Mail Response Rates from Distributors - Science Direct

21 downloads 934 Views 393KB Size Report
Mail Response Rates from Distributors. Sudhir K. Chawla. P. V. (Sundar) Balakrishnan. Mary F. Smith. To forecast demand for a new product, mformutlon wan.
Mail Response Rates from Distributors Sudhir K. Chawla P. V. (Sundar) Balakrishnan Mary F. Smith To forecast demand for a new product, mformutlon wan sought from a nutlonul sample of 600 medic al equipment dealcry by means of a mud survev The crltrcul munugerml que>tlon for thus reyeurch way whether u new product Inventor wlshrng to obtain mformutmn from the sumple should employ u umverslty sponsor, offer monetur?; rncentrveh, or go It alone The results Indicate thut monetary mtentlveJ und umvetxt) sorshlp are effective wuy~ to ohtum mformutlon

upon-

INTRODUCTION Crucial for manufacturers developing new products before national launch IS the accurate estimation of demand Information regardmg demand 1s usually avadable from dealers, who have a better understandmg of the market or potential markets for the product Given that the mformation desired by the manufacturer exists with members of the dlstrlbutlon channel, the question becomes how best to ehclt this mformatlon from the mdlvldual dealers Since m most sltuatlons conductmg personal or

Address correspondence IO Mary F Snuth, Dept of Mdrketmg. Cdldormd State Unwenlty Sdn Bernardmo, 5500 UnlverWy Pdrkwdy, Sdn Berndrdmo, CA 92407

lndusrrlal Murkrtlng Monqrment 21, 307-310 (1992) 0 Elsewer Science Pubhshmg Co Inc , , 1992 655 Avenue of the Americas,, New York, NY 10010

telephone mtervlews IS prohlbltlvely expensive because of the large number of dealers to be sampled, a mall survey may be appropriate The maJor disadvantage of mall surveys, however, IS their low response rates [I] A number of studies have attempted to find ways to increase mad survey response rates [2-51 However, as noted by Duhan and Wilson [6], academic researchers have largely focused their attention on the Impact of the alternative mechamsms on consumers Notable exceptions, as pomted out by Jobber [7] Include such research as Mautz and Neumann’s [S] study of chief executive officers, directors, and accountants, Pressley and Tullar’s [9] study of marketing research directors m four different mdustrlal classes, and Tyagl’s [ IO] study of salespeople The preponderance of research studies of consumer populations IS understandable Unfortunately, for developers of new products and manufacturers attemptmg to increase response rates for their mad surveys from retailers and wholesalers, the results of these studies may not be directly apphcable In this context, Chdders, Pride, and Ferrell [2] have noted that different populations tend to respond differently to various mechamsms for mcreasmg response rates In-other words, what works best for consumers may not be the most sultable way to elicit responses from wholesalers and retaders, and may,

307 001 Y-8501/92/$5

00

University

affiliation

on the contrary, offend them Thus, the generallzablllty of these results to different population\ needs to.be exammed One of the major problems m mall survey research IS that there has been little attempt to evaluate the mcremental advantages of \uccessIve stlmuh For Instance, response rates to various types of appeals were exammed by Chllders, Pride, and Ferrell [2] However, they used an addItIona mechamsm, follow-up letter\ Consequently, It IS difficult to determme the Incremental effect\ of each mechanism Monetary

Incentives

The use and effectiveness of monetary mcentlves on the consumer population IS well-documented In their review articles, Lmsky [ 111 and Kanuk and Beren\on [ I] mdlcate that of the po\slble premiums and reward\, the most effective seems to be monetary mcentlves The results of \ub\equent \tudles [4, S] mdlcate that a $1 00 personal monetary mcentlve achieves higher response rates than promised mcentlves on return of the survey or lottery-onented cash prizes For bu\mess market\, Pre\\ley and Tullar 191 found that the very small monetary mcentlve of $0 10 Increased response rates Hansen [ 121 found that a small monetary mcentlve generated a higher response rate In a shorter period of time than a nonmonetary mcentlve However, when compared to a nonlncentlve control group, both ylelded reduced response quality Survey Sponsorship There has been considerably les\ research on the use of survey sponsorshlp Kanuk and Berenson [ 1] pomt out

SUDHIR K CHAWLA IS Associate Umvemty, San Angelo, Texas P V (SUNDAR) BALAKRISHNAN Ohlo State University, Columbus,

Professor

at Angelo

IS AssIstant Professor at Ohlo

MARY F SMITH IS Associate Professor at Callfornla Unwerslty, San Bernardino, California

308

State

State

helps response. that although many writers have recommended official support of surveys, there I\ “httle experlmental evidence on the effects of sponsorshlp ” Most of the studies conducted to date have used umverslty sponsors More recently, Tyagl [ IO] has called for research to examme “the effect\ of u\mg different types of sponsors, umverslty versus commercial sponsors, on response rates ” RESEARCH

