differences in turnover predictors between multilevel and single level direct selling organizations

Differences in Turnover Predictors between Multilevel and Single Level Direct Selling OrganizationsTHOMAS R.WOTRUBA *,STEWART BRODIE **&JOHN STANWORTH ***San Diego State University,San Diego,CA,USA,**Westminster Business School,University of Westminster,London,UKA BSTRACT Turnover among its salespeople is a significant issue for direct selling firms because attrition impacts the size and continuity of revenue generation by a firm’s sales force.While turn-over rates in direct selling are high overall,turnover rates and intentions to quit differ significantly between multilevel (ML)and single level (SL)forms of direct selling organizations.This study examines whether specific demographic and behavioural/attitudinal characteristics of direct sales-people correspond to differences in turnover between ML and SL salespeople.For many demo-graphic variables there are significant differences between ML and SL salespeople,but none of these differences correspond to differences in quitting intentions.There are also significant differ-ences between ML and SL salespeople on the behavioural and attitudinal variables studied.Ana-lysis revealed that the relationship between some of these variables and quitting intentions differed substantially between ML and SL salespeople.These variables included job satisfaction,organi-zational commitment,perceived image of direct selling in the marketplace,and the importance of the job characteristics of work rewards and career growth.Specific managerial implications fol-low from these findings.K EY W ORDS :Direct selling,organization,single level,multilevel,turnover,quittingBackground:Turnover and Direct SellingTurnover is often used synonymously with quitting or attrition.It involves the severing of employment relationships between an organization and its employee(s).It often (but not always)leads to the need for the organization to replace the person who has left his or her job with a new person so that the job will continue to be performed.Direct selling is a method of distribution of consumer goods and services through personal contact between buyer and seller away from a fixed business location (Peterson &Wotruba 1996).Turnover is a somewhat unique occurrence in direct selling as compared with other jobs,even with other selling jobs,since there is no formal quitting procedure by the direct salesperson or official notice of employmentCorrespondence Address:Stewart Brodie,Westminster Business School,University of Westminster,35Marylebone Rd,London NW15LS,UK.Tel.:+44(0)2079115000ext.3036;Fax:+44(0)2079115839;Email:brodies@ Int.Rev.of Retail,Distribution and Consumer Research,Vol.15,No.1,91–110,January 2005ISSN 0959-3969Print/1466-4402Online/05/010091-20#2005Taylor &Francis LtdDOI:10.1080/095939604200029914892T.R.Wotruba et al.termination by the company.Direct salespeople are self-employed independent contractors,not company employees,and in that sense direct selling has some char-acteristics in common with franchising(Brodie et al.,2002b;Stanworth et al., 2004).As independent contractors,direct selling activity is discretionary to many in that position,and many direct salespeople will cycle periodically from little or no activity to high levels of activity.Temporary inactivity does not always portend quitting,though many direct sellingfirms specify some length of inactivity as a likely indicator of turnover.In general,a direct sellingfirm will welcome back an inactive direct salesperson who chooses to become active once again.As independent contractors,direct salespeople are essentially their own bosses. They are free to spend whatever time they want as they seefit.Many participate in direct selling as a part-time activity,or as a supplement to other employment commitments,or as a temporary means for income until another employment opportunity is found.Maintaining direct selling activity requires considerable initia-tive and self-motivation(Pratt&Rosa,2003).Financial compensation depends essentially on commission or incentive pay since there are no salaries(Coughlan& Grayson,1998).For these reasons,inactivity among direct salespeople is common, and it is not unusual tofind annual turnover rates of100percent in the sales force of a typical direct selling company(Biggart,1989;Peterson&Wotruba,1996). Turnover can be anticipated by determining a salesperson’s‘intentions to quit’or‘propensity to quit’.These terms have been used as the basis for measures designed to predict inactivity or quitting behaviour among direct salespeople as well as others in various occupations.Numerous studies of turnover exist,and there is considerable evidence that measures