Consumer attitudes toward digital newspaper subscriptions: factors affecting WTP

As for demographic factors, the concept has been frequently used in the previous research as the important variable for analysing WTP in online news field of study. Among the key factors that were used as significant predictors there were mentioned age, gender, level of education and income. The data of listed factors was analysed with standard survey measurements (Julmi, 2020; Blattman et al., 2016). The chosen measurement scale for age variable: 1 = under 18; 2 = 18-24 years old; 3 = 25-34 years old; 4 = 35-44 years old; 5 = 45-54 years old; 6 = Over 55. Scale for gender variable: 1 = Male; 2 = Female; 3 = Other. Scale for the highest level of education: 1 = Less than High School; 2 = High School; 3 = Associate Degree or Some College; 4 = Bachelor's; 5 = Master's; 6 = Doctorate; 7 = Professional; 8 = Other. Scale for monthly income: 1 = $0 - $499; 2 = $500 - $999; 3 = $1000 - $1499; 4 = $1500 - $1999; 5 = $2000 - $2499; 6 = $2500 - $2999; 7 = $3000 - $3499; 8 = $3500 - $3999; 9 = $4000 - $4499; 10 = $4500 - $4999; 11 = $5000 +.

The concept of types of online newspapers refer to newspapers that can be categorized by size, business models, type of content, etc. Local newspapers are newspapers that are circulated in the local areas (Fogarty, 2012). National newspapers are newspapers that are circulated throughout a country (Bowden et al., 2019). International newspapers are newspapers that are sold and circulated and serve around the world (d'Haenens et al., 2004). These definitions, more precisely measurement variables for a concept of types of newspapers, are described as the kinds of newspapers by size. Moreover, newspapers can be divided into two categories: special interest and general interest newspapers (Kim & Kim, 2018). Special interest newspapers are newspapers that focus on specific topics whilst general interest newspapers are newspapers that cover a wide variety of topics and non-specialized. (Saavedra & Gonzбlez, 2015). The scales for each of newspapers' types are Likert scales that define the degree of WTP in correspondence to each variable.

3.2 Sample

The total population of the research analysis is people living in USA who have online subscriptions to any of the newspapers and people interested in the online content of newspapers that have a paywall model. For making the data collection possible the chosen population includes people who follow social media accounts (Facebook, Twitter, etc.) of variety of newspapers that use a paywall model on their websites and apps as well as social media groups and pages on media and journalism topics. The research has the objective to analyse American population and based on the responses of people living in USA explore the research questions and make relevant conclusions. The choice of the nationality for the analysis is explained by different reasons. Firstly, during pretesting with all kinds of nationalities the following issues were identified -- there is a difficulty to analyse the responses of each nationality as the number of people of each nationality differ significantly and that does not allow to represent the viewpoint of the whole population. Thus, the responses of this kind of sample are not representative for the obtained results. The nationality factor would be appropriate to analyse if there was a tool to collect a sufficient amount of responses corresponding to each nationality. For that reason, one nationality was chosen to examine in the study. Secondly, an American market of media is developed and that is common to read online news in this area (Newman & Levy, 2019). One of the first newspapers that introduces a paywall model online was The New York Times newspaper (Himma-Kadakas & Kхuts, 2015). That is why the chosen population provides an opportunity to examine participants among large amount of people who are interested in reading online news. But the choice of the population gives the research certain limitations and threats that will be discussed further.

The sampling method which is used in the research is probability sampling method that allows to obtain results of the sample that are representative for the entire population (Julmi, 2020). More precisely, cluster sampling and random sampling are used. Since the research focuses on the entire population of people (living in USA) who are interested in the online content on news publications and follow social media accounts of online newspapers that have a paywall on their websites, 10 newspapers (the most popular ones) were chosen randomly out of the large number of newspapers. The lists of the chosen newspapers involves:

The Guardian;

The Wall Street Journal;

The New York Times;

The Washington Post;

China Daily;

The Times of India;

The Sydney Morning Herald;

The Daily Mail;

USA Today;

The Independent.

