Professional Documents
Culture Documents
Energy Conservation in University Buildings Encouraging and Evaluating Reductions in Occupants' Electricity Use
Energy Conservation in University Buildings Encouraging and Evaluating Reductions in Occupants' Electricity Use
The basic design of the study was 10 contrast the effectiveness of two
ENERGY CONSERVATION IN methods of encouraging conserva tion in matched pairs of buildings and
UNIVERSITY BUILDINGS dormitories. Both methods are used frequently in businesses and other
large organizations to bring about changes in employee practices. The
Encouraging and Evaluating first . a traditional "manaKemem" method. emphasi7.ed organizational
efficiency and use of existing hierarchical lines of communication.
Reductions in Occupants' Persuasive written communications from building authorities (for
Electricity Use e~ample, department heads and deans) urging consen·ation were cen-
tral to this method . In the second "user participation" method. the
building head delegated responsibility for development and implemen-
LOU McCLELLAND and STUART W. COOK tation of energy conservation plans to small groups of occupants. The
lnstitutr of Btha••iorol Sciencr. groups discussed use and waste of energy in their sections of the build ing
Unfrf'Tsity of Colorado and then devised their own conservation plans. This method is deri,·ed
from research on small groups (Bennett. 1955: Coch and French. 19..t8:
119
McClelland, Cook I ENERGY CONSERVATIO:"\ 1~I 122 EVALU,\TION REVIEW/ FEUIWARY 1980
the universiiy's efforts to conserve: and (d) to monitor rrogram events sized the need for conservation and included a page of how-to-save-
in their buildings. Program implementation began February I, 1977. encrgy information. Its purpose was to pince the e~perimcntal conser-
vation programs in the context of a university-wide effort and, in so
Bl'll.DING SEl.ECTION
doing, to mask their "experimental" character.
Immediately after the memo. posters (yellow on black "Turn out
Pair members were matched on several dimensions related lo energy lights and equipment") and labels for light switches were delivered to
use-si1.c, age (and associated construction tyre), number and type of each building head. Thereafter, program implementation and mainte-
occupants, functions (office,_library.13boratory, classrooms)-to maxi- nance were left lo the building heads, although the program staff did
mi7.e similarity in the proportion of electricity use controlled by occu- issue periodic rcminde1s lo them about previously scheduled activities
pants. This was done so that the conservation programs would have and monthly reports on their building's consumption. These reports.
equal "opportunity.. to reduce consumption in pair members. Whereas henceforth called "feedback,';' included expected a·nd actual electricity
the dormitory matches are excellent, those for the classroom buildings use and percentage of savings for the first and last two-week periods of
are not as good. Si1.e, age. and amount of research equipment vary the month.
somewhat within building pairs; furthermore, measurement of conser-
vation .. opportunity" is impossible and may vary with factors other than ACTUAi. rROGHA!\1 orEHATION
those used for matching. The pairs can be characterized as small. older
classroom (Education. Economics): large. new classroom (Business. The programs in all buildings and dormitories 'operated as normal
Law): small. old laboratory (Geology. Biology); and large, new labora- university or departmental activities, not as experiments; this was done
tory (Psychology, Microbiology). Two control buildings-a large. new to maximize the generalirnbility of the results. lkcausc actual impk-
laboratory (Physics) and an older, large classroom (Liberal Arts)-were mentation of the programs-posting of notices, memo distribution,
also monitored. Of the six dormitories, four (A, D, C, and D) are virtu- section meetings, and so on-was administered by building heads and
ally identical in size, construction type. and class mix: two of these house occupants, program details varied considerably across buildings. Inter-
females, two house males . The remaining two dormitories (X and Y) arc views with building and section heads were conducted at the end of the
older, coed and very similar to one another in age, size, and construct ion ter m to determine how and when the prescribed programs had been
tyre. No separately metered dormitories were available to serve a~ a implemented in each building.
