OPINION OF THE COURT
GARTH, Circuit Judge.
This case, involving the Pittsburgh Press Club’s action for an income tax refund, comes before us for the second time. Because crucial findings of fact made by the district court were based on inadmissible evidence, we must once again reverse.
I
The Pittsburgh Press Club (PPC or the Club), organized in 1885 under the laws of Pennsylvania, and reactivated in 1955, ap
plied in 1956 for a federal income tax exemption as a social club. In 1959 the Internal Revenue Service (IRS) granted the exemption pursuant to IRC § 501(c)(7), which exempts from federal income taxation:
Clubs organized and operated exclusively for pleasure, recreation, and other nonprofitable purposes, no part of the net earnings of which inures to the benefit of any private shareholder.
In its application for tax exemption, the Club stated that its purpose was:
To provide a professional and social meeting place for those engaged in news and advertising work on newspapers, magazines, radio and television and in public relations activities for the furtherance of common interests.
In 1964 IRS examined PPC’s books and records for fiscal year 1962, and determined that “outside” groups (i.e., nonmember groups) had been permitted to use the Club’s facilities twelve times during that fiscal year. IRS did not then challenge PPC’s exempt status, but warned the Club’s officers that continued use by outside groups jeopardized PPC’s exemption. PPC responded by a letter stating that nonmember use of its facilities would be discontinued.
In 1970 an IRS agent conducted an extensive audit of the Club for fiscal years 1967, 1968 and 1969. He determined that during the period of these three years more than 800 functions had been held by nonmember outside groups, although each such group had been member sponsored.
Based on these findings the agent further determined the amount of “outside” revenues for the three years in question, as well as the percentage those revenues were of the Club’s gross receipts (defined to exclude receipts from initiation fees). These determinations were:
Fiscal Outside Percentage of
Year Revenue Gross Receipts
1967 $ 62,311 11.18%
1968 101,472 16.31%
1969 118,130 17.43%
On February 1, 1972, IRS revoked the Club’s exempt status as of fiscal year 1967, and assessed income taxes for fiscal years 1967 through 1971. The taxes so assessed were paid under protest. IRS’s revocation of PPC’s exempt status was based on two factors. First, observing that voting members paid lower dues than the other classes of members (associate or affiliate members), IRS concluded that the differential in dues constituted an inurement of net earnings to the benefit of the voting members, who were considered by IRS to be private shareholders. IRS charged that this practice violated IRC § 501(c)(7).
See generally
Rev.Rul. 70-18, 1970-1 C.B. 133. Second, IRS claimed that PPC’s continued practice of permitting its facilities to be used for outside affairs, resulting in PPC’s receiving substantial revenues from these outside sources, constituted engaging in business. Thus, it charged that the Club was not
being operated exclusively for nonprofit purposes, and was therefore in violation of IRC § 501(c)(7).
See generally
Treas.Reg. § 1501(c)(7)-l(b); Rev.Rul. 60-324, 1960-2 C.B. 173.
PPC brought suit in the district court for a refund of the taxes which it had paid. IRS counterclaimed for interest which had accrued prior to payment. The district court ruled in the Club’s favor on both issues. 388 F.Supp. 1269 (W.D.Pa.1975).
On appeal (at No. 75-1757) this court upheld the district court’s ruling that the differential dues scale did not violate the “inurement of net earnings” provision of IRC § 501(c)(7). 536 F.2d 572, 574 (3d Cir. 1976). With respect to the nonmember revenue issue, however, this court remanded for more specific findings,
“including the percentage of gross receipts from nonmembers disclosed by the record.”
Id.
at 575 (footnote omitted).
The court also directed that the district court make findings respecting other relevant factors, such as the amount of “outside” business, net profits derived from nonmembers, the purpose for which PPC’s facilities were made available to nonmember groups, and the frequency of such use.
Id.
On remand the district court (426 F.Supp. 553 (W.D.Pa.1977)) reopened the record over IRS’s objection to allow PPC to present additional evidence. The Club’s new evidence consisted principally of a survey (the “second survey”) and of expert testimony by an economist (Dr. Kenkel) and by PPC’s accountant. The second “survey” was comprised of mailings to the member-sponsors of 690 of the 815 challenged affairs (the addresses of 125 member-sponsors could not be located). This survey resulted in 281 usable responses. Based on extrapolations from these responses, Dr. Kenkel concluded that outside revenue for the three audited years was between $41,502 and $83,004, or two to four percent of total income. Using these figures, the Club’s accountants then determined that the net profit attributable to “outside business” was $150 in fiscal 1967 and $72 in fiscal 1968.