Thus, clusters of stated newspapers include participants whose responses are representative for the whole population of people having an interest to consuming digital news. The audience of each newspaper cluster provides a point of view of the entire population and represents a sample frame of the research. The clusters do not differ significantly by size so all clusters were combined and analysed with a random sampling method.

The overall population provides sample units which are people who follow social media accounts of newspapers that offer online subscriptions to readers on their websites. The population size is more than 100000 people. According to the sample size formula, sample size equals to 384 with 5% margin of error at a 95% confidence level. Thus, 384 responses were needed to make the sample representative.

3.3 Data collection and processing tools

The data was gathered through an online questionnaire in order to analyse the responses and find out the relationship between necessary variables and thereby examine the influence of different factors on the respondents' willingness to pay for online newspaper subscriptions. A survey is divided into several sections in accordance with its objective to test each stated group of hypotheses, and each question is related to a particular hypothesis in order to analyse the influence of each factor on consumers' WTP. The first section is referred to the group of hypotheses about personal factors and focuses on the factors that are connected to people willingness to pay for an online newspaper that corresponds to their interests/hobbies, professional field or their attitude to the brand of a newspaper. The next section is devoted to different types of online newspapers and sets the questions of what the maximum amount of money a person is ready to pay of each of those types of online newspapers (local, national international, special interest and general interest newspapers). The last step for respondents in a questionnaire is to specify their demographic characteristics such as age, gender, the highest level of education and monthly income. Moreover, the data collected with questionnaire provides the information about the number of people who already have an online subscription to any of the newspapers and the number of those who do not have it but are asked to consider the probability of paying for it under several conditions and according to the factors that motivate them most.

The survey was sent to more than 2000 people using special tools allowing to collect database of users subscribed to social media accounts of news publications, and to send messages to people on social media platforms and to their email addresses (e.g. InstaPlus, FBSender). A survey was conducted in March and April 2020. A sample of a questionnaire is represented in Appendix 2.

3.4 Analysis

Firstly, in order to summarize and describe the collected data set so that the sample of population and observations are been understood (median, percentile values, mode, etc.) the descriptive statistics is carried out.

In order to evaluate the correlation between the following variables -- types of newspapers, the correspondence of an online newspaper to such personal factors as interests/hobbies, professional field and attitude to a brand -- the correlation analysis, based on a correlation matrix is used. Thus, the correlation between given variables is measured and interpreted. The exact method of a correlation analysis used is Spearman rank-order correlation coefficient as the data is measured at ordinal scale (Wang et al., 2019). media remuneration newspaper

The ordinal logistic regression was used in the previous research for evaluating WTP. This type of regression is carried out when a dependent variable is ordinal (when the categories are ordered) (Koletsi & Pandis, 2018; Chyi et al., 2010). During testing the assumptions of ordinal logistic regression, the proportional odds assumption was violated according to the test of parallel lines that indicated statistical significance. The proportional odds model represents the extension of the Binomial logistic regression when the variables are measured at the ordinal level. The proportional odds assumption means that there is a consistent relationship between each pair of independent and dependent variables (Brant, 1990). For that reason, other alternatives have been considered. It is possible to use multinomial logistic regression when the proportional odds assumption is not met in ordinal logistic regression. However, multinomial logistic regression provides less thrifty representation (Osborne, 2015). As a result, in order to understand how WTP level of various analysed factors differ across different groups, the chosen binomial logistic regression model have been chosen. The binomial logistic regression predicts the probability that an observation is being in one of two categories of a dependent variable (DeMaris, 1995). The binomial logistic regression model is the following:

logit(Y) = в0 + в1X1 + в2X2 + вkXk + е.

where: в0 -- the constant (also called the intercept);

в1 -- the slope coefficient of variable X1;

Y -- a dependent variable;

Logit -- the logarithm of the odds of the occurring event;

е -- the error term.