control. The four "management" condition buildings represented that meth-
od very well: All communication was from the building head to individ-
ual residents, most communications were written, and little or no discu s-
PROGRAM IMPLEMENTATION sion of the program among building occupants occurred. The amount·
of small group interaction, discussion, and delegation of responsibility
The building heads or their representatives met with the Vice Chan-
varied within the four " user-participation.. condition buildings from
cellor and with program staff early in the fall ( 1976) term. At this time
very low (only one section met, one time) to very high (all sections met
they heard about the university's need to conserve, received information
at least once and developed explicit conservation plans).
about the experiment and the two methods, and were asked 10 partici-
In the dormitories, responsibility for the consen•ation programs fell
pate. Instructions on program implementation were sent to building
primarily to members of student government. Evidently. because of
heads in early December and reviewed with them shortly before the
their enthusiastic support of the need for energy conservation, these
programs began. In the dormitories, elected student government groups
students held group meetings (and sometimes even spoke individually
and resident advisors on each floor helped the administration plan and
with all section occupants about the program) regardless of whether
implement the conservation programs.
they were in the management or user condition . Thus, there were essen-
The programs began February I, 1977. with a memo from the Chan-
tially no program differences among the six dormitories corresponding
cellor to all university faculty, staff, and students. This memo empha-
McClelland, Cook I E:--:ERGY CO:\'SERVATION 12J
a set of points randomly drawn from a normal distribution. Two outliers pared with those from the 10-weclc program period . 1 In addition, the
were found: Microbiology November 15, 2.9 standard deviations below baseline period figures were compared with those from each of three
the mean. and dormitory A January 17, 2.9 standard deviations above subsets ofwceks-1(2/14 through J/ 13), II (3/28through4/17)and Ill
the mean. No explanation for these large de\'iations could be found . The (4/ 18 through 5/ l); these comparisons were by means of a priori or
analyses reported below were all performed with these points excluded. planned contrasts. Outliers within the program period were identilied
A variety of weather measures. including heating degree days. plus first; only Microbiology March 28 (2.6 standard deviations below the
hours of daylight and a cloudiness rating were tested in the regression mean) was climinateu . The results arc displayed in Table I; consump·
analyses. Only statistically significant predictors (p<.05) were allowed tion for each period is shown as percentage of expected consumption.
to enter the stepwise analyses. Electricity consumption rclntes signifi- Consumption was significantly below expected levels during the
cantly to weather (percent days over 60°) in only three of the 10 class- program in all experimental buildings and dormitories except Biology,
room buildings: Business (R? =. 52). Law (R? =.89). and Microbiology where new air conditioning equipment was installed near the beginning
(R? =.58). In addition, signilieantly more electricity is used in the dorm- of the conservation program. However, small but statistically signifi·
itories A-Don days over 70° (R? =.61 10 .82). For these seven buildings cant reductions also occurred in Liberal Arts and Physics. the control
weekly consumption figures were "corrected" to a common basis of zero buildings.
da\'S over 600 or 70°. For no building arc hours of daylight or cloudiness
rai°ings signilicantly rebted to consumption. Comparison ll'ith controls. Direct comparisons across buildings of
Engineers from the university's physical plant d~par.tment wer_e c_on- the deviation of actual consumption from expected are meaningless: A
sulted regarding the p:lltern of electricity use over time m each build mg. drop of 906 Kwh a day ·in Microbiology is only 1% of consumption.
outliers. and the results of the regression analyses. They were able to whereas a drop of 102 Kwh in Geology is 20%! Although a percentage
account for some outliers (for example, a week when the fan in Liberal reduction measure is not without its own interprctational problems (if
Arts was broken) and to confirm that the prediction equations were as occupants of Microbiology can only control 5% of that building's con-
expected given the nature of the equipm~nt i~ t~e buildings_. . sumption, but those in Geology can control 15%. Geology has the ad-
Week-to-week variations in consumption w1thm the baseline pcnod vantage in such a comparison) this measure was used to compare reduc-
are quite small after the effects of weather have been removed. Coeffi- tions in experimental and control buildings and within building pairs.
