Market Segmentation In Radio Listening Habits

Market Segmentation In Radio Listening Habits Market segmentation plays an important role in radio listening habits due to three important reasons. Firstly, it allows marketers to identify groups of listeners with similar needs and facilitate the analysis of characteristics and listening behaviour of these groups (Soutar Clarke 1983; Kent 1994; Arjona et al. 1998). Secondly, segmentation provides marketers with critical information that are necessary for designing marketing mixes that are consistent with the characteristics and desires of one or more segments (Arjona et al. 1998; Gatfield 2006). Thirdly, segmentation allows radio stations achieved its objectives while satisfying the needs and wants of its customers and listeners (Fitzgerald 2004; Gatfield 2006). According to past research done on radio listening habits had indicated that lifestyle segmentation is an appropriate and effective approach adopted by marketing managers to reach its target audiences (Massy 1971; Soutar Clarke 1983; Evans, Lawson Todd 2006). Therefore, this quantitative study will be focusing on a key research question Do lifestyle predicts radio listening patterns for 6WF and 96FM Radio Station? Firstly, an external market research company was engaged to conduct phone interviews within the Perth metropolitan area and respondents were asked to respond to 43 key sets of AIO (activity, interest and opinion statement). The AIO approach is one of the most common approach use by scholars to measure consumers lifestyle (Li 2004). Respondents lifestyle can be assess through a 1-7 likert scale measurement (where 1 stands for strongly disagree and 7 stands for strongly agree) (Cicia et al. 2010). Appendix 1 shows the AIO questionnaire. Responses were obtained from 400 household in metropolitan Western Australia. Secondly, factor analysis will be used to identify the latent construct of questionnaire as this is commonly use in business research (Hair Jr et al. 2010). A Confirmatory Factor Analysis (CFA) will be used to verify all the questionnaire are categorized under the latent dimensions as proposed by previous theories or literatures (Soutar Clarke 1983) because the research will be using the 43 seven-point Lickert scale items to gauge if lifestyle influence a respondents listening habits. Thirdly, according to Leung, Fund and Lee (2009) a stepwise regression analysis is an appropriate method used to predict if lifestyle plays an important role in affecting radio listening habits. But before a factor analysis can be conducted the assumption of sample size, normality, linearity, outliers among cases, multicollinearity and singularity, factorability of the correction matrix and outliners among variables before analysis must be conducted. Results: The assumption on sample size was adequate as the research is based on 400 responses which are higher than the rule of thumb of 100. Thus, the sampling size is adequate for a factor analysis. When testing for normality on the AIO items, most of the AIO items seem to be normal with the exception of AIO8, 10,11,14,25,31,32,38. These 8 cases also had a series of outliners. Therefore, a data transformation will be required to determine if these questions can be kept for the analysis. After a series of data transformation, normality could not be achieved. Even with the deletion of outliners, normality was not achievable. Therefore these questions were eliminated from the analysis. Thus a total of 35 AIO items will be used. AIO8: AIO10 AIO11 AIO14 AIO25 AIO31 AIO32 AIO38 Reliability Statistics Cronbachs Alpha Cronbachs Alpha Based on Standardized Items .732 .739 The reliability statistic as indicated above has a result higher than Cronbachs alpha .70, which indicates an acceptable degree of internal consistency. Confirmatory Factor Analysis will be conducted to verify the hidden dimension of lifestyle towards listening habits as well as to determine the number of items categorized under each hidden dimension. The factorability of a correlation matrix can be detected via the Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy. In relation to this, the