Finally, the Chi-Square test is used to identify the difference between the distributions of variables and define whether the result is statistically significant.

3.5 Limitations

One of the limitations that should be noted related to the data collection process. The targeting tool, which allows to send messages to people in social media who are subscribed to the accounts of any of studied online newspapers and who are living in USA, sets the limit to 3000 people. Due to expected response rate of 20%, the necessary number of responses was expected to be collected but that was not enough at the stage of conducting the analysis. That is why other ways to collect the responses were used but they did not consider the factor of being subscribed to social media accounts of online newspapers (the factor that can demonstrate the interest in reading online news). Moreover, the fact of being subscribed to social media accounts of online newspapers can be considered as opposite to the motivation to pay for online news as the content consumed on social media is free of charge. However, the only method, which could allow to gather the necessary number of responses, relates to the getting responses from people who are subscribed to online newspapers' social media accounts.

The next limitation follows from the violation of one of the assumptions of Ordinal logistic regression. As a result, Binomial Logistic regression is used as an alternative. However, having ordinal data, it is more appropriate to use Ordinal logistic regression or Generalized ordinal logistic regression which is more complicated and cannot be performed in SPSS Statistics, but solves the problem with the proportional odds assumption (Williams, 2016).

Finally, due to impossibility to gather data from different countries in the way to make results representative, the analysis focuses on US market that creates limitations for the results obtained.

4. Results

4.1 Descriptive analysis

First and foremost, the results of obtaining descriptive statistics are presented. The given results demonstrate the summary of important features of a sample of entire population. The obtained information allows to provide the organized summary that demonstrates the information about analysed variables and can be useful for understanding and potential justification of the relationships between variables.

For the following ordinal variables -- WTP in correspondence to interest/hobbies of a person; WTP in correspondence to professional field of a person; WTP for an online newspaper which brand is considered as well-known worldwide for a person; WTP for a local online newspaper; WTP for a national online newspaper; WTP for an international online newspaper; WTP for a special interest online newspaper; WTP for a general interest online newspaper -- the Frequencies procedure is used. The statistics that represent the information about ordinal variables are the mode, median, quartiles, range, maximum and minimum values. The descriptive statistics is displayed in Table 1.

The frequency tables represent the necessary information for the further analyses. The number of valid observations equal to 385 (N=385). According to the mode as one of the main indicators of frequency, the obtained information represents the most frequent WTP level that respondents have chosen in accordance with each factor. The WTP at a «high» level (3) as the most frequent answer was for all factors except for local online newspapers and online newspapers that are perceived as well-known worldwide (the WTP is the lowest (1) was chosen for these factors). As for percentiles, a «low» level of WTP (1) is indicated as the value below which 25% of observations falls for such variables as the WTP degree for a newspaper that corresponds to a person's interests or hobbies, which is considered as worldwide brand, and for local, national and international online news publications. For the rest of ordinal variables, the value equals to 2 (medium level of WTP). As for 50% percentile, which is also considered as a median, 50% of the values are below the value of 2 (medium level of WTP) for such variables as the correspondence to interests, worldwide well-known brand, and local, national and international types. The correspondence to professional field, special interest and general interest types of newspapers are the variables for which high level of WTP (3) is indicated as the value below which 50% of other values are. Finally, the 75% percentile provides the information that high WTP degree is the value below which 75% of observations falls for all variables expect for local online newspapers for which medium WTP level is a value below which 75% of observations fall for this variable.

Table 1 The descriptive statistics

For the other categorical variables, which are demographic characteristics of a sample and whether a person already have an online subscription to a newspaper, the mode, minimum and maximum values are chosen to represent the frequency. Frequency table shows that the most frequently occurring value for gender is Female (where 2 = Female), for educational level -- Bachelor's (where 4 = Bachelor's), for monthly income group -- $2500 - $2999 (where 6 = $2500 - $2999) for age group -- 18-24 years old (where 2 = 18-24 years old). The value that occurs most often in the data set related to whether a person is currently subscribed to an online newspaper or not is Yes (where 1 = Yes).