DESIGN AND METHOD

This study mvestlgated the Impact of two different mechamsms for Increasing response rates from dealersthe use of survey spon\orshlp and the use of monetary mcentlves This basic objective of the study, albelt slmple, IS of crItIca managerial Importance Should a new product Inventor wlshmg to obtam mformatlon from natlonal retailers employ a survey sponsor, use monetary mcentlves, or go it alone? A sample of 600 dealers was randomly chosen trom a national hst of dealer\ of medlcal equipment Each member of the sample received a five-page questlonnalre assessmg consumer demand and acceptance for a new product Consistent with the recommendation of Childer\, Pride, and Ferrell [2], a control group was utlllzed to more effectively asse\\ the dlfferentlal magnitude and dlrectlon of the relative advantages of monetary mcentlve and survey sponsor\hlp Therefore, the 600 dealers were partltloned randomly Into one of three groups Control group (CC) m this case subjects were petltloned on ordinary stationery (I e , wlthout umver\lty letterhead but with sponsor Identlficatlon) Monetary mcentlve group (MC) subjects were sent a letter ldentlcal to that received by the control group In addltlon, they were told to find enclosed “a token of appreclatlon,” that is, a $1 00 bill Survey sponsorshlp group (SC) subjects received a letter on umver\lty stationery They were informed that the survey on the vlablhty of a new product was bemg conducted m conJunctIon with the umverslty Two hundred survey\ were mailed to dealer\ m each of the three groups In all three cases, except as noted above,

TABLE

TABLE 2 Response

1

Response

Rates for Dfferent

Mechanisms Surveys Returned

Letter

Surveys Sent

Undehvered

n

%

Completeness Number of Incomplete

Cost per Response

Mean

Standard Devlatlon

F Value

R’

2 23%

0 04

Control Group

200

40

15

Control Group

0 80

I 57

Monetary Group

200

42

62

39 24

464-l

Monetary Group

4 29

6 87

Survey SponsorshIp

200

43

31

19 75

4 169

Survey Sponsorstup Group

5 81

IO 31

9 315

$8 667

Items

Group *Not vgmficant dt 0 10 level

Responseratesbetween

the groups dre slgmficantly dlfferent dl the 0 0.5

level

the mformatlon provided to the respondents m the cover letter was identical In addition, all dealers were Informed that “we seek your expert opmlon ” Measures of response effectiveness employed m this study were response rate and response completeness In addition, we examined the blasmg effect of the alternative mechanisms employed on the various questlons relating to demand forecasts and product vlablhty RESULTS The number of undehverable surveys m the CG, MG, and SG was 40,42, and 43, respectively These numbers were relatively high for two reasons (1) the samplmg frame was not current (about three to four years out of date), and (2) there 1s high volatlhty m this type of buslness The response rates ranged from a low of 9 4% m the CG to 19 8% m the SG to a high of 39 2% m the MG All palrwlse differences m response rates are slgmficant at the 0 05 level Response rates and cost per completed response are presented m Table 1 These data suggest that a manufacturer IS better off usmg a university to sponsor its study than gomg It alone, especially when the manufacturer does not have a national reputation or 1s a new entrant to the busmes\ Along with doublmg the response rates, the cost per completed response 1s halved The 19 8% response rate compares favorably to response rates obtamed m other channel studies employmg umverslty letterheads [ 13, 141 On the other hand, If a manufacturer wishes to have complete control over the research process and Improve the response rate, It could employ monetary Inducements In this study, enclosing $1 00 with the survey (wlthout any addltlonal Inducements) resulted m a fourfold mcrease m response rate relative to the CG Note that the MG response rate IS twice that of the SG, whde the cost

per completed response was only about 10% higher. However, the cost figures should be interpreted with caution as they do not take mto account the “fees” a manufacturer would have to pay to obtain sponsorship A critical question 1s whether use of such mechamsms results m the respondents provldmg mformatlon “pleasmg” to the researcher In this study, dealers were given a photograph and description of the new product and were asked to forecast demand for the new product ’ There were no statlstlcal differences for the forecast demand for the new product by the three groups, nor any dlscemlble response bias This 1s a managerially important finding, since any response bias may Invalidate the use of either monetary mcentlves or survey sponsorship as appropriate ways to increase response rates The returned surveys were also analyzed for response completeness This was difficult to do given the nature of the survey due to the presence of a number of branchmg questions, respondents did not have to respond to all questions For the purpose of consistency with prior hterature, response completeness was defined as the number of Items omltted only from the number of questions respondents were asked to answer As shown m Table 2, analysis of variance results indicate that there was no significant difference m response completeness across the three groups The overall R* for all effects taken together 1s 0 04 The Independent groups do not account for much of the variance m response completeness DISCUSSION

AND CONCLUSION

We investigated whether a new product inventor wlshmg to obtain mformatlon from a dealer sample should

‘Because the study wd\ conducted for d real product currently under htlgdtlon detail\ of the demdnd forecast ddtd dre proprletdry