of intentions to quit are accurate predictors of turnover or permanent inactivity(Porter&Steers,1973;Price,1977; Muchinsky&Morrow,1980;Bluedorn,1982;Steel&Ovalle,1984).Studies speci-fically of salespeople reinforce this pattern(Donnelly&Ivancevich,1975;Busch& Bush,1978;Parasuraman&Futrell,1983),and further studies involving direct salespeople specifically confirm thesefindings(Wotruba,1989,1990;Wotruba& Tyagi,1991,1992).Many of these studies have also succeeded in identifying the links between turnover or intentions to quit and other behavioural or attitudinal variables such as job satisfaction,organizational commitment,and met expecta-tions.Strong evidence has been presented to show that measures of intentions to quit are more direct causal antecedents of turnover than are the behavioural or attitudinal variables mentioned.Thus,turnover is a significant issue for direct sellingfirms because it impacts the size and continuity of the sales force producing thefirm’s revenues.An under-standing of the antecedents of turnover can help management devise programmes and practices that could reduce turnover if properly implemented.But for such programmes and practices to be most effective,they mustfit the nature of the organization and its workers.Direct sellingfirms operate under two major types of organizational structures, called multilevel and single level.In a multilevel(ML)organization,direct sales-people recruit,train and supervise other direct salespeople who become part of the recruiter’s downline and whose sales generate compensation for the recruiter.In a single level(SL)organization,salespeople do not build an organization via recruit-ing and training but rather devote all efforts to selling and achieve allTurnover Predictors in Direct Selling93 compensation based on their own sales.Recent evidence shows that salespeople in these two types of direct sellingfirms have some specific differences in personal characteristics,attitudes and behavioural measures.For instance,MLs show significantly greater organizational commitment than do SLs while SLs place greater importance on job characteristics that involve proving themselves to others when compared with MLs(Brodie et al.,2002a).This raises two important ques-tions:(1)Does turnover or intentions to quit differ between these two types of direct sellingfirms;and if so(2)are these attrition measures linked differently in ML versus SL organizations to attitudes or behaviours previously determined to precede turnover among direct sellingfirms in general.We address these questions as the focus of this study in a series of analysis stages,using thefindings in each stage to provide direction for analysis in subsequent stages.Because of the somewhat intricate nature of turnover in direct selling–invol-ving both permanent and temporary inactivity as well as the propensity to quit–the measure of turnover used successfully in previous studies is used here as well. Turnover is measured from a single scale incorporating both quitting and inten-tions to quit,so that answers can be obtained from respondents who had already quit and others who were still active but in various stages of intending to quit. That scale was framed as the following question,and generated the results indi-cated when posed to a sample of direct salespeople in theirfirst few months of activity:Which of these statements best describes your feelingsright now about continuing your direct selling activity?I have never thought about quitting.53.0%I seldom think of quitting.23.8I sometimes think of quitting.15.1I frequently think of quitting. 3.3I am just about ready to quit. 3.1I am no longer active in this direct selling activity. 1.7Total(n=638)100.0%This sample included respondents provided from22member companies of the Direct Selling Association in the UK.Each co-operating company agreed to send a mail questionnaire to a sample of their salespeople with a cover letter from the researchers.Selection of companies was based on gaining adequate representation of ML and SL salespeople from a spectrum of companies that varied in size and product or service sold.Out of the722responses received from the4,050mailed questionnaires(response rate of18percent),49did not report their type of organi-zation and an additional35did not answer the question on intentions to quit,so these were excluded from the analysis.The proportions of respondents who could be categorized as ML or SL closely matched corresponding proportions available from the UK Direct Selling Association.For instance,a recent UK direct selling report(Direct Selling Association UK,2000)indicated that67.2percent of UK Direct Selling Association memberfirms were ML.Among respondents to this94T.R.Wotruba et