Figure 1. Distribution of WTP by correspondence to interests

Figure 2. Distribution of WTP by correspondence to professional field

The distributions by gender, level of education, age group and monthly income group, whether a person is currently subscribed to an online newspaper as well as the distribution of data by factor affecting WTP are shown visually in Appendix 3.

The frequency distribution graphs for personal factors and different types of online newspapers are represented in Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6.

Figure 3. Distribution of WTP by brand which is considered as well-known worldwide

Figure 4. Comparison of distributions of local and national online newspapers

Figure 5. Comparison of distributions of local and international online newspapers

Figure 6. Comparison of distributions of special and general interest online newspapers

4.2 Correlation analysis

In order to study the strength of relationship for given personal factors and various types of online newspapers, the correlation analysis is used. The analysis is carried out to determine whether there is an association between analysed variables and to what extent these variables are related. Spearman's Rank-Order Correlation analysis is used as the variables are measured at the ordinal level of measurement.

The correlation coefficient results show that there is a strong positive and significant correlation between the following variables representing the level of WTP in accordance with personal factors: the conformity of a newspaper to interests/ of a person and a newspaper's brand which is well-known (0.868), the conformity of a newspaper to personal interests and correspondence of a newspaper to professional field (0.725) . The correlation coefficients show a moderate (close to strong) positive and significant correlation between the conformity of a newspaper to professional field of a person and a newspaper's brand which is considered as well-known (0,677). That is, in each case the variables move in the same direction.

The correlation matrix for variables related to WTP for various types of online newspapers provides the information about the strength of correlation between described variables. The relationship is positively strong between: local and national kinds of newspapers (0,794); national and international newspapers (0,812); national and non-specialize newspapers (0,736); international and non-specialize newspapers (0,739); specialized and non-specialized newspapers (0,756). The relationship is positively moderate for the rest variables. The correlation matrices are presented in Appendix 4.

4.3 Regression analysis

With the obtained results of a logistic regression the independent variables that have a statistically significant effect are identified. Moreover, the predictions for each of independent variables are made. The adequacy of the model is analysed.

The following demographic predictors are used -- gender, age group, level of education, income group. Moreover, the influence of the factor related to whether a person is subscribed to an online newspaper or not on the dependent variable is analysed. It should be noted that in a regression analysis ordinal independent variables are included in the covariates, for each nominal variable the categorical variables are defined with reference groups.

Firstly, WTP for a newspaper that corresponds to personal interests is chosen for the regression analysis as an outcome variable. According to the analysis results, there are no missing cases. The information is represented by Case Processing Summary. From the Omnibus Tests of Model Coefficients it is clear that the model is statistically significant (p < .000). That means that the model fits significantly well. According to Cox & Snell R Square and Nagelkerke R Square values, which are useful for analysing how much variance in a dependent variable might me explained, the explained variation ranges from 38% to 51%. Overall classification accuracy is 78.2% which is higher than the same indicator in the null model (which is 50.9%).

The Variables in the Equation table (Table 2) demonstrates the given results:

Gender is a positive and significant predictor of the probability of high WTP (B = 1.176, p = .000), with the Exp (B) determining that for each increase in one unit of this predictor the odds of high WTP change by 3.241 (the odds are increasing). B coefficients predict the likelihood of the occurring event (having a high WTP). That indicates that females are more likely to have higher WTP for online newspapers that correspond to their interests or hobbies.

Age group is a negative and statistically significant variable of the probability of high WTP (B = - .615, p = .000). OR indicates that for every one unit reduction of an independent variable the odds of high WTP increase by a factor of .541. The negative coefficient indicates that people who are older are more likely to have lower WTP for online newspapers that correspond to their interests.