The accountants also concluded that no profit was realized for the following three fiscal years because total banquet operations had resulted in a net loss to the Club for those years.
Based on the new exhibits and testimony, the district court again found in favor of PPC. Its specific findings of fact were as follows:
(1) the amount of outside business is not more than $83,004;
(2) the percentage of outside business is not more than 4%, and may be nearer 2%;
(3) the amount of net profits from outside business is negligible or de minimis;
(4) the purpose of events at which the club facilities are used by outsiders are principally social events such as wedding or bar mitzvah receptions, and business-
oriented gatherings for sales presentations and the like, in connection with the trade, profession, or business of the sponsoring member;
(5) the frequency of such events is no more than normal in a club of plaintiffs size, 3550 banquet department events having been held in the years 1967-1971 (PX C — 1), of which only 815 were questioned by the IRS, and of these not over 245 were found to be outside business.
426 F.Supp. at 555-56. This appeal by IRS followed.
II
A
IRS initially objected to the district court’s decision which permitted PPC to introduce new evidence after remand. On appeal, IRS at first contended that when this court remanded to the district court for more specific findings, we had precluded the reopening of the record or the adducing of additional evidence by our direction that “[t]he district court should make more specific findings on remand, including the percentages of gross receipts from nonmembers
disclosed by the record,"
536 F.2d at 575 (emphasis added). IRS also singled out our directive that “[o]n remand, the district court should make more specific findings on all relevant factors, including findings on the amount of ‘outside business’ of the taxpayer
revealed in the record
. . . ,”
id.
at 576 (emphasis added). This language, according to IRS, indicated a narrow mandate requiring that the district court refer only to the then existing record in making its new findings. IRS argued that in reopening the record the district court exceeded the scope of our mandate. We do not agree.
This court’s language in the decision at No. 75-1757 cannot be read to have the restrictive effect claimed by IRS. Whether a trial court will reopen the proceedings on remand is, of course, a matter for its sound discretion. In
Rochez Bros., Inc. v. Rhoades,
527 F.2d 891 (3d Cir. 1975),
cert. denied,
425 U.S. 993, 96 S.Ct. 2205, 48 L.Ed.2d 817 (1976), this court enumerated the criteria with which a district court should be concerned in the exercise of that discretion. We noted that the district court should consider several factors: (1) what burden that would be placed on the parties and witnesses by additional proceedings; (2) what “undue prejudice” may result by not taking new testimony; and (3) whether judicial time and resources would be unnecessarily drained. 527 F.2d at 894, n.6.
In the case
sub judice,
IRS pointed to no specific burdens placed upon it or its witnesses as a result of reopening the record at the second trial. At the same time, the prejudice to PPC had not been permitted to adduce further evidence was apparent, since PPC had the burden of demonstrating its entitlement to an exemption. Moreover, the two days of trial time required to receive the new evidence cannot be said to have significantly or unnecessarily drawn upon judicial resources, at least where the additional testimony and exhibits had the potential of aiding in the resolution of difficult questions. We conclude, therefore, that the district court did not abuse its discretion by reopening the proceedings.
B
IRS vigorously asserts that PPC’s second “survey” was nothing more than a compilation of hearsay, none of which was admissible under any exception to the hearsay rule of Fed.R.Evid. 802. We agree.
After our remand, the Club decided to undertake a new survey of those affairs which had been classified as “outside busi
ness” by IRS. The survey was prepared and taken with the assistance of Dr. James L. Kenkel, an economist and statistician, as well as with the assistance of the Club’s attorneys and various of its employees. PPC mailed a questionnaire and covering letter (the entire text of which is set forth in the margin
) to the sponsors of 690 of
the questioned functions.
The survey produced 281 usable responses, which Dr. Kenkel believed were reliable. Based upon these responses, he classified the particular affairs (all of which had been classified as outside by the IRS) as “clearly outside”, “clearly inside”, or “doubtful”. Extrapolating from this data to the entire universe of 815 events, he concluded that the Club’s
maximum
outside revenue (i.e., including “doubtful” events) for the fiscal years 1967 through 1969 was $83,004 (as opposed to IRS’s figure of $276,683.51), and that the Club’s
minimum
outside income
(i.e.,
excluding “doubtful” events) for that period was $41,502. Comparing these estimates to PPC’s total income for the three fiscal years in question, Dr. Kenkel determined that the Club’s outside income for the three years was only two to four percent of its gross receipts (as contrasted with the 11% to 17% figures computed by IRS).