measure of sampling adequacy for each item (as shown on the diagonal of the anti-image correlation matrix) is larger than .5. Anti-image Matrices Extraction Method: Principal Component Analysis. The Communalities table shows the communalities value for each of the 35 items in the AIO questionnaire. Note that aio39 (I listen to the radio for a specific announcer or DJ) shows the highest communalities (.832), whereas aio5 (I have traditional ideas about most things) shows the lowest communalities (.413). The value of the initial communalities represents the percentage of variance in each item that can be explained by all possible factors. Hence, the value of initial communalities of (1.000) means that 100% of the variance in an item can be explained by all possible factors. On the other hand, the value of extracted communalities represents the percentage of variance in each item that can be explained by the extracted factors via the Principal Component (PC) Analysis. Hence the value of the extracted communalities is smaller as compared to the value of initial communalities. Based on the Total Variance Explained table, we can conclude that only 13 factors (with eigen value of more than 1) have been extracted via Principal Component (PC) Analysis. The first extracted factor can explain 11.448% in the items; the second factor can explain 10.115% of the variance in the items, all the way to the 11th extractor factor which explained 3.015% of the variance in the items. These 11 extracted factors can explain 65.518% of the variances in the AIO questionnaire items. Total Variance Explained Extraction Method: Principal Component Analysis. The Rotation Sums of Squared Loadings column reports information on the extracted factors with their respective eigen values, the percentage of variance and the cumulative percentage of variance explained by the extracted factor after the Varimax rotation. Only 11 factors (with eigen value of 1.055 to 4.007 respectively have been extracted. The first extracted factor can explain 7.744% of the variance in the items whereas the 11th factor can explain 3.727% of the variance in the items. As a whole, these 11 factors can explain 65.518% of the variance in questionnaire items. The Varimax rotation has changed the percentage of variance explained by the 11 factors (for example Factor 1 from 11.448% to 7.744%). The scree plot displays the eigen value for each of the factor. The plotted eigen value is based on the eigen value reported in Extraction Sum of Squared Loadings column. From the scree plot, observation can be made that there are two dominant factors with an eigen value of greater than 3.540. The Varimax rotation with Kaiser Normalization is conducted to better categorize the component matrix. Each item of the questionnaire that best suited a particular component will be categorized together and highlighted in different colors. For example, component 1 consists of items with a factor loading of 0.831 to 0.874. Appendix 2 provides the series of AIO questions associated to each component. Examination of the items grouping in each component allows the representation of a conceptually distinct aspect of lifestyle to radio listening as indicated in Appendix 2. The Rotated Component Matrix provide a form of content validity as it provide an assessment of the correspondence of the variables to be included in each component and its conceptual definition. This form of validity, also known as face validity, subjectively assesses the correspondence between the individual items and the concept through ratings (Hair Jr et al. 2010). For example, items in component 1 can be classified under a main concept or variables TV Addicts. The mean score of items associated to each component will be computed through the Transform Compute Variables function, for example, mean(aio27,aio20,ai04) of each respondent to form a new variables call Lifestyle_TV_Addicts. Rotated Component Matrixa Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. The computed data had resulted in 11 newly created variables that can be used to perform a multiple regression and provide information if lifestyle influence radio listening behaviour. The list of newly created variables can be found in Appendix 2. A stepwise linear regression will be conducted to address the below mentioned research question: What contributions do the 11 variables make to the prediction of radio listening preference of each radio station? Null Hypothesis: Lifestyle influence radio listening preference Alternate Hypothesis: Lifestyle does not influence radio listening preference Radio Station: 6WF Variables Entered/Removeda Model Variables Entered Variables Removed dimension0 1 Lifestyle_Cultural_Classical . 