al.study69.7percent of those who could be categorized were ML and30.3percent were SL(for more details,see Brodie et al.,2002a).Comparison of Turnover Between ML and SL Respondents.Among the638respondents in this study who reported their quitting intentions and could be identified as associated with an ML or SL direct selling organization, 438were ML and200were SL.While very few of these respondents had actually quit their direct selling activity,a substantial number had given at least some thought to quitting,hence had some degree of‘intention’or‘propensity’to quit. Their responses to the turnover question,tallied by type of organization,appear in Table1.These results indicate that MLs are significantly different from SLs in their intentions to quit.MLs have lesser and SLs have greater turnover propensity.This conclusion applies regardless of whether the difference is measured by a chi-square test comparing the distributions of responses on the six scale positions or by a t-test comparing the mean scale positions of the two respondent groups.Since MLs and SLs are significantly different in their turnover propensity,the next step is to identify what variables relate to this measure of turnover,and whether MLs and SLs differ on those variables.This analysis is donefirst for demographic and personal history variables and then for behavioural and attitudinal variables. Analysis of Turnover Related to Demographic and Personal History CategoriesBefore examining whether MLs and SLs differ on demographic characteristics,it is useful to determine whether there are any relationships within demographic categories and the turnover measure of intentions to quit.Perhaps turnover is significantly more(or less)likely within certain demographic categories regardless of whether a direct salesperson was an ML or an SL.In some cases,however,a turnover-demographic relationship may not occur for MLs and SLs combined, Table1.Intentions to quit:ML and SL respondents comparedML respondents SL respondents Scale value Statement No.%No.%1Never thought about quitting25257.58643.0 2Seldom think of quitting9922.65326.5 3Sometimes think of quitting5512.64120.5 4Frequently think of quitting15 3.46 3.0 5Just about ready to quit11 2.59 4.5 6No longer active6 1.45 2.5 Total438100.0200100.0 Notes:Chi square:14.932(degrees of freedom=5),p=0.011Mean scale position:ML=1.75SL=2.07Standard deviation: 1.11 1.25Significance of difference(t-test,2-tailed)p=0.001Turnover Predictors in Direct Selling95 but can occur for one group rather than the other because those groups differ demographically and those differences correspond to differences in propensity to quit.For this reason a demographic comparison of MLs and SLs is also necessary to indicate where these two groups differ demographically and whether such differences correspond to differences in intentions to quit.An examination of the638direct salespeople in this study(ML and SL combined)was done for each demographic and personal history variable in the study.A comparison among the categories in each variable on intentions to quit produced the results in Part A of Table2.These results show that in the tradi-tional demographic variables there is no significant difference among the categories of those variables on intentions to quit.In other words,regardless of age,gender,marital status,income,or other traditional demographic measures the direct salespersons in these categories did not differ significantly in their intentions to quit.But with two measures of respondents’personal history regarding their direct selling jobs,including number of hours worked per week and tenure on that job,there are significant differences among respondent cate-gories.Regarding hours worked per week,those working50or more hours per week show the greatest intentions to quit.When the tenure categories are compared,those with between4and10months of tenure likewise show the greatest intentions to quit.Comparing MLs and SLs on these personal history measures as well as on the demographic variables is the next step to determine whether there are differences between those two groups that might correspond with the difference between them in intentions to quit.Part B of Table2presents these comparisons.Only those respon-dents that provided an answer on intentions to quit are included in these tables. Based on these results,there are significant differences between ML and SL direct salespeople in age,education,income,type of housing,previous experience, and hours per week worked in direct selling.From an inspection of these results, it appears that MLs are older,have higher education and higher income,are more likely to live in owner-occupied or private-rented housing,are more likely to have work