PPC’s accountants next calculated the Club’s net profit or loss from its total banquet business
for the tax years 1967 to 1971. They then applied Dr. Kenkel’s percentages to derive net profit attributable to outside banquet business. They concluded that there was no net profit on outside business for fiscal years 1969 through 1971, since they had determined that in those three years the Club’s total banquet business had produced a net loss. As for fiscal years 1967 and 1968, they calculated net profits on outside business as $150 and $72, respectively.
It is clear that PPC’s second survey, which the district court admitted into evidence, was hearsay. The survey was no more than a summary and distillation of 281 extrajudicial declarations offered to prove the truth of the matters asserted in those declarations,
viz,
whether the banquets challenged by the IRS were “inside” or “outside” affairs. While some polls are not hearsay in that they are admitted to show the beliefs or attitudes of the respondents (as opposed to the truth of the respondents’ answers), other polls are unquestionably hearsay.
Such hearsay polls are not necessarily inadmissible as evidence in that they may fall into a recognized class exception
to the hearsay rule, as, for example, when they involve a present sense impression
or a presently existing state of mind.
See, e. g., Zippo Mfg. Co. v. Rogers Imports, Inc.,
216 F. Supp. 670, 680-86 (S.D.N.Y. 1963) (Feinberg, J.) (unfair competition case, in which a survey was admitted; survey respondents were shown a cigarette lighter and were asked which firm had manufactured it). If a survey or poll does not fit
into one of the class exceptions to the hearsay rule, it may nevertheless be admissible under Fed. R. Evid. 803(24).
In other words, the survey is admissible if it is material; if it is more probative on the issue than any other evidence; and if it has “circumstantial guarantees of trustworthiness” equivalent to those of the class exceptions (/. e., if the “hearsay dangers” — principally, faulty memory and lack of sincerity — are greatly reduced, see, e.
g., Zippo Mfg. Co. v. Rogers Imports,
supra). The proponent of such evidence, of course, has the burden of establishing these elements of admissibility.
In the context of polls and surveys, the circumstantial guarantees of trustworthiness are for the most part satisfied if the poll is conducted in accordance with generally accepted survey principles, and if the results are used in a statistically correct way, since proper survey and statistical methods are intended to assure a poll’s reliability.
See Allen v. Morton,
333 F.Supp. 1088, 1094 (D.D.C.1971), which cautions that surveys, since they involve hearsay, must be conducted with proper safeguards to insure accuracy and reliability. The court in
Allen
gave little weight to a survey because of its technical deficiencies. The Judicial Conference Study Group, in its Handbook of Recommended Procedures for the Trial of Protracted Litigation, 25 F.R.D. 351, 429 (1960), discusses the several factors which must be examined when determining whether a poll meets generally accepted survey principles. A proper universe must be examined and a
representative
sample must be chosen; the persons conducting the survey must be experts; the data must be properly gathered and accurately reported. It is essential that the sample design, the questionnaires and the manner of interviewing meet the standards of objective surveying and statistical techniques. Just as important, the survey must be conducted independently of the attorneys involved in the litigation. The interviewers or sample designers should, of course, be trained, and ideally should be unaware of the purposes of the survey or the litigation.
A fortiori,
the
respondents
should be similarly unaware.
See Berman
v.
New Hampshire Jockey Club,
324 F.Supp. 1156, 1168 (D.N.H.1971).
See generally
Annot.,
Admissibility and Weight of Surveys or Polls,
76 A.L.R.2d 619 (1961).
In
Zippo,
for example, the court admitted into evidence a consumer survey. The survey in that case was scientifically designed. The random sample was carefully chosen to obtain a representative cross-section of the universe. Methods were used which would minimize sampling and procedural errors. The interviewers were all experienced surveyors and were not informed about the litigation or the survey’s purpose. The respondents — who had no interest in the litigation — were likewise uninformed. In such circumstances the court ruled that the survey was trustworthy because there were no substantial “hearsay dangers”. There was no danger of faulty memory because of the nature of the question asked (which firm manufactured a cigarette lighter). There was no substantial danger of insincere responses because of the scientific sampling methods used. The district court judge noted, however, that a poll might well be inadmissible if, for example, the questions were “unfairly worded to suggest answers favorable to the party sponsoring the survey”,
because in such a case the circumstantial guarantees of trustworthiness would be lacking. 216 F.Supp. at 684.
See also Union Carbide Corp. v. Ever-Ready Inc.,
531 F.2d 366 (7th Cir. 1976);
President & Trustees of Colby College v. Colby College-N.H.,
508 F.2d 804 (1st Cir. 1975);
Standard Oil Co. (Indiana) v. Standard Oil Co. (Ohio),
252 F.2d 65 (10th Cir. 1958);
United States v. Healthco, Inc.,
387 F.Supp. 258 (S.D.N.Y.),
aff’d,
535 F.2d 1243 (2d Cir. 1975).
The PPC survey clearly cannot meet the standards specified in the Handbook on Protracted Litigation or in
Zippo.