Monthly income group is a positive and significant predictor of the probability of high WTP as B = .597, p = .000, Exp (B) = 1.817. For every one unit increase on the predictor the odds of high WTP increase by 1.817. That means that increasing monthly income of respondents leads to increase in the probability of the given dependent variable.

Whether a person is subscribed to a newspaper or not is a positive and significant predictor of the probability of high WTP (B = 2.104, p = .000). Odds Ratio shows that for every one unit increase on the predictor the odds of high WTP increase by a factor of 8.199. Thus, those who have a subscription to any of the online newspapers are 8.2 times more likely to have higher WTP.

Educational level is a non-significant independent variable (B = .017, p = .876).

Table 2 Variables in the Equation for Correspondence to personal interests

The next dependent variable which is analysed is WTP for an online newspaper that corresponds to professional field of a person. A binomial logistic regression is carried out to evaluate the influence of gender, age, educational level, monthly income and presence of an online subscription on the probability of having a high WTP for an online newspaper subscription. The model is statistically significant (p = .000) based on the Omnibus Tests of Model Coefficients table. The logistic regression model explains 38% of the variance in an outcome. According to the Classification Table, sensitivity indicator is 70 % and specifity is 82.9%. Overall percentage is 78.2%, the addition of independent variables makes an improvement in the overall prediction of variables as in the beginning block the model correctly classified 63.6 % of all cases.

Proceeding to the most important table, 3 independent variables are statistically significant whilst 2 other predictors are non-significant. Older groups of respondents are ,567 less likely to have a high WTP when it comes to the correspondence to their professional field (B = - .567, p = .001). Increasing monthly income related to the increase of the likelihood to have a high WTP by a factor of 1.492 (B = .400, p = .000). Those people who have subscriptions to any of the online newspapers are associated with the rise in the probability of having a high WTP. They have 5.859 higher odds to have a high WTP (Table 3).

Table 3 Variables in the Equation for Correspondence to professional field

The relationship between different factors and WTP for a newspaper which brand is considered as well-known worldwide for a person is analysed. There are no missing cases. Omnibus Test of Model Coefficients represents a good model fit (p = .000). The Model explains 48% of the variation in the dependent variable. The Model correctly classifies 74.5% cases (higher than in the null model where this indicator equals to 61.8%. The last table shows that all factors except for educational level are statistically significant. Women are 3.988 times more likely to have a high WTP for a newspaper brand which they know well than men. Older people are associated with the decrease in the probability of having a high WTP. Increase in monthly income is connected to the increase in the likelihood of having a high WTP. People who are subscribed to any of the online newspapers are much more likely to have a high WTP (Table 4).

Table 4 Variables in the Equation for Brand which is well-known worldwide

The likelihood of having a high WTP for local newspapers is analysed in relationship with other factors. The model correctly classifies 85.5% of cases. Regression model is statistically significant (p = .000). Only one of the analysed predictors is non-significant -- a level of education. (Table 5). Females are 5.543 times have higher odds to have a high WTP than males. Age group and monthly income group are positive and significant predictors and it means that the increase in each predictor is associated with the increase of the likelihood to have a high WTP for a local newspaper. People with online subscriptions are 5.074 times more likely to have a high WTP.

Table 5 Variables in the Equation for Local online newspapers

The likelihood of having a high WTP for a national online newspaper is analysed among the given predictors. The Model is statistically significant. Overall, it explains 74.5% of cases. Only age group predictor is non-significant. Females are 3.039 times more likely to have a high WTP than males. The increase in educational level and monthly income is associated with the increase in the likelihood of having a high WTP. Presence of subscriptions is associated with an increased probability to have a high WTP (Table 6).