experience that did not involve selling,and spend fewer hours per week in direct selling in comparison with the SLs.There are no significant differences between MLs and SLs on gender,marital status,location of residence,and tenure with their direct selling company.Thus far the data show that intentions to quit is related to no demographic variable and only one personal history variable on which MLs and SLs differ–hours per week worked in direct selling.But it is possible that there are relation-ships between intentions to quit and personal history or demographic characteristics that are unique to MLs only or to SLs only,but not to both.This investigation is accomplished by use of a two-way analysis of variance.This analy-sis technique is illustrated using hours worked and type of direct selling organization simultaneously.As explained below,two-way analysis of variance indicates that intentions to quit is related independently to both hours worked and type of direct selling organization(that is,ML versus SL),but not to the interaction between hours worked and organization type.In other words,the effect of hours worked on turnover intentions is the same regardless of type of organization.Since SLs work more hours per week than MLs,and the number of96T.R.Wotruba et al.Table2.Part A:Intentions to quit compared among categories of demographic and personalhistory variablesPart B:Personal history variables compared between ML and SL respondentsPart A Part BIntention to quit%of ML%of SL Variables and categories n Mean S.D.respondents respondents AgeUnder186 1.500.5480.5 2.018–2488 1.89 1.07711.020.025–2992 1.710.97814.015.530–34121 1.72 1.06619.019.035–44171 1.87 1.21228.822.545–5499 2.05 1.32016.513.555–6446 2.07 1.3237.8 6.065+14 1.57 1.284 2.5 1.5 Total637 1.85 1.162n=437n=200 Significance*p=0.258p=0.027 EducationNo formal education67 1.600.9549.213.6 CSE/YT/RSA47 1.87 1.244 5.312.1 GSCE(grades A-C),GCE‘O’level156 1.92 1.17521.431.7 GCE‘A’level,BTEC,City&Guilds134 1.91 1.18620.722.1 Higher education,below degree level95 1.78 1.18717.79.0 BA,BSc degree,or equivalent81 1.78 1.00015.27.5 Post-graduate degree53 2.08 1.41210.4 4.0 Total633 1.85 1.165n=434n=199 Significance*p=0.350p=50.001 GenderFemale455 1.86 1.17470.573.0 Male183 1.81 1.13329.527.0 Total638 1.85 1.162n=438n=200 Significance*p=0.627p=0.296Marital StatusMarried363 1.87 1.18259.552.3 Living as married75 1.87 1.07010.315.1 Separated,divorced74 2.00 1.39510.813.6 Widowed6 1.170.4080.5 2.0 Single,never married110 1.75 1.02417.116.6 Other6 1.170.408 1.10.5 Total634 1.85 1.163n=435n=199 Significance*p=0.287p=0.113Annual Household IncomeUnder£8,00092 1.86 1.22812.422.0£8,000to£13,999107 1.81 1.13415.822.6£14,000to£21,999156 1.91 1.19324.529.6£22,000to£36,000145 1.91 1.14826.219.9 Over£36,00098 1.81 1.18121.1 5.9 Total598 1.87 1.172n=412n=186 Significance*p=0.924p=50.001 (continued)Turnover Predictors in Direct Selling97Table2.(continued)Part A Part BIntention to quit%of ML%of SL Variables and categories n Mean S.D.respondents respondents Location of residenceUrban196 1.91 1.24932.830.5 Suburban272 1.89 1.14644.145.5 Rural143 1.76 1.07423.124.1 Total611 1.87 1.164n=424n=187 Significance*p=0.463p=0.853Type of housingOwner-occupied402 1.86 1.14165.758.4 Private rented108 1.89 1.16318.314.2 Local authority rented61 1.70 1.070 6.915.7 Housing association35 1.97 1.445 4.67.6 Other28 1.82 1.362 4.6 4.1 Total634 1.85 1.164n=437n=197 Significance*p=0.832p=0.003 Previous experiencePrevious direct selling experience190 1.88 1.14629.931.0 Other self-employment86 1.81 1.22215.010.7 Other selling102 1.75 1.05013.921.3 Other non-selling195 1.83 1.13833.126.4 None56 2.11 1.4108.110.7 Total629 1.85 1.165n=432n=197 Significance*p=0.435p=0.049Hours per week workedLess than5193 2.12 1.44437.016.25–9219 1.730.97933.337.410–19112 1.68 1.01517.218.720–2946 1.670.7627.17.630–3924 1.67 1.007 2.5 6.640–4923 1.91 1.125 1.68.150or more16 2.31 1.401 1.1 5.6 Total633 1.85 1.164n=435n=198 Significance*p=0.003p=50.001 Tenure with direct selling companyLess than1month259 1.660.97340.940.01–3months158 1.82 1.14824.725.04–6months95 2.22 1.39316.212.57–9months16 2.50 1.461 2.1 3.510–12months9 1.67 1.0000.9 2.513–24months27 1.70 1.103 3.9 5.0 More than24months74 2.04 1.28711.412.0 Total638 1.85 1.162n=438n=200 Significance*p=50.001p=0.457Note:*Significance in Part A determined by analysis of variance.Significance in Part B determined by Pearson Chi Square.hours worked is positively related to turnover intentions(that is,the more hours worked,the greater the intention to quit),it follows that SLs will have greater intentions to quit.But among the MLs who work the most hours per week,their intentions to quit are likewise high.Of course,there might be other