2 Lifestyle_Fashion . 3 Lifestyle_Outdoor . a. Dependent Variable: q3wf The table above shows the order in which the variables were entered and removed from the model. 3 variables were added and none were removed. In addition, an examination of the Mahalanobis distance MAH_1 values had indicated that there are no multivariate outliers among the independent variables as there are no values that are greater than or equal to the critical chi square value of 13.8 at an alpha level of .001 (Coakes, Steed Ong 2010). Model Summaryd Model R R Square Adjusted R Square Std. Error of the Estimate Change Statistics R Square Change F Change df1 df2 dimension0 1 .279a .078 .075 1.991 .078 33.576 1 398 2 .332b .111 .106 1.958 .033 14.603 1 397 3 .375c .141 .134 1.927 .030 13.968 1 396 a. Predictors: (Constant), Lifestyle_Cultural_Classical b. Predictors: (Constant), Lifestyle_Cultural_Classical, Lifestyle_Fashion c. Predictors: (Constant), Lifestyle_Cultural_Classical, Lifestyle_Fashion, Lifestyle_Outdoor d. Dependent Variable: q3wf The above model summary indicated that model 1, which included only Lifestyle_Cultural_Classical accounted for 7.5% of the variance (adjusted R Square = 0.075). The inclusion of Lifestyle_Fashion in model 2 resulted in an additional 3.1`% of the variance explained. The inclusion of Lifestyle_Outdoor into model 3 resulted in an additional 2.8% of the variance explained (R Square change = 0.03). The whole model accounted for 13.4% of the variance in radio listening preference, which is highly significant, as indicated by the F-value of 21.635 in the ANOVA table below. ANOVAd Model Sum of Squares df Mean Square F Sig. 1 Regression 133.105 1 133.105 33.576 .000a Residual 1577.792 398 3.964 Total 1710.897 399 2 Regression 189.082 2 94.541 24.663 .000b Residual 1521.815 397 3.833 Total 1710.897 399 3 Regression 240.931 3 80.310 21.635 .000c Residual 1469.967 396 3.712 Total 1710.897 399 a. Predictors: (Constant), Lifestyle_Cultural_Classical b. Predictors: (Constant), Lifestyle_Cultural_Classical, Lifestyle_Fashion c. Predictors: (Constant), Lifestyle_Cultural_Classical, Lifestyle_Fashion, Lifestyle_Outdoor d. Dependent Variable: q3wf The ANOVA assess the overall significance of the model. As p Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) 1.120 .267 Lifestyle_Cultural_Classical .384 .066 2 (Constant) 2.016 .352 5.732 Lifestyle_Cultural_Classical .444 .067 .323 6.625 Lifestyle_Fashion -.266 .070 -.186 -3.821 3 (Constant) 1.280 .398 3.213 Lifestyle_Cultural_Classical .452 .066 .329 6.856 Lifestyle_Fashion -.306 .069 -.214 -4.413 Lifestyle_Outdoor .221 .059 .176 3.737 Dependent Variable: q3wf Therefore the overall strength of the model in predicting lifestyle influence on radio listening preference for 6WF is as follow: Adjusted R square = .134; F3,396 = 21.635, p Predictor Variable Beta p Lifestyle_Cultural_Classical .329 p Lifestyle_Fashion -.214 p Lifestyle_Outdoor .176 p Radio Station: 96FM Variables Entered/Removeda Model Variables Entered Variables Removed dimension0 1 Lifestyle_Radio_Addicts . 2 Lifestyle_Tradition . 3 Lifestyle_Cultural_Classical . 4 Lifestyle_Conservative . 5 Lifestyle_Follower . 6 Lifestyle_Fashion . 