Table 6 Variables in the Equation for National online newspapers

The probability of having a high WTP for international newspapers is analysed in relationship with other factors. The model correctly classifies 74.5% of cases. Regression model is statistically significant. Only educational level predictor is not statistically significant. Females are 2.765 times have higher odds to have a high WTP than males. The increase in monthly income is associated with the increase in the likelihood to have a high WTP by 63%. People with online subscriptions are 10.068 times more likely to have a high WTP. Moreover, the rise in age is connected to have a low WTP for international newspapers (Table 7).

Table 7 Variables in the Equation for International online newspapers

The last two dependent variables -- WTP for special interest and general interest -- are analysed in relationship with the given predictors. Both models are statistically significant. The increase in educational level and monthly income relates to the increase of likelihood to have a high WTP for special interest newspaper but the increase in age relates to the reduction of this occurring event. Those who have subscriptions are more likely to have a high WTP for special interest newspapers (Table 8). As for general interest newspapers, females are more likely to have a high WTP than males, the rise in monthly income indicates the increase of the probability to have a high WTP as well as the presence of an online subscription. However, age group and educational level predictors are not statistically significant (Table 9).

Table 8 Variables in the Equation for Special interest online newspapers

Table 9 Variables in the Equation for General interest online newspapers

4.4 Chi-Square analysis

The Chi-Square test is used to identify the difference between the distributions of variables. Firstly, personal factors are analysed with the given type of test. Test Statistics in Table 10 shows that there is a statistically significant result for each variable (p-value < 0.05).

Table 10 Test Statistics

The relationship between different types of online newspapers is also analysed with the Chi-Square test. Table 11, Table 12 and Table 13 represent statistical significance for the relationship between local and national newspapers. local and international newspapers, special interest and general interest newspapers.

Table 11 Chi-Square Tests for local and national online newspapers

Table 12 Chi-Square Tests for local and international online newspapers

Table 13 Chi-Square Tests for special and general interest online newspapers

4.5 Discussion

The obtained results should be interpreted in connection to the stated hypotheses. The following discussion determines whether the hypotheses are confirmed or rejected, the justification of the results and their connection to the finding of previous studies.

The first group of hypotheses related to demographic factors is discussed. The model for the test of each hypothesis is statistically significant (p < 0.05). As for the monthly income, all models show this predictor as positive and significant. For all dependent variables the increase in monthly income is associated with the increase of the probability to have a high WTP (high or very high levels in comparison to low or medium levels). Therefore, H1a hypothesis is confirmed. In the research of Chiou & Tucker (2013) the authors note that the number of readers with lower income fell down after introducing new pay models by some newspapers on their websites. As for the age group, it is a significant predictor for each outcome except for WTP for national and general interest online newspapers. According to the results where this predictor is significant, the hypothesis H1b can be confirmed. The increase in age rate demonstrates the reduction of the probability to have a high WTP for all models except for WTP for local newspapers. However, the hypothesis can be significantly affected by other factors that are not considered in the present research. The obtained result can be supported by the previous findings on younger audience - younger people are more likely to spend more time in digital environment and consume online news than older people (Olsen, 2020). Moreover, the significant predictor in the research of Chyi (2005) is age that indicated that younger audience showed more intend to pay for online news. The probability of having a high WTP for women is positive (B is positive) in each model except for WTP connected to special interest newspapers where the predictor is not statistically significant. Odds Ratio is particularly high for WTP for local and national newspapers as well as WTP for newspapers that corresponds to personal interests and are considered as well-known worldwide for a person. That means that females have higher odds to have a high WTP for the aforementioned outcomes. Thus, the hypothesis H1c is confirmed. Theoretical foundation confirms the fact that women are more likely to feel social connection and actively keep up conversations while consuming news information on the websites and as a result are more likely to have higher WTP for news information than men (Punj, 2015). Finally, educational level of a person is significant in only two models that does not allow to confirm the H1d. However, the predictor is significant and positive for WTP for national and special interest online newspapers. Thus, those who have higher level of education are more likely to have a high WTP for the given outcomes. Chyi et al. (2010) in their research indeed find out that more than a half of readers who pay for the online news have higher educational levels than those who do not pay. The previous research shows that younger audience and people with less educational level are less likely to buy print newspapers. The factor with age differs between print and online newspapers based on the findings of present research.