factors influ-encing the number of hours worked that are not considered here but that would also influence the propensity to quit.A further explanation of the components of this analysis of variance(Table3) is as follows.Corrected ModelThe sum of squares shows the amount of variability in the dependent variable (intentions to quit)that can be explained or attributed to all of the other variables or components specified(organization type,hours worked,and the interaction between them).Corrected TotalThe sum of squares shows the aggregate of how much the observed values differ from the mean of all the cases combined.It is‘corrected’because the mean is subtracted from each observed value before squaring the difference.The‘total’is the same calculation but without subtracting the mean before squaring the values. When the corrected model sum of squares is divided by the corrected total sum of squares,the result(0.058or5.8percent in this case)indicates what share of the variability in intentions to quit can be explained by the variables selected for analysis.This calculation also produces the measure of R2.ErrorThis represents the combined variance within the groups analysed.In this case there are14groups(2organization types67hours worked categories).These values from each group are then summed and divided by the total number ofponents of analysis of varianceSource Sum ofsquaresDegrees offreedomMeansquare F ratio SignificanceCorrected model49.29613 3.792 2.90750.001 Intercept808.4531808.453619.85950.001 Organizational type9.33119.3317.1540.008 Hours worked25.2006 4.200 3.2200.004 Interaction 2.17260.3620.2780.948 (Organizational type6hours worked)Error807.333619Total3034.000633Corrected total856.62963298T.R.Wotruba et al.Turnover Predictors in Direct Selling99 degrees of freedom in all groups combined.The resulting mean square is the aver-age of the variances from each of the analysis groups.Organization Type,Hours Worked and InteractionThese are the analysis groups or variables being tested.The mean square in each is compared with the error mean square to test the null hypothesis that the varia-bility is the same among the means in each analysis group.The measure F is the ratio from this comparison and the significance is the probability of how often such a difference or F-ratio can occur if the true ratio in the populations repre-sented is1.For example,if the variability in intentions to quit was the same in the populations of both organization types,how often would we draw a sample of both organization types with the variability measures obtained?The table indicates it would happen with p=0.008or very rarely.Hence,the null hypothesis of no difference is rejected and it is concluded that the two organization types differ in their intentions to quit.We have seen this already in Table1.InterceptThe intercept is sometimes interpreted as a surrogate for other unknown or unidentified variables not included in the analysis that might be significant if they were included.The significance of the intercept suggests that there are other vari-ables that would help explain the variance represented by the error term.The data also show that MLs and SLs differ onfive other demographic vari-ables that,when examined for all direct salespeople combined,are not related to intentions to quit.It is possible,however,that one or more of these variables might relate to quitting intentions for one or the other group–MLs or SLs–because those two groups differ on those variables.To examine this possibility a two-way analysis of variance was performed for thosefive demographic variables, and those results are shown in summary form in Table4.These results all indicate that there is no interaction between organization type and any of the demographics in the explanation of intentions to quit.In other words,within demographic categories,there are no instances where the pattern of intentions to quit is different for one organization type than for the other.The overall conclusion from this analysis of demographic and personal history variables is that the difference in intentions to quit between MLs and SLs is not explained by any pattern of demographic or personal history characteristics unique to either group.While these two groups of direct salespersons do differ on demographic and personal history characteristics,those differences do not explain the difference between the two groups in their intentions to quit.Analysis of Turnover Related to Behavioural and Attitudinal VariablesThe previous analysis on demographic and personal history variables is now repeated with regard to behavioural and attitudinal variables.The eight variables chosen for this analysis are those used in a prior comparison of ML