cients of variation for all buildings except Geology are less than 5%. Weekly percentage reduction figures (the weather-adjusted deviation
However, there is immense variation in consumption le_vels across build- of actual from e x pected consumption, divided by expected consump-
ings. Education, Economics, Geology, Biology. and Liberal Ar~s all use tion) were subjected to analyses of variance for each experimental-
under t 000 K wh per day; Business, Law, Psychology. and Phys1_c s 3000- control pair. In no case (except Biology) is the control's reduction
4500; and Microbiology 9064 Kwh per day, probabl_y due to its larg_c greater; for every building except Microbiology the total reduction is
store of experimental equipment. Dormitories A-D. with ab~ut 255 rcs~ signilicantly greater than that of either control.
dcnts each. use an averuge of 1150 K wh per day; X and Y. w11h 420 resi-
dents each. use about 1700 K wh per day. /ntrapair comparisons. The same procedure used to compare reduc-
tions in experimental and control buildings wa5 used to compare user
Did conwmption drop? If the programs had no effect, c_onsumption and management buildings. In the three (testable) building pairs no
during the program period should average the same as .d~nng the ba se- consistent superiority of either method is apparent-of the eight com-
line period (barring nonprogram· effects on consumption). In oth~r parisons for these three pairs , seven show no significant difference.
words. expected average corrected consumption during the r.rogram is 13ccause the conditions were not actually implemented differently in the
the same as that before the program . One-way analyse~ o_fvanance test- dormitories. no user-management comparisons were made for them.
ing this (null) hypothesis were pcrformcll for cac_h building separately:
the weather-adjusted weekly b~seline consumption ligurcs were com-
J McClelland, Coolc I ENERGY CONSERVATION 127 12R EVALUATION REVIEW f FEBRUARY 191!0
I TABLE 1
Electricity Use as Percentage of Expected Consumption
Period
experimental buildings and in all six dormitories. but that neither the
user nor management method was more effective in doing so. Savings in
the seven e xperimental buildings and six dormitories totaled over
200.000 Kwh with a value of over SJ500.
Co11dirio11 I II /fl Toto/
baths in all six dormitories; 10 of the 12 drops were statistically signifi- tion programs: prediction problems, interpretation difficulties, and the
cant. effects of initial waste levels.
several buildings. Second, energy-saving actions (primarily turning off The preceding arguments arc not meant to suggest that energy con-
lights) were made part of the job responsibilities of individuals in several servation programs cannot or should not be evaluated . Rather. the need
buildings. These might be called policy changes. Third, some building for carefully planned evaluations. with ample lead time prior lo pro-
occupants voluntarily made energy-saving changes in their daily activi- gram implementation and ample resources for collection of more than
ties, such as turning off office and restroom lights. All building occu- consumption data, must be acknowledged.
pants had to accept the physical and policy changes (or their results) in
their buildings; however, such changes required no attention from or
maintenance by most building occupants. Estimating how much each
type of activity contributed to the energy reductions that did occur is
impossible; in particular, it is possible that the changes in daily energy-
using behaviors at which the programs were primarily aimed contri- NOTES
buted insignificantly to reduced consumption. If this is so. then the
I. Baseline consumption figures for e:ich building were subjected 10 time-ser:es an:11l-
programs' effectiveness might not transfer to settings in which physica l
ysis with pro!!ram CORREL (Gl:iss et :ii., 1975). These :anal~·ses sho"'c:d a_ll s~ri_es 1~ be
and policy changes could not be made or already had been made. A RI MA (0. o. 0) o r .. white: noise"; in othrr words. the successive weeks :ire 1ndu11ngu1th·
ahle from indepc:ndc:nl c:iscs. thereby meeting the cond itions for A~OVA.