The second group of the hypotheses is associated to personal factors. Based on the correlation analysis, the correlation is positive and statistically significant for all pairs of variables (p-value = .000). Given the results of frequency tables and Chi-Square Test Statistics, it is clear that a high level of WTP is chosen more often for online newspapers that correspond to personal interests and professional field. Thus, the hypothesis H2a is confirmed. Based on the previous study, indeed personal attitude to the topic and a brand is associated to the motivation to pay for online content as well as willingness to pay for print newspapers (Saavedra & Gonzбlez, 2015). However, the present research finds out that high and low levels of WTP for an online newspaper's brand that is perceived as well-known are almost similar, for that reason it is not possible to confirm the hypothesis H2b.

Finally, there is a statistically significant positive correlation between WTP indicators for all types of newspapers. Comparison of distributions of the variables and the results of Chi-Square Test represent the following results that confirm the hypotheses H3a, H3b. National and international online newspapers as factors demonstrate higher levels of WTP than local online newspapers. It is seen from the graphs that a prevailing level for local online publications is low. Indeed, the scholars of previous research note that attitudes to local, regional and global news differ (Goyanes, 2015). The hypothesis H3c states that special interest online newspapers have higher WTP than general interest online newspapers, however the present results show that the levels for each factor are almost the same. That is why H3c cannot be confirmed or rejected. Further studies on this topic need to be developed considering other factors that might influence the results. The previous research indicates that those who read special interest newspapers is a separate niche so further analysis on this topic could consider this group separately.

In addition to the described predictors, the current presence of a subscription to any of the online newspapers shows the strong increase in the likelihood of having a high WTP in all regression models.

Conclusion

The conducted analysis made it possible to answer the research questions, test stated hypotheses and make conclusions on the obtained results. Among the factors that strongly affect willingness to pay and are of high value for people considering the decision to pay for a subscription to an online newspaper, there are the correspondence of the newspaper's content to personal interests or hobbies and professional field. These factors demonstrate the motivation to pay for an online subscription and can be critically explained with the assumption that people are ready to pay for the product which is highly correlated to their interests and which is valuable and beneficial for their professional areas. However, the research indicates that if a brand of a newspaper is well-known worldwide, it does not mean that a person is highly motivated to pay for it. The attitude towards brand showed a low level of willingness to pay but still this topic needs to be studied in detail further. The analysis of other topics related to a brand of a newspaper, such as brand loyalty, brand commitment, brand satisfaction, etc., could give more appropriate and detailed results on the topic of willingness to pay. Furthermore, it is worth paying attention to the readiness to pay for different online newspapers and conclude whether there is the difference across different types of online news editions or not. It is identified that people are willing to pay for national and international online newspapers more than for local newspapers. Local newspapers struggle to compete with larger players of the market. Local companies should consider other pay models or experiment with other business models and different marketing techniques associated with attraction and retention of consumers. People consider larger newspapers by geography more valuable and as a source to get much more global information, therefore the audience is motivated to pay for this type of information at higher extent. As for the comparison of WTP for special interest and general interest online newspapers, the results showed that the degree of intention to pay is approximately same across these two types of editions. That is why more in-depth analysis of these factors need to be performed for better understanding readers' preferences and needs towards a kind of content offered by an online newspaper. Separately, conclusions should be made about the effect of demographic factors on levels of WTP. It was identified that people who have higher monthly income are more likely to pay for online newspapers' content. Moreover, younger people are more likely to pay for an online subscription to a newspaper. As for age group, females showed higher willingness to pay for subscriptions to online newspapers then males. However, the association between educational level and willingness to pay was not significantly identified. In addition, there were concerns regarding to what extent the results of t...