farrell moved to North Carolina before it was decided. The North Carolina Supreme Court has stated that “[ujnder long-established North Carolina law, a decision of a court of supreme jurisdiction overruling a former decision is, as a general rule, retrospective in its operation.” Cox v. Haworth, 304 N.C. 571, 284 S.E.2d 322, 324 (1981). The Court further explained that: “Unless compelling reasons ... exist for limiting the application of the new rule to future cases, we think that the overruling decision should be given retrospective effect.” Id. The Court in Cox also noted that the decision as to whether a case ought to be applied retroactively should include consideration of policy issues as well, such as “reliance on the prior decision, the degree to which the purpose behind the new decision can be achieved solely through prospective application, and the effect of retroactive application on the administration of justice.” Id. at 324. Farrell, citing MacDonald v. Univ. of North Carolina, 299 N.C. 457, 263 S.E.2d 578 (1980), argues that protecting previously vested contract rights is a compelling reason which requires limiting the application of the rule announced in Kurtzman to cases involving promises made after it was decided. In MacDonald, the North Carolina Supreme Court explained: When the law has received a given construction by a court of last resort, and contracts have been made and rights acquired under and in accord with such construction, such contracts may not be invalidated nor vested rights under them impaired by a change of construction made by a subsequent decision. MacDonald, 263 S.E.2d at 581. However, as the District Court recognized, the facts of Farrell’s ease fail to support the exception described by MacDonald. Farrell - would have us disregard the point made in Kurtzman that the “moving residence” exception to the at-will doctrine had never been established by the North Carolina Supreme Court. See Kurtzman, 493 S.E.2d at 423 (“Plaintiffs contention that this exception is well established in our jurisprudence is incorrect. This Court has not heretofore expressly passed upon it.”). It is telling that the North Carolina Supreme Court in Kurtzman did apply its rule to the case before it, remanding for a judgment notwithstanding the verdict to be entered. We find no basis for not following its lead in this regard. We also believe that policy considerations point to the application of Kurtz-man to bar claims invoking contracts allegedly made before it was decided. In Kurtzman, the Court focused upon at-will employment as a fundamental precept of law noting that exceptions to the rule should not be found to exist without “substantial justification grounded in compelling considerations of public policy.” Id. at 423. The Court explained that allowing a “moving residences” exception, in a mobile society would create instability in an otherwise stable area of employment law. See id. at 423-24. The strength of these policy rationales suggests that the North Carolina court would not create a subset of employees who could avoid the holding in Kurtzman, and bring a claim based upon the “moving residences” exception solely because the plaintiff moved before Kurtz-man was decided. IV. Accordingly, we will reverse the District Court’s order granting summary judgment in favor of Planters on Farrell’s federal claims, and remand for further proceedings. We will, however, affirm the District *289 Court’s ruling precluding Farrell’s state law breach of contract claim. Footnotes [1] Planters Lifesavers Company was an operating company of Nabisco, Inc. Subsequently, Planters Lifesavers Company divided into Planters Company and Lifesavers Company. Both are operating companies of Nabisco, Inc. For the sake of simplicity, we will refer to the defendants collectively as “Planters.” [2] Farrell does not appeal the District Court’s decision granting summary judgment in favor *275 of Planters on her wage discrimination claim. See Farrell, 22 F.Supp.2d at 388-90. [3] The District Court refused to consider certain evidence regarding DeLong's change in attitude after Farrell rejected his advance because it contradicted her deposition testimony. See Farrell, 22 F.Supp.2d at 381 n. 22 (citing Martin v. Menell Dow Pharmaceuticals, Inc., 851 F.2d 703, 705-06 (3d Cir.1988)). As we explain, see infra section II. C, we disagree and will consider the allegation as part of the record in front of us. [4] We note that the District Court held that the rejection of a sexual advance was a protected activity, see Farrell, 22 F.Supp.2d at 392, and that determination has not been questioned on appeal. Therefore, we do not need to address it. Farrell's termination clearly establishes the second prong. [5] We do note that our pronouncements regarding temporal proximity and causation need to be assessed with the understanding that the relative evidentiary impact of temporal evidence may vary depending upon the stage of the McDonnell Douglas proof analysis, and the procedural circumstance. We caution, therefore, that each case must be considered with a careful eye to the specific facts and circumstances encountered. See Kachmar, 109 F.3d at 178 (“Our cases set no *280 parameters but were decided in the context of the particular circumstances before us.”). There is clearly a difference between two days and nineteen months. Compare Jalil, 873 F.2d 701, 708 (3d Cir.1989), with Krouse, 126 F.3d at 503. There is also a difference between a plaintiff relying upon temporal proximity to satisfy her prima facie case for the purpose of summary judgment, see Jalil, 873 F.2d at 708, and to reverse a verdict. See Quiroga v. Hasbro Inc., 934 F.2d 497, 500, 501-02 (3d Cir.1991). [6] In its brief, the EEOC’s