7 Lifestyle_TV_Addicts . a. Dependent Variable: q396fm The table above shows us the order in which the variables were entered and removed from the model. Seven variables were added and none were removed. In addition, an examination of the Mahalanobis distance MAH_2 values had indicated that there are no multivariate outliers among the independent variables as there are no values that are greater than or equal to the critical chi square value of 13.8 at an alpha level of .001 (Coakes, Steed Ong 2010). Model Summaryh Model R R Square Adjusted R Square Std. Error of the Estimate Change Statistics R Square Change F Change df1 df2 dimension0 1 .273a .074 .072 2.448 .074 31.961 1 398 2 .365b .133 .129 2.371 .059 27.051 1 397 3 .417c .174 .168 2.318 .041 19.607 1 396 4 .449d .201 .193 2.282 .027 13.358 1 395 5 .468e .219 .209 2.260 .017 8.795 1 394 6 .479f .229 .218 2.248 .011 5.388 1 393 7 .488g .239 .225 2.237 .009 4.788 1 392 a. Predictors: (Constant), Lifestyle_Radio_Addicts b. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition c. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical d. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative e. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower f. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower, Lifestyle_Fashion g. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower, Lifestyle_Fashion, Lifestyle_TV_Addicts h. Dependent Variable: q396fm Model 1, which included Lifestyle_Radio_Addicts accounted for 7.2% of the variance (Adjusted R Square 0.072). The inclusion of Lifestyle_Tradition into model 2 resulted in an additional 6% of the variance being explained (R Square Change = .059). The inclusion of Lifestyle_Cultural_Classical in model 3 resulted in an additional 4% of the variance being explained (R Square Change = .041). The inclusion of Lifestyle_Conservative in model 4 resulted in an additional 3% of the variance being explained (R Square Change = .027). The inclusion of Lifestyle_Follower in model 5 resulted in an additional 2% of variance explained (R Square Change = .017). The inclusion of Lifestyle_Fashion into model 6 resulted in an additional 1% of the variance explained (R Square Change = .011). Lastly, the inclusion of Lifestyle_TV_Addicts into model 7 resulted in an additional 1% of the variance explained (R Square Change = .009). The whole model accounted for 22.5% of the variance, which is highly signifi cant as indicated by the F-value of 17.547. ANOVAh Model Sum of Squares df Mean Square F 1 Regression 191.481 1 191.481 31.961 Residual 2384.479 398 5.991 Total 2575.960 399 2 Regression 343.593 2 171.797 30.552 Residual 2232.367 397 5.623 Total 2575.960 399 3 Regression 448.908 3 149.636 27.858 Residual 2127.052 396 5.371 Total 2575.960 399 4 Regression 518.486 4 129.622 24.885 Residual 2057.474 395 5.209 Total 2575.960 399 5 Regression 563.413 5 112.683 22.060 Residual 2012.547 394 5.108 Total 2575.960 399 6 Regression 590.630 6 98.438 19.486 Residual 1985.330 393 5.052 Total 2575.960 399 7 Regression 614.586 7 87.798 17.547 Residual 1961.374 392 5.004 Total 2575.960 399 a. Predictors: (Constant), Lifestyle_Radio_Addicts b. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition c. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical d. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative e. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower f. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower, Lifestyle_Fashion g. Predictors: (Constant), Lifestyle_Radio_Addicts, Lifestyle_Tradition, Lifestyle_Cultural_Classical, Lifestyle_Conservative, Lifestyle_Follower, Lifestyle_Fashion, Lifestyle_TV_Addicts h. Dependent Variable: q396fm The ANOVA assess the overall significance of the model. As p Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .534 .537 Lifestyle_Radio_Addicts .599 .106 2 (Constant) 1.858 .579 3.208 Lifestyle_Radio_Addicts .566 .103 .258 5.505 Lifestyle_Tradition -.346 .067 -.243 -5.201 3 (Constant) 3.021 .624 4.841 Lifestyle_Radio_Addicts .624 .101 .284 6.152 Lifestyle_Tradition -.390 .066 -.274 -5.921 Lifestyle_Cultural_Classical -.348 .079 -.206 -4.428 4 (Constant) 4.330 .711 6.088 Lifestyle_Radio_Addicts .651 .100 .296 6.503 Lifestyle_Tradition -.353 .066 -.248 -5.371 Lifestyle_Cultural_Classical -.292 .079 -.173 -3.696 Lifestyle_Conservative -.363 .099 -.170 -3.655 5 (Constant) 4.539 .708 6.413 Lifestyle_Radio_Addicts .723 .102 .329 7.084 Lifestyle_Tradition -.330 .065 -.232 -5.042 Lifestyle_Cultural_Classical -.312 .079 -.185 -3.975 Lifestyle_Conservative -.351 .098 -.164 -3.568 