Why per­for­mance ap­praisal sys­tems fail

Many com­pa­nies have dropped per­for­mance ap­praisal be­cause it’s just not up to the job

The Irish Times - Business - - WORLD OF WORK - Man­ag­ing Work Kevin Mur­phy Prof Kevin Mur­phy is the Kemmy chair of work and em­ploy­ment stud­ies at Univer­sity of Lim­er­ick

Per­for­mance ap­praisal is the job man­agers and em­ploy­ees love to hate. Man­agers dis­like giv­ing per­for­mance feed­back and em­ploy­ees dis­like re­ceiv­ing it.

Sur­veys of sat­is­fac­tion with per­for­mance ap­praisal and per­for­mance man­age­ment sys­tems rou­tinely re­port abysmal re­sults. It is com­mon for 80-90 per cent of man­agers, ex­ec­u­tives and hu­man re­source pro­fes­sion­als to re­port that the per­for­mance ap­praisal and per­for­mance man­age­ment sys­tems in their or­gan­i­sa­tions are fail­ing.

This al­most rou­tine fail­ure of per­for­mance ap­praisal sys­tems is not for lack of try­ing, as I found in re­view­ing 100 years of re­search on per­for­mance ap­praisal and per­for­mance man­age­ment in Jour­nal of Ap­plied Psy­chol­ogy .A num­ber of strate­gies for im­prov­ing per­for­mance ap­praisal, rang­ing from rater train­ing and mem­ory aids to the devel­op­ment of bet­ter rat­ing scales have been tried, with gen­er­ally dis­ap­point­ing re­sults.

Or­gan­i­sa­tions such as Deloitte, Adobe and Ac­cen­ture have ei­ther dropped per­for­mance ap­praisal al­to­gether or have rad­i­cally sim­pli­fied their ap­praisal sys­tems.

Other or­gan­i­sa­tions have gone in the op­po­site di­rec­tion, ad­ding mul­ti­ple lay­ers of eval­u­a­tion by in­clud­ing peer ratings, self-ratings, or ratings from cus­tomers and clients as part of a mul­ti­source feed­back sys­tem. Un­for­tu­nately, there are few rea­sons to be­lieve that any of these strate­gies will suc­ceed.

It is time to stop re­ar­rang­ing the deck chairs; the Ti­tanic is clearly sinking. Rather than con­tin­u­ing to tinker with fail­ing sys­tems, we need to ex­am­ine the struc­tural flaws that cause per­for­mance ap­praisal sys­tems to fail, and use this in­for­ma­tion to take rad­i­cal ac­tion.

Two prob­lems are present in vir­tu­ally ev­ery ap­praisal sys­tem which cause them to fail: con­flict­ing pur­poses and con­flict­ing per­spec­tives.

The first struc­tural flaw is that com­pa­nies and or­gan­i­sa­tions rou­tinely try to use per­for­mance ap­praisal sys­tems for con­flict­ing pur­poses. I re­cently an­a­lysed data from a sur­vey of sev­eral thou­sand mid-sized and large or­gan­i­sa­tions in 24 coun­tries. Over 80 per cent have some sort of for­mal per­for­mance ap­praisal sys­tem, and the great ma­jor­ity of them use ap­praisals for two pur­poses: iden­ti­fy­ing train­ing and devel­op­ment needs; and iden­ti­fy­ing can­di­dates for salary in­creases, pro­mo­tions, sep­a­ra­tions, etc.

Con­flict­ing de­mands

But these place con­flict­ing de­mands on per­for­mance ap­praisal sys­tems. Us­ing ap­praisals for train­ing and devel­op­ment re­lies on in­for­ma­tion about in­di­vid­ual strengths and weak­nesses – ie dif­fer­ences within in­di­vid­u­als – while their use for ad­min­is­tra­tive pur­poses (eg salary ad­min­is­tra­tion) re­lies on in­for­ma­tion about dif­fer­ences be­tween peo­ple.

It is al­most im­pos­si­ble to get clear in­for­ma­tion about dif­fer­ences be­tween peo­ple and dif­fer­ences within peo­ple from the same set of per­for­mance rat­ing. Ratings that show clear peaks and val­leys, dis­tin­guish­ing strengths from weak­nesses, will end up pro­duc­ing av­er­age ratings that are very close to the mid­dle of the scale for just about all em­ploy­ees, mak­ing it dif­fi­cult to dis­tin­guish strong per­form­ers from av­er­age or weak per­form­ers.

Per­for­mance ratings that clearly dis­tin­guish good per­form­ers from av­er­age and poor per­form­ers tend to pro­duce flat pro­files be­cause they fo­cus on over­all per­for­mance and ef­fec­tive­ness as op­posed to in­di­vid­ual strengths and weak­nesses.