It was not scientifically designed; indeed, Dr. Kenkel had never before taken a poll. The sample consisting of 281 responses, while large in proportion to the universe of 815 challenged banquets, was not designed to be representative.
The respondents, who were all Club members and thus interested in the litigation, were told the precise nature of the litigation and the purpose of the survey. They consequently knew which responses would be helpful to PPC, and conversely, which would be harmful. Moreover, it was possible that a recipient of the questionnaire would fail to respond because he knew an honest response would be harmful to the Club’s position. Thus the respondents might have contained a higher percentage of those who could answer in a way helpful to the Club.
It therefore appears that PPC’s survey suffers from a severe dearth of any circumstantial guarantees of trustworthiness. Both hearsay dangers — faulty memory and insincerity — loom large in the Club’s poll. The respondents were not being asked for a present impression; rather they were being asked for details about banquets which had taken place many years before. In fact the covering letter, recognizing that the members’ memories might be “dimmed or imperfect,” nevertheless asked for a “best recollection.”
See
note 7
supra.
The respondents were all interested in PPC’s prevailing in the lawsuit. Yet they were expressly advised about the nature of the litigation and the survey, as well as which answers would benefit the Club.
In sum, the survey admitted by the district court was neither objective, scientific nor impartial.
See Berman v. New Hampshire Jockey Club, supra.
As a result of its technical
and non-technical deficiencies, the PPC survey lacked the necessary safeguards of accuracy and reliability — the circumstantial guarantees of trust
worthiness.
See Allen v. Morton,
333 F.Supp. 1088, 1094 (D.D.C.1971). Since the survey was hearsay, and was not admissible under any class exception to the hearsay rule, and since it lacked those guarantees required by Fed.R.Evid. 803(24), we hold that PPC’s second survey was inadmissible, and that the district court erred by receiving it into evidence.
See United States v. Article of Drug, etc.,
415 F.2d 390 (5th Cir. 1969).
The survey — which we have just held to be inadmissible — was the foundation for certain of the plaintiff’s exhibits, and for much of Dr. Kenkel’s testimony. Since this second survey should have been excluded, it follows that the evidence based on the survey should have been excluded as well. In particular, Dr. Kenkel’s determinations as to amount of outside income and the percentage of gross receipts attributable to outside affairs could not have been admitted into evidence once the foundation for that testimony was excluded. For the same reason, the exhibits prepared by PPC’s accountants and the testimony of accountant Di Mario, respecting the Club’s net profits attributable to outside business, should not have been admitted into evidence.
The district court’s findings of fact with respect to the amount of outside business, the gross revenues percentages, and net profits were all premised on the evidence which we have now held to be inadmissible. Since we have held that the survey evidence should have been excluded by the district court, the district court’s findings are no longer supported by the record, and are therefore clearly erroneous.
Ill
Inasmuch as we have determined that the district court erred in admitting the survey evidence, we are obliged once again to remand this case to the district court for additional proceedings.
See
Part IV of this opinion,
infra.
Because new calculations must be made to support any finding respecting net profits, we conclude that it is necessary to furnish the district court with an appropriate standard so as to avoid any possibility that we will once again be faced with an appeal raising this issue.
As we observed previously, PPC’s accountants calculated net profits attributable to the Club’s banquet activities by charging against banquet income an allocation of fixed and variable costs.
The net profits attributed to
outside
banquet income was determined to be two to four percent of total banquet net profits. (The percentage figure used was, of course, that derived from the second survey by Dr. Kenkel.) The record reveals that the costs so charged against banquet income were: cost of goods sold, wages and salaries, rent, depreciation, insurance, sales taxes, employees’ insurance, supplies expenses, utilities, maintenance, and miscellaneous expenses.
In our opinion at No. 75-1757 we directed that the district court on remand make specific findings as to net profits derived from nonmembers. 536 F.2d at 575. In its attempt to comply with our mandate PPC proffered, and the district court accepted, the profit figures calculated in the manner described above. When we directed that “net profits” be examined, however, we assumed that the net profit calculation would be made within the context of the issue being disputed. Although under accounting principles in general, the net profit calculation might be deemed to include all the categories of charges listed above, we are not here concerned with normal methodology of accounting practice.
It is clear in this case that the literal accounting definition of “net profits” is not relevant to the ultimate inquiry of whether an exempt so
cial club is or is not engaged in sufficient nonmember business to justify the revocation of its exemption.