Lifestyle_Follower -.195 .066 -.137 -2.966 6 (Constant) 4.187 .720 5.815 Lifestyle_Radio_Addicts .679 .103 .309 6.567 Lifestyle_Tradition -.325 .065 -.228 -4.984 Lifestyle_Cultural_Classical -.351 .080 -.208 -4.394 Lifestyle_Conservative -.359 .098 -.168 -3.663 Lifestyle_Follower -.220 .066 -.155 -3.316 Lifestyle_Fashion .193 .083 .110 2.321 7 (Constant) 4.560 .737 6.191 Lifestyle_Radio_Addicts .666 .103 .303 6.463 Lifestyle_Tradition -.297 .066 -.208 -4.490 Lifestyle_Cultural_Classical -.360 .080 -.213 -4.525 Lifestyle_Conservative -.333 .098 -.156 -3.395 Lifestyle_Follower -.203 .067 -.143 -3.053 Lifestyle_Fashion .210 .083 .120 2.526 Lifestyle_TV_Addicts -.162 .074 -.101 -2.188 a. Dependent Variable: q396fm Therefore the overall strength of the model in predicting lifestyle influence on radio listening preference for 96FM is as follow: Adjusted R Square = .225, F7,392 = 17.547, p Predictor Variable Beta p Lifestyle_Radio_Addicts .303 p Lifestyle_Tradition -.208 p Lifestyle_Cultural_Classical -.213 p Lifestyle_Conservative -.156 p = 0.001 Lifestyle_Follower -.143 p = 0.002 Lifestyle_Fashion .120 p = 0.012 Lifestyle_TV_Addicts -.101 p = 0.029 The models representing both radio stations, 6WF and 96FM, are not a good model as they only explained 13.4% and 22.5% of the variance (R Square) in radio listening preference. Therefore, lifestyle can be seen as an insignificant predictor of radio listening preference. Therefore, the H0 will be rejected and H1 will be accepted and conclude that lifestyle does not influence radio listening preference. Conclusion: It is apparent that, the two radio stations, 6WF and 96FM, did not have distinct audiences with different lifestyle. This is a direct contrast to the previous research conducted by Soutar and Clarke (1983) that concluded lifestyle plays a role in influencing radio listening patterns. Therefore, the respective radio stations program manager need not have distinct radio programming policy to attract a different group of audiences. However, the research has indicated that lifestyle plays a more important role in predicting the radio listening preference for 96FM than 6WF because the model represented in the regression analysis managed to explained 22.5% of the variance, which is 9.1% more than the 6WFs model (13.4% of the variance explained). Bibliography Arjona, LD, Shah, R, Tinivelli, A Weiss, A 1998, Marketing to the Hispanic Consumer, The McKinsey Quarterly, vol. 1, no. 3. Cicia, G, Corduas, M, Del Giudice, T Piccolo, D 2010, Valuing Consumer Preferences with the CUB Model: A CaseStudy of Fair Trade Coffee, International Journal on Food System Dynamics, vol. 1, pp. 82-93. Coakes, SJ, Steed, L Ong, C 2010, SPSS Version 17.0 for Windows: Analysis without anguish, Wiley, Milton, Qld. Evans, S, Lawson, R Todd, S 2006, New Zealand in the 21st century: A consumer lifestyles study, NZ Post, Loyalty NZ, University of Otago Fitzgerald, J 2004, Evaluating Return on Investment of Multimedia Advertising with a Single-Source Panel: A Retail Case Study, Journal of Advertising Research, vol. 44, no. 3, pp. 262-270. Available from: ufh. Gatfield, T 2006, Community Radio Broadcasting and Positioning an Australian Perspective, Marketing Review, vol. 6, no. 2, pp. 183-189. Available from: buh. Hair Jr, JF, Black, WC, Babin, BJ Anderson, RE 2010, Multivarate Data Analysis: A Global Perspective, Seven edn, Pearson, Upper Saddle River, New Jersey. Kent, R 1994, Measuring Media Audiences, Routledge, London. Leung, L, Fung, AYH Lee, PSN 2009, Embedding into out lives: New opportunities and challenges of the internet, The Chinese University Press, NT, Hong Kong. Li, S-CS 2004, Examining the factors that influence the intentions to adopt internet shopping and cable television shopping in Taiwan, New Media Society, vol. 6, no. 2, pp. 173-193. Massy, WF 1971, Discriminant Analysis of Audience Characteristics, in Multivariate Analysis in Marketing, ed. D Aaker, Wadsworth, California. Soutar, JN Clarke, YM 1983, LIFE STYLE AND RADIO LISTENING PATTERNS IN PERTH, WESTERN AUSTRALIA, Australian Journal of Management, vol. 8, no. 1, p. 71. Available from: buh. Appendix 1 AIO Questionnaire