Initial wasre levels. The criterion for program cffectiwness must be 2. Fullow-up data for the classroom -office buildings wc:r( not collected b~causc _ '''
ehangc-<lid reductions in energy consumption from prcprogram le vels prohibiti ve costs. In these buildings the ckctric meters arc interior. :ind ~mHr~11y ro_l_icy
stipul:it c• that read ings c:in be made only in the prc~cncc of a fully qu:ilt ficd dcctr1c1.1n .
occur? Comparisons of change across buildings and experimental con-
ditions presume that equal opportunities for change were present before
program initiation. In the present study experiment~! buildings were
matched in pairs on building size, function, and age in an attempt to
REFERENCES
ensure equal opportunity for change in the buildings being compared.
The quality of the matches was limited by the small number and great
BE~:-IETT . E. n. (1955) " D iscussion, decision. commitment. and consensus in •group
variety of available buildings. However, matches of the si:< dormitories
decision·.· Huma n Relat ions 8: 251-273.
were excellent. The first two dormitories. X and Y, are virtually identical HOWERS , D. G. ( 1976) Systems of Organi1ation: Management of the Human Resource.
in size and construction type. both are coed, both have kitchens and Ann Arbor: Univ. of Michigan Press. . . . . .
cafeterias serving equal numbers of students . The four remaining dormi- BROWER. M. ( 1975) "Ellpcricnce w it h self-management and pnruc1pa11on 1n United
tories are part of a .. residence complex .. ; although they have slightly States industry.· Admin. and Society 7 : 65-84.
COCH , L. and J. R. r . FRENCH, Jr. ( 1948) ..O"crcomin& n:sist3nce 10 change." 11 uman
different arrangements of resident wings and lounges, they are identical
Rcl31ions I: S 12-532. . .
in construction type, function, and type of occupants. Nevertheless, v v L WILLSON . and J. M. GOTTMAN (197S) Dest~n and Analysis of
GLA SS . G . ·• · · · d U · p
initial per capita consumption levels within the X- Y and C-D pairs Time-Series Experiments. Boulder: Colondo Assoc13IC .. "'~· rcn._ . •.
differed significantly. In both pairs the dormitory with higher initial LEVINE, L. and J . OUTLER ( 1952) .. Lecture \'S . group dec1s1011 in ch3ngrng behavior.
consumption achieved a significantly larger reduction; however. in J. or Applied Psychol ogy 36: 29-33. .
LEWI N, K. [ 195)) .. St ud ies in ~roup decision." in D. Cartwright :1nd A. Zander (ed(.)
neither pair were consumption levels within pairs significantly different Group D ·n:imics : Resea rch and Theory. E":inston. IL:_ Row. Peterson. .
during the program. These results arc consistent with the hypothesis McCLELLi\~'\l>. L anti I.. flEl.STFN (1979) .. f'romut1t1i; cncri:y COJhCr\'at.'."n 1n
that the programs sim ply servcJ to move d ormito ry pair members with · · d orm1· 10 r·1 ,~s by ph)·si"cal . polic\·
university . • :ind resident bcha\IOr ch:ini;cs. J . uf
different initial waste levels to the same level of waste. Comparisons En\'ironmcnt:il Systems. .
• ., C J DARLEY and L. J. BECKER (1977) .. Hc:ha,·1or:il approachrs to
of change in energy use, even percentage change, can be as misleading as S[l1(1MA., . .. · .' •. . · . . . '5-)37
rc sidt·ntial energy c1H1serva11on . Energy and llu1ld1np L )_ · _
direct comparisons of consumption level s. VROOl'-·l, \'. U. (191>9) .. lmlu>tri:il >1lCial p>y.:holOl!Y·.. in Ci . l.111tl7~y :inJ_L Ari:~""" (t•J'-1
11ie Handbook of So.:ial Psychology. Vol. S. Read ing. MA . Aod"on-\h,k) ·
McClcll3nd, Cook I ENEl~GY CONSERVATION IJJ