and SL salespeo-ple(Brodie et al.,2002a).These include a nine-item job satisfaction(JOBSAT)scale100T.R.Wotruba et al.Table4.Analysis of variance results relating demographic and personal history variables toorganization type(ML and SL)Analysis and variables p=SignificantAge0.139No Organization type0.054Yes*Interaction0.255NoCorrected model***0.014YesEducation0.427No Organization type0.028YesInteraction0.566NoCorrected model***0.040YesIncome0.767No Organization type0.008YesInteraction0.768NoCorrected model***0.093Yes*Type of housing0.620No Organization type0.147No**Interaction0.354NoCorrected model***0.041YesPrevious experience0.281No Organization type0.001YesInteraction0.342NoCorrected model***0.036YesNotes:*Significant at p40.10.**If the non-significant interaction term is omitted and the analysis is re-run,organization type becomes significant(p=0.001).***Dependent variable: Intentions to quitbased on the job satisfaction subscales of the Job Diagnostic Survey(Hackman& Oldham,1974,1975),a15-item organizational commitment(ORGCOMIT)ques-tionnaire(Porter et al.,1974),a25-item met expectations(METEX)scale(Wotruba &Tyagi,1991),and a12-item image(IMAGE)scale assessing the respondents’impression of how people feel about direct selling and direct salespeople(Wotruba, 1990).In addition,four job characteristics scales are used resulting from a factor analysis of the26items used in previous studies(Wotruba&Tyagi1992)that produced four subscales.Respondents rated the importance of each item to them, and the four resulting subscales were labeled as follows:JOBCHAR1,‘‘Proving oneself to others’’;JOBCHAR2,‘‘Personal feelings of success’’;JOBCHAR3,‘‘Work rewards and career growth’’;and JOBCHAR4,‘‘My own boss,in control’’. Thefirst objective of this analysis is to determine whether the intentions to quit measure(ITQUIT)is related to each of the behavioural/attitudinal variables regardless of whether the salesperson was an ML or an SL.This is done by exam-ining correlations between ITQUIT and each of the behavioural/attitudinal constructs individually.The second objective is to establish whether MLs and SLs differ significantly on each of those attitudinal constructs.Both of these aims are accomplished in Table5.The first column of Table 5shows negative and significant correlations between ITQUIT and each of the eight constructs examined for all respondents combined.Thus,higher scores on each construct,such as JOBSAT,correlate with low scores on ITQUIT.Turnover propensity is therefore related to each of these eight beha-vioural/attitudinal variables when analysed separately or one at a time.This completes the first objective.Comparisons between the mean scores of MLs and SLs on each behavioural/attitudinal construct are also shown,together with a measure of significance of the difference between those means.The last column of the table indicates that these two salesperson groups differ significantly on only two of the eight constructs examined –ORGCOMIT and JOBCHAR1(see Brodie et al .(2002a)for descrip-tions of the components of these factors).Based on the mean scores,MLs exhibit higher organizational commitment than do SLs,but MLs attribute lesser impor-tance than do SLs on job characteristics 1(‘‘proving oneself to others’’).There are no significant differences in mean scores between MLs and SLs on JOBSAT,IMAGE,METEX,and the other three measures of job characteristics.This table also gives some additional information of interest.Since the eight constructs are compilations of individual scaled items as borrowed from previous studies,the reliability measure of each scale is reported (coefficient alpha).A coef-Table parison of means of behavioural and attitudinal constructs between ML and SLrespondents and correlations of those constructs with intentions to quitConstruct All ML and SLRespondentsML only SL only Difference in means between ML and SL*Job satisfaction mean:33.5333.7333.10p =0.18(9items;alpha =0.894)R :0.4930.4680.537p :mit.Mean:54.4255.2452.61p =0.001(15items;Alpha =0.899)R :0.5170.4640.592p :50.00150.00150.001Image of Dir.Sell.Mean:38.5738.5238.68P =0.78(12items;alpha =0.694)R :0.3060.3080.320p :50.00150.00150.001Met expectations Mean:79.7880.0779.15p =0.35(25items;alpha =0.884)R :0.4140.3860.474p :50.00150.00150.001Job Char.Factor 1Mean:34.0133.2935.68p =50.001(10items;alpha =0.893)R :0.2070.2140.273p :50.00150.00150.001Job Char.Factor 2Mean:27.1927.1827.22p =0.91(7items;alpha =0.866)R :0.2330.2050.307p :50.00150.00150.001Job Char.Factor 3mean:19.4219.4219.44p =0.94(5items;alpha =0.800)R :0.1220.1370.092p :0.0020.0050.202Job Char.Factor 4Mean:11.5811.5411.66p =0.45(3items;Alpha =0.560)R :0.1450.1440.166p :50.0010.0030.020Notes :*Two-tailed probabilities,based on t-tests。