309

Use monetary incentives. employ umverslty sponsorshlp, monetary mcentlves, or go It alone Obtaining higher response rates IS desirable because of higher sample size, less need for follow-up contacts, and less concern with nonresponse bias Our results demonstrate that managers can Increase response rate and reduce the cost per completed survey wlthout Incurring validity problems The use of surveys sponsored by umversltles and the employment of monetary mcentlves are effective mechanisms by which managers can gather more mformatlon for less from members of dlstrlbutlon channels Using a umverslty sponsor doubled the response rate, while offering a monetary mcentlve mcreased the response rate fourfold compared with a control group ’ By way of caution, given the very low response rate m the control group, we suggest that the sample size, at least for the control group, be 200 for more Of course, this may require that other cells m the research design be smaller, especially If the samplmg frame as a whole is not very large In addition, we recommend for researchers with larger research budgets that future studies examme the relative impact of combining the two mechanisms studied here Alternatlvely, researcher\ might consider the theoretlcal and practical aspects of having a survey sponsored by a governmental agency \uch as the Small Busmess Admmistratlon

REFERENCES 1

Kanuk.

,

L

dnd Beremen,

2

ChIlden,

T

,

L

,

C

Ma11 Surveys dnd Respome

of Marketmg

Ltterdture Revtew, Journal Pride, W

,

M

dnd Ferrell, 0

the Effects of Appedl\ on Re\pon\e\ krtq 3

Rrsrurc h 17, 365-370

Golden,

L

,

L

Salutdtlon,

turner Resecrrth. K Comumer 4

W

,

Hubbdrd, R

and Shdrpe, L

7

.

ed

L

we Jam Chllder\

Pride. dnd Ferrell [2] m \ugge\tmg

thdt tuture resedrcher\ include control group\,

c\peciJlly

11 the refdtlve Im-

portdnce or ddVantdge ol edch mechdnl~m I\ to be fully determlned

310

for

Cd\h Prize\ dnd Mdll Survey Response

( 1988) P , and Lorenzl, ,

Duhan.

Dale F

Survey

Response\,

P

dnd Wllson.

, Monetary

R

Indu\rncrl

Jobber

Ddvtd,

8

Dale

SC1.

Incentive\ dnd Mdll Survey

13, 46-48

(1984)

Prenotlficdtlon

Murhetrng

dnd Indu\tnal

Manqement

19,

Pre\\ley,

95-105,

,

F L

dnd Tulldr

. The

Effective Corporate Audit Com-

WllhJm

(1970) L

.

A bdctor Interactive ln-

of Mall Survey Response Rates from A Commercldl

Idtlon, Journal oj Marker/q 10

Mad Surveys.

15, 183- 195 ( 19X6)

Butrnrcc Rewew 48, V-65

Milton M

ve\tlgatlon

Respome Rdte\ m Indu\tndl

M~mqemmt

, dnd Neumann,

MdutL, R F

mtttee, Harvurd 9

Improvmg

Marketq

, The

Tydgt, P K

Rewurth

14, 108-I

Effect5 of Appedls, Anonymity,

Popu-

11 ( 1977) dnd Feedbdck on Mad

Survey Response Pattern\ from Sdlespeople. Journal of the Atudem? of

( 1989)

Marketrng Jc wnt c 17, 235-246 11

Lmsky.

A

S

Hansen,

,

Stlmulatmg Reapon\e\

Oprnron Quartrrl\

Robert A

,

to Mdlled QueQlonndlre\

39, 82- 101

A Self-PerceptIon

Monetary dnd Nonmonetary

using d control group helped fdcllttdte the drdwlng of the\e con-

m Con-

A\\ocldtlon

Joumcrl o/ the Acudemy of Marketmg

Incentive\

13

Brown, J R

, and

DJ~, R

L

lnterpretdtmn

14

Ehdshberg. J

and Mtchle.

D

A

.

( 1980)

of Mdnlfe\t

Re\eurth

(1984)

Conflict m Dtstrt-

18, 263-274

Multtple Busme\\

termmants of Mdrketmg Chdnnel Conflict of Murketrng ReAeunh 21, 75-88

of the Effect of

on Mdll Survey Respondent Be-

, Measures

butmn Channels, Journal of Murkrtmg

A Re-

( 1975)

havlor, Journcrl of Market/n,? Resrcrrc h 17, 77-83

Consequently

on Mall

m Advuntrt

1980. Provo. UT

Rdte\, Journcrl of Advrrtlcq

Indu,trral

12

‘Indeed

The Effects of

( 1990)

ACKNOWLEDGMENTS

cluslon\

.

K

dnd Degree of Urbanlzdtlon

S Monroe,

JIId LIttIe. E

Pdohllo, J G Re\pon\e

view. Publu

This work was supported in part by the Unlverslty of Evansville We thank Pramod Goel, Lisa Wllhams, and Ian Brennan for their assistance with this study

of

Research, pp 292-298

ence 16, 42-44

6

A Redses\ment

to Mdll Surveys, Journul of Mur-

Rdte, Speed and Qudllty,

Rates A Threshold AndlyV\,

5

,

T

Incentive,

Questlonndlre Re\pon\e

,

C

A

(1975)

( 1980)

Anderson.

Monetdry

Rates

Research 12, 440-443

(1989)

Goal\ Set a\ De-

An Emplrlcdl Study, Journal