Or­gan­i­sa­tions that at­tempt to use the same per­for­mance ap­praisal sys­tem for train­ing and devel­op­ment and for salary ad­min­is­tra­tion of­ten end up with sys­tems that are not par­tic­u­larly ef­fec­tive for ei­ther pur­pose.

An even more im­por­tant struc­tural flaw in­volves dif­fer­ences in per­spec­tive. One of the most ro­bust find­ings from re­search in this area is that peo­ple al­most al­ways rate their own per­for­mance higher than do their peers, their su­per­vi­sors or al­most any other set of ex­ter­nal raters. This self-serv­ing bias is a prod­uct of a num­ber of psy­cho­log­i­cal pro­cesses (such as a de­sire to main­tain a pos­i­tive self-im­age and dif­fer­ences in the way we process in­for­ma­tion about our­selves as against oth­ers) that are strongly in­grained and al­most uni­ver­sal in scope (in some Asian cul­tures, self-ratings are closer to ratings pro­vided by oth­ers).

The im­pli­ca­tions for per­for­mance eval­u­a­tion and feed­back are dev­as­tat­ing. If a su­per­vi­sor gives per­for­mance ratings that are fair and ac­cu­rate, it is al­most cer­tain that the em­ploy­ees who re­ceive them will feel that they are be­ing un­fairly crit­i­cised and un­der­val­ued. One of the rea­sons for per­va­sive rat­ing in­fla­tion – in most or­gan­i­sa­tions, 80 per cent or more of the em­ploy­ees who are eval­u­ated re­ceive ratings of “above av­er­age” to “ex­cel­lent” – is that su­per­vi­sors and man­agers un­der­stand that ac­cu­rate ratings will lead to con­flict and to the re­jec­tion of feed­back.

Sub­stan­tial changes

Are there so­lu­tions to these struc­tural prob­lems? In our re­search, my col­leagues and I have rec­om­mended two sub­stan­tial changes in how or­gan­i­sa­tions con­duct and use per­for­mance ap­praisals. First, de­cide whether you want to use per­for­mance ap­praisals for train­ing and devel­op­ment or for salary ad­min­is­tra­tion, pro­mo­tion, and the like. At­tempt­ing to do both with the same ap­praisal sys­tem will fail.

Our re­search sug­gests that us­ing per­for­mance ap­praisals for high-stakes de­ci­sions, such as salary ad­min­is­tra­tion, pro­mo­tion and the like is of­ten a fool’s er­rand. Or­gan­i­sa­tions that fol­low this path cre­ate strong pres­sure on su­per­vi­sors and man­agers to in­flate ratings, mak­ing it dif­fi­cult to mean­ing­fully dif­fer­en­ti­ate su­pe­rior per­for­mance from av­er­age and even poor per­for­mance.

The re­search sug­gests that, if you are go­ing to do per­for­mance ap­praisals at all, you should use them solely as a tool for iden­ti­fy­ing train­ing and devel­op­ment needs. This means ap­praisal sys­tems should be ruth­lessly tai­lored to that pur­pose.

Most per­for­mance ap­praisal sys­tems ask su­per­vi­sors or man­agers to rate each em­ployee on sev­eral per­for­mance di­men­sions. A much bet­ter ap­proach would be to rank-or­der per­for­mance ar­eas, re­gard­less of the em­ployee’s over­all per­for­mance level. For ex­am­ple, sup­pose you are asked to rate per­sonal care aides on the key di­men­sions of their job (com­mu­ni­ca­tion, as­sist­ing oth­ers, get­ting in­for­ma­tion from clients, doc­u­ment­ing and record­ing client con­di­tions and ac­tiv­i­ties). Rather than rat­ing each di­men­sion in­di­vid­u­ally, it would be bet­ter to iden­tify the rel­a­tive strength of the em­ployee on var­i­ous as­pects of per­for­mance. A num­ber of rank­ing strate­gies might be ap­plied to pro­vide a de­tailed pic­ture of each in­di­vid­ual’s train­ing and devel­op­ment needs.

A strat­egy like this means or­gan­i­sa­tions need to give up on us­ing per­for­mance ratings to de­ter­mine salary raises, pro­mo­tions and the like, but it is clear that at­tempt­ing to use the same sys­tem for both pur­poses al­most al­ways ends up mak­ing a mess of both de­ci­sions.

It is cer­tainly bet­ter to de­cide which task is im­por­tant and do it well than to con­tinue to muddle the whole process of per­for­mance ap­praisal and per­for­mance man­age­ment.

Two prob­lems are present in vir­tu­ally ev­ery ap­praisal sys­tem which cause them to fail: con­flict­ing pur­poses and con­flict­ing per­spec­tives

Newspapers in English

Newspapers from Ireland

© PressReader. All rights reserved.