principal argument is that the temporal relationship between the rejection of DeLong’s advance and Farrell’s termination is sufficient on its own to demonstrate the inference of causation Farrell needs to establish a prima facie case at this procedural stage. We do not need to decide this issue. Following argument, the EEOC brought to our attention a recent opinion of the Court of Appeals for the Eleventh Circuit, Farley v. Nationwide Mutual Ins., Co., 197 F.3d 1322 (11th Cir.1999). In Farley, the Eleventh Circuit found "that a period of seven weeks was sufficient to establish the required causal connection. See id. We note, however, that the Eleventh Circuit in Farley seems to apply a less stringent test for causation in general, see id. ("To prove a causal connection, we require a plaintiff only to demonstrate 'that the protected activity and the adverse action were not wholly unrelated.’ ”), and, thus, its ruling is inapposite. [7] The District Court also cited Bonenberger v. Plymouth Township, 132 F.3d 20, 22, 28 (3d *283 Cir.1997) and Kidwell v. Sheetz, Inc., 982 F.Supp. 1177, 1179-81 (W.D.Va.1997) in a footnote. See Farrell, 22 F.Supp.2d at 388 n. 32. In Bonenberger, we did state, as the District Court contended, that the plaintiffs supervisor “did not suggest, either by word or action, that sexual favors were the price for keeping her job.” Farrell, 22 F.Supp.2d at 388 n. 32 (quoting Bonenberger, 132 F.3d at 28). However, we read this statement in Bo-nenberger as part of the court’s analysis of a subsection (1) claim that the plaintiff's supervisor made a "veiled threat to have her fired for rejecting his sexual advances." See Bonenberger, 132 F.3d at 27. In Robinson v. City of Pittsburgh, we said that a plaintiff could allege a quid pro quo sexual harassment claim when a supervisor implicitly or explicitly made a coercive request for sexual favors even without a subsequent effect on the terms and conditions of employment. See Robinson, 120 F.3d at 1297. Supreme Court precedent has, however, eroded that cause of action since our opinions in Robinson and Bonenberger. See Hurley v. Atlantic City Police Department, 174 F.3d 95, 120 (3d Cir.1999) (discussing Burlington Industries, Inc. v. Ellerth, 524 U.S. 742; 118 S.Ct. 2257, 141 L.Ed.2d 633 (1998) and Faragher v. City of Boca Raton, 524 U.S. 775, 118 S.Ct. 2275, 141 L.Ed.2d 662 (1998)). In Kidwell, a case relying upon the law of the Court of Appeals for the Fourth Circuit, the district court dismissed the claim because the plaintiff failed to establish the second prong of the prima facie case which the court found required evidence of some form of threat. See Kidwell, 982 F.Supp. at 1180. [8] A careful reading of Robinson makes it clear that the events substantiating the plaintiff’s quid pro quo claim which we found sufficient to go to a jury were distinct from those we found insufficient to establish her claim of retaliation. Robinson’s quid pro quo claim involved the alleged block of her transfer request in 1993. See id. at 1298-99. Robinson’s retaliation claim revolved around a complaint she filed with the EEOC in May of 1994. See id. at 1292, 1301-02. Robinson did not plead the facts of her quid pro quo allegations in the alternative as a retaliation claim, as Farrell has done here. [9] We note that Planters suggests the entire analysis of both the quid pro quo and retaliations claims should be joined together because of the similarity of the claims. See Appellee’s Br. at 24 n. 3. [10] Planters points out in its brief that Farrell’s affidavit accompanying her EEOC complaint did not state that DeLong put his hand on her knee. We also place no evidentiary consequence on this omission. [11] Since. Robinson, which described the quid pro quo cause of action, we have not had occasion to focus on the extent to which the McDonnell Douglas burden shifting test applies to, or has implications for the elements of the prima facie case of, quid pro quo claims of this type. See Hurley, 174 F.3d at 120-22. We see no reason to explore these issues in this case. Neither party questions the District Court's standard or the applica *287 tion of the burden shifting structure. See Appellant’s Br. at 28-29 (adopting the District Court’s standard and citing Kauffman v. Allied Signal, Inc., 970 F.2d 178, 186 (6th Cir.1992)); Appellee’s Br. at 25-26 (adopting the District Court's standard and citing Perkovich v. Roadway Express, Inc., 1997 U.S.App. LEXIS 1155, *9-10 (6th Cir. Jan. 22, 1997), and Bonenberger v. Plymouth Township, No. CIV. A.96-403, 1996 WL 729034, at *22 (E.D.Pa. Dec. 18, 1996), aff'd in part and rav’d in part on other grounds, 132 F.3d 20 (3d Cir.1997)); EEOC’s Br. at 26 (arguing that a plaintiff need not show intervening hostility to prove quid pro quo harassment and applying the McDonnell Douglas test to the claim). Farrell only contends that the District Court erred in finding that she had not set forth enough evidence to establish the threshold causal relationship. In two cases decided before Robinson, we considered claims brought by plaintiffs under the same Title VII section alleging that they were terminated because they rejected a supervisor’s explicitly coercive request for sexual favors. See Craig v. Y & Y Snacks, 721 F.2d 77, 78-80 (3d Cir.1983); Tomkins v. Public Serv. Elect. & Gas Co., 568 F.2d 1044, 1047-49 (3d Cir.1977); see also Bonenberger, 132 F.3d at 28 (citing Craig and Tomkins as quid pro quo precedent). Neither case detailed the elements of a prima facie test. However, in Craig we noted that the district court found causation as part of the prima facie case and applied the McDonnell Douglas paradigm in affirming the trial court's verdict for the plaintiff. See Craig, 721 F.2d at 79-80.