The exemption of IRC § 501(c)(7) extends only to social clubs “supported solely by membership fees, dues, and assessments.” Treas.Reg. § 1.501(c)(7)-l. Congress intended that individuals be permitted to join together in appropriate organizations “to provide recreational or social facilities or other benefits on a mutual basis, without tax consequences . . . .” S.Rep.No. 91-552, 91st Cong., 1st Sess. 71, 1969-3 C.B. 469-70, U.S.Code Cong. & Admin.News 1969, pp. 1645, 2100. Such mutuality requires that the members share the benefits and burdens of the club. Consequently, when one member benefits through the use of the facilities of the club, it is only because all members have been “burdened” with the cost and support of those facilities. In other words, resources are merely shifted between members of such a group, and no tax consequences attach to that shifting.
See McGlotten v. Connally,
338 F.Supp. 448, 458 (D.D.C.1972) (three-judge court).
It is obvious, then, that a club which derives substantial income from nonmember sources has eschewed that mutuality which is the basis for its exemption. In such a situation revenue derived from nonmembers is used to benefit members since the outside revenue permits the club to assess lower dues than would otherwise be required to support the club’s facilities and operations. The club would be able to carry “as large and luxurious a plant as the members might like without the payment of burdensome dues.”
West Side Tennis Club v. Commissioner,
111 F.2d 6 (2d Cir. 1940).
Clearly, in light of the purpose of § 501(c)(7), the relevant inquiry with respect to outside profits is the amount by which members are benefited by the contributions of nonmembers. Put another way, it is necessary to determine by what amount nonmembers may be subsidizing the club’s members.
The amount of this subsidy is the proper measure of “net profits” as this court intended that concept to be employed when it first remanded for a finding as to net profits. 536 F.2d at 575.
Hence, it was undoubtedly proper to charge against outside banquet income the cost of goods sold, since those goods were used in generating that particular income. It was also proper to charge against outside banquet income other costs directly attributable to outside banquets, such as sales taxes, the wages of PPC employees earned while assigned to an outside function, and supplies used specifically for an outside banquet.
It was improper, however, to charge against outside banquet income those costs which were not incurred directly as a result of the outside banquet, but were rather fixed costs which the Club’s members would have to bear in the absence of the nonmember income. We include in this category rent, depreciation, overhead salaries and wages (as contrasted with wages of an employee earned while assigned to a specific outside affair), utilities, maintenance, employees’ insurance, and those taxes and supplies not attributable directly to an outside function. While we may not have listed all such items, the ones noted are representative of those charges which constitute the subsidy, the amount by which nonmembers benefit members. Had there been no outside functions, and therefore no nonmember income, the members of PPC would have been required to absorb these costs themselves, either through higher dues and assessments, or higher restaurant and bar fees charged members. In the alternative they may have opted for reduced services or less “large and luxurious” facilities. In ei
ther event it is the PPC membership which bears the burden.
The district court erred, therefore, by accepting PPC’s “net profits” figures, since those figures included costs which were not directly attributable to the outside functions. Fixed costs and variable costs not incurred directly on account of an outside function cannot be charged against outside banquet income in calculating “profits” if the inquiry to be answered is whether a social club exemption was properly revoked. On remand, the district court should make its finding of “net profits” in a manner consistent with the instructions which we have outlined.
IV
As we have observed, we must once again remand to the district court — albeit reluctantly — so that the findings called for by this court’s first opinion may be made. The district court should, of course, utilize the proper measure of “net profits” (see Part III
supra).
It must also rule on the admissibility of and weight to be given the first survey.
Moreover, the district court should give careful consideration to the burden of proof placed upon the taxpayer to prove its exempt status. In this regard, the taxpayer has an obligation “to maintain records adequate to establish a right to the tax exemption”, 536 F.2d at 576, and in particular to maintain adequate records which clearly show the frequency of nonmember use and the gross receipts derived therefrom, Rev.Proc. 71-17, 1971-1 C.B. 683, § 4 (superseding Rev.Proc. 64-36, 1964 — 2 C.B. 962, § 4);
see
IRC § 6001.
The district court, on remand, shall make findings based on the record of the first trial, and so much of the record of the second trial as has not been held to be inadmissible by this court. It shall not, however,, reopen the record or take additional evidence. PPC has already had two opportunities to produce evidence to satisfy
its burden of proof. Having had two bites from the apple, PPC has had more than most litigants receive. PPC must therefore rest on the evidence adduced in the two trials it has already had.
The judgment of the district court will be reversed and the case remanded for further proceedings consistent with this opinion and with our opinion at No. 75